LM97593 Dual ADC / Digital Tuner / AGC General Description Features The LM97593 Dual ADC / Digital Tuner / AGC IC is a two channel digital downconverter (DDC) with integrated 12-bit analog-to-digital converters (ADCs) and automatic gain control (AGC). The LM97593 further enhances National’s Diversity Receiver Chipset (DRCS) by integrating a wide-bandwidth dual ADC core with the DDC. The complete DRCS includes one LM97593 Dual ADC / Digital Tuner / AGC and two CLC5526 digitally controlled variable gain amplifiers (DVGAs). This system allows direct IF sampling of signals up to 300MHz for enhanced receiver performance and reduced system costs. A block diagram for a DRCS-based narrowband communications system is shown in Figure 1. The LM97593 offers high dynamic range digital tuning and filtering based on hard-wired digital signal processing (DSP) technology. Each channel has independent tuning, phase offset, filter coefficients, and gain settings. Channel filtering is performed by a series of three filters. The first is a 4-stage Cascaded Integrator Comb (CIC) filter with a programmable decimation ratio from 8 to 2048. Next there are two symmetric FIR filters, a 21-tap and a 63-tap, both with independent programmable coefficients. The first FIR filter decimates the data by 2, the second FIR decimates by either 2 or 4. Channel filter bandwidth at 52MSPS ranges from ±650kHz down to ±1.3kHz. At 65MSPS, the maximum bandwidth increases to ±812kHz. The LM97593’s AGC controller monitors the ADC output and controls the ADC input signal level by adjusting the DVGA setting. AGC threshold, deadband+hysteresis, and the loop time constant are user defined. Total dynamic range of greater than 123dB full-scale signal to noise in a 200kHz bandwidth can be achieved with the Diversity Receiver Chipset. ■ 100% Software compatible with the CLC5903 ■ Pin compatible with the CLC5903 except for the analog ■ ■ ■ ■ ■ ■ ■ ■ ■ input and reference section 123 dB dynamic range with CLC5526 DVGA (200kHz) On-chip precision reference User Programmable AGC with enhanced Power Detector Channel Filters include a Fourth Order CIC followed by 21tap and 63-tap Symmetric FIRs Flexible output formats Serial and Parallel output ports JTAG Boundary Scan 8-bit Microprocessor Interface 128 pin PQFP Key Specifications ■ ■ ■ ■ ■ ■ ■ Internal ADC Resolution Sample Rate SNR (fIN = 250MHz, 11-bit, Nyquist) SNR (fIN = 250MHz, 200kHz) SFDR (fIN = 250MHz, 11-bit, Nyquist) Full Power Bandwidth Power Consumption (65MSPS) 12 Bits 65 MSPS 62 dBFS (typ) 83 dBFS (typ) 68 dBFS (typ) 650 MHz (typ) 560 mW (typ) Applications ■ ■ ■ ■ ■ Cellular Basestations GSM / GPRS / EDGE / GSM Phase 2 Receivers Satellite Receivers Wireless Local Loop Receivers Digital Communications Block Diagram 1 30008701 FIGURE 1. Diversity Receiver Chipset Block Diagram © 2008 National Semiconductor Corporation 300087 www.national.com LM97593 Dual ADC / Digital Tuner / AGC March 12, 2008 LM97593 Connection Diagram 30008702 FIGURE 2. LM97593VH PQFP Pinout Ordering Information Industrial (−40°C to +85°C) Package LM97593VH 128 Pin PQFP LM97593EB Evaluation Board Block Diagram 2 30008703 FIGURE 3. LM97593 Block Diagram www.national.com 2 LM97593 Pin Descriptions and Equivalent Circuits Pin No. Symbol Equivalent Circuit Description 13 27 VINA− VINB− Analog Input Negative differential input signal for the 'A' channel Negative differential input signal for the 'B' channel 14 26 VINA+ VINB+ Analog Input Positive differential input signal for the 'A' channel Positive differential input signal for the 'B' channel Control / Analog Input Reference Select Pin / External Reference Voltage Input Input differential full scale swing = 2 * VREF VREF = VA to VA - 0.3V: Reference Voltage = 1.0 V (Internal) VREF = 0.8V to 1.5V: Reference Voltage = VREF (External) ANALOG I/O 21 VREF 15 24 VCOMA VCOMB Analog Output Common Mode reference voltage for the 'A' channel Common Mode reference voltage for the 'B' channel These pins may be loaded to 1 mA for use as temperature stable 1.5V references. 16 23 VRPA VRPB Analog Output Upper reference voltage for the 'A' channel Upper reference voltage for the 'B' channel 17 22 VRNA VRNB Analog Output Lower reference voltage for the 'A' channel Lower reference voltage for the 'B' channel REFSEL/DCS Control Input This is a three-state pin. VCOM = VCOMA or VCOMB. REFSEL/DCS = AGND: the internal reference is enabled and duty cycle correction is applied to the ADC input clock (CK). REFSEL/DCS = VCOM: the internal reference is enabled and no duty cycle correction is applied to the ADC input clock (CK). REFSEL/DCS = VA: DCS is on, the internal reference is disabled. Apply A 0.8-1.2V external reference to the VREF pin. 30 PD Input POWER DOWN, when high both ADCs are powered down, when low, both ADCs are enabled 45 MR Input MASTER RESET, Active low Resets all registers within the chip. ASTROBE and BSTROBE are asserted during MR. 8 DIGITAL I/O 82 78 AOUT BOUT Output SERIAL OUTPUT DATA, Active high The 2's complement serial output data is transmitted on these pins, MSB first. The output bits change on the rising edge of SCK (falling edge if SCK_POL=1) and should be captured on the falling edge of SCK (rising if SCK_POL=1). These pins are tri-stated at power up and are enabled by the SOUT_EN control register bit. See Figure 13 and Figure 34 timing diagrams. In Debug Mode AOUT=DEBUG[1], BOUT=DEBUG[0]. 127:125 40:42 AGAIN[2:0] BGAIN[2:0] Output OUTPUT DATA TO DVGA, Active high 3 bit bus that sets the gain of the DVGA determined by the AGC circuit. 124 43 ASTROBE BSTROBE Output DVGA STROBE, Active low Strobes the data into the DVGA. See Figure 7 and Figure 41 timing diagrams. Output SERIAL DATA CLOCK, Active high or low The serial data is clocked out of the chip by this clock. The active edge of the clock is user programmable. This pin is tri-stated at power up and is enabled by the SOUT_EN control register bit. See Figure 13 and Figure 34 timing diagrams. In Debug Mode outputs an appropiate clock for the debug data. If RATE=0 the input CK duty cycle will be reflected to SCK. 80 SCK 3 www.national.com LM97593 Pin No. 99 81 Symbol SCK_IN SFS 84, 86:88, 90, 91, 93:97, POUT[15:0] 104:106, 108, 109 Equivalent Circuit Description Input SERIAL DATA CLOCK INPUT, Active high or low Data bits from a serial daisy-chain slave are clocked into a serial daisychain master on the falling edge of SCK_IN (rising if SCK_POL=1 on the slave). Tie low if not used. Output SERIAL FRAME STROBE, Active high or low The serial word strobe. This strobe delineates the words within the serial output streams. This strobe is a pulse at the beginning of each serial word (PACKED=0) or each serial word I/Q pair (PACKED=1). The polarity of this signal is user programmable. This pin is tri-stated at power up and is enabled by the SOUT_EN control register bit. See and timing diagrams. In Debug Mode SFS=DEBUG[2]. Output PARALLEL OUTPUT DATA, Active high The output data is transmitted on these pins in parallel format. The POUT_SEL[2:0] pins select one of eight 16-bit output words. The POUT_EN pin enables these outputs. POUT[15] is the MSB. In Debug Mode POUT[15:0]=DEBUG[19:4]. 112:114 POUT_SEL[2:0] Input PARALLEL OUTPUT DATA SELECT, Active high The 16-bit output word is selected with these 3 pins according to . Not used in Debug Mode. For a serial daisy-chain master, POUT_SEL [2:0] become inputs from the slave: POUT_SEL[2]=SFSSLAVE, POUT_SEL[1]=BOUTSLAVE, and POUT_SEL[0]=AOUTSLAVE. Tie low if not used. 111 POUT_EN Input PARALLEL OUTPUT ENABLE. Active low This pin enables the chip to output the selected output word on the POUT[15:0] pins. Not used in Debug Mode. Tie high if not used. Output READY FLAG, Active high or low The chip asserts this signal to identify the beginning of an output sample period (OSP). The polarity of this signal is user programmable. This signal is typically used as an interrupt to a DSP chip, but can also be used as a start pulse to dedicated circuitry. This pin is active regardless of the state of SOUT_EN. In Debug Mode RDY=DEBUG[3]. Input INPUT CLOCK. Active high The clock input to the chip. The The VINA and VINB analog input signals are sampled on the rising edge of this signal. SI is clocked into the chip on the rising edge of CK. Input SYNC IN. Active low The sync input to the chip. The decimation counters, dither, and NCO phase can be synchronized by SI. This sync is clocked into the chip on the rising edge of CK. Tie this pin high if external sync is not required. All sample data is flushed by SI. To properly initialize the DVGA ASTROBE and BSTROBE are asserted during SI. Input/Output DATA BUS. Active high This is the 8 bit control data I/O bus. Control register data is loaded into the chip or read from the chip through these pins. The chip will only drive output data on these pins when CE is low, RD is low, and WR is high. 48, 50, 52:57 A[7:0] Input ADDRESS BUS. Active high These pins are used to address the control registers within the chip. Each of the control registers within the chip are assigned a unique address. A control register can be written to or read from by setting A [7:0] to the register’s address and setting CE, RD, and WR appropriately. 59 Input READ ENABLE. Active low This pin enables the chip to output the contents of the selected register on the D[7:0] pins when CE is also low. 77 37 46 RDY CK SI 62, 63, 69:73, D[7:0] 75 www.national.com RD 4 58 Symbol WR Equivalent Circuit Description Input WRITE ENABLE. Active low This pin enables the chip to write the value on the D[7:0] pins into the selected register when CE is also low. This pin can also function as RD/ CE if RD is held low. See for details. 60 CE Input CHIP ENABLE. Active low This control strobe enables the read or write operation. The contents of the register selected by A[7:0] will be output on D[7:0] when RD is low and CE is low. If WR is low and CE is low, then the selected register will be loaded with the contents of D[7:0]. 116 TDO Output TEST DATA OUT. Active high 117 TDI Input TEST DATA IN. Active high with pull-up 118 TMS Input TEST MODE SELECT. Active high with pull-up 119 TCK Input TEST CLOCK. Active high. Tie low if JTAG is not used. 121 TRST Input TEST RESET. Active low with pull-up Asynchronous reset for TAP controller. Tie low or to MR if JTAG is not used. 122 SCAN_EN Input SCAN ENABLE. Active low with pull-up Enables access to internal scan registers. Tie high. Used for manufacturing test only! Digital Power Supplies 38, 39, 64, 79, 92, 102, 107, 128 VDR DDC Output Driver Power I/O Power Supply, 3.3V nominal. Quantity 8. 1, 47, 61, 68, 83, 89, 98, DRGND 110 DDC Output Driver Ground I/O Ground Return. Quantity 8. 49, 74, 85, 115, 123 VD18 DDC Core Power DSP Digital Core Power Supply, 1.8V nominal. Quantity 5. 49, 74, 85, 115, 123 VD18 DDC Core Power DSP Digital Core Power Supply, 1.8V nominal. Quantity 5. 44, 51, 65, 66, 76, 103, 120 D18GND DDC Core Ground DSP Digital Core Ground Return. Quantity 7. 4, 6, 31, 34 VD ADC Digital Power ADC Digital Logic Power Supply, 3.3V nominal. Quantity 4. 5, 7, 32, 33 DGND ADC Digital Ground ADC Digital Logic Ground Return. Quantity 4. Analog Power Supplies 10, 11, 19, 25, 29 VA ADC Analog Power ADC Analog Power Supply, 3.3V nominal. Quantity 5. 2, 9, 12, 18, 20, 28 AGND ADC Analog Ground ADC Analog Ground Return. Quantity 6. NC Not Connected. These pins should be left floating. Unconnected Pins 3, 35, 36, 67, NC 100, 101 5 www.national.com LM97593 Pin No. LM97593 Absolute Maximum Ratings Operating Ratings Soldering process must comply with National Semiconductor's Reflow Temperature Profile specifications. Refer to www.national.com/packaging. (Note 6) Operating Temperature −40°C ≤ TA ≤ +85°C Range ADC Analog, Digital and IO +3.0V to +3.6V Supply Voltages (VA, VD and VDR) Digital Core Supply Voltage (VD18) +1.6V to +2.0V Difference Between AGND, DGND, DRGND and D18GND ≤ 100 mV Voltage on Any Input or Output Pin 0V to +3.3V VCM 1.0V to 2.0V Clock Duty Cycle 30% to 70 % (Notes 1, 2) If Military/Aerospace specified devices are required, please contact the National Semiconductor Sales Office/ Distributors for availability and specifications. ADC Analog, Digital and IO Supply Voltages (VA, VD and VDR) Difference between VA, VD, and VDR Positive Core Supply Voltage (VD18) Voltage on Any Input or Output Pin (Not to exceed 4.2V) Input Current at Any Pin other than Supply Pins (Note 3) Package Input Current (Note 3) Max Junction Temp (TJ) −0.3V to 4.2V ≤ 100 mV −0.3V to 2.35V −0.3V to (VDR +0.3V) ±5 mA ±50 mA +125°C 39°C/W Thermal Resistance (θJA) Package Dissipation at TA = 25°C (Note 4) ESD Susceptibility (Note 5) (Notes 1, 2) 3.2W Reliability Information Transistor Count 2000 V Human Body Model (1.5kΩ, 100pF) 1.3 million 200 V Machine Model (0Ω, 200pF) Charge Device Model Storage Temperature 750 V −65°C to +150°C LM97593 Electrical Characteristics Unless otherwise specified, the following specifications apply: AGND = DGND = DRGND = D18GND = 0V, VA = VD = VDR = +3.3V, VD18 = +1.8V, Internal VREF = +1.0V, fCLK = 65 MHz, VCM = VCOM, tR = tF = 1 ns, CL = 5 pF/pin. The ADC’s 11 most significant bits observed at the mixer output debug tap with NCO = 0Hz. Typical values are for TA = 25°C. Boldface limits apply for TMIN ≤ TA ≤ TMAX. All other limits apply for TA = 25°C. (Notes 7, 8, 9) Symbol Parameter Conditions Typical (Note 10) Limits Units (Limits) STATIC CONVERTER CHARACTERISTICS Resolution with No Missing Codes 11 Bits (min) INL Integral Non Linearity (Note 11) Ramp, End Point ±0.7 2 -2 LSB (max) LSB (min) DNL Differential Non Linearity Ramp, End Point ±0.3 0.85 -0.85 LSB (max) LSB (min) VOFF Offset Error −40°C to +85°C -4.1 LSB REFERENCE AND ANALOG INPUT CHARACTERISTICS 1.0 V (min) 2.0 V (max) VCM Common Mode Input Voltage 1.5 VCOMA VCOMB Reference Output Voltage 1.5 V CIN VIN Input Capacitance (each pin to GND) (VIN= 1.5Vdc ±0.5V) (Note 12) CK LOW 8 pF CK HIGH 7 VREF External Reference Voltage (Note 14) 1.0 Reference Input Resistance 1 pF 0.8 V (min) 1.2 V (max) MΩ DYNAMIC CONVERTER CHARACTERISTICS FPBW SNR www.national.com Full Power Bandwidth Signal-to-Noise Ratio fIN = 20MHz, VIN = -3dBFS fIN = 249MHz, VIN = -3dBFS fIN = 249MHz, VIN = -9dBFS 6 650 MHz 66.2 63.7 63.9 dBFS dBFS dBFS (min) 62.2 Typical (Note 10) Signal-to-Noise and Distortion fIN = 20MHz, VIN = -3dBFS fIN = 249MHz, VIN = -3dBFS fIN = 249MHz, VIN = -9dBFS 62.8 62.0 63.4 ENOB Effective Number of Bits (Relative to Full Scale) fIN = 20MHz, VIN = -3dBFS fIN = 249MHz, VIN = -3dBFS fIN = 249MHz, VIN = -9dBFS 10.6 10.0 10.3 Bits Bits Bits (min) THD Total Harmonic Distortion fIN = 20MHz, VIN = -3dBFS fIN = 249MHz, VIN = -3dBFS fIN = 249MHz, VIN = -9dBFS -77.1 -57.9 -64.6 -54.1 dBc dBc dBc (max) Second Harmonic Distortion fIN = 20MHz, VIN = -3dBFS fIN = 249MHz, VIN = -3dBFS fIN = 249MHz, VIN = -9dBFS -82.7 -59.9 -67.3 -59.3 dBc dBc dBc (max) Third Harmonic Distortion fIN = 20MHz, VIN = -3dBFS fIN = 249MHz, VIN = -3dBFS fIN = 249MHz, VIN = -9dBFS -91.7 -69.0 -72.5 -56.8 dBc dBc dBc (min) fIN = 20MHz, VIN = -3dBFS fIN = 249MHz, VIN = -3dBFS fIN = 249MHz, VIN = -9dBFS 79.7 67.0 dBFS dBFS (min) dBFS SINAD H2 H3 SFDR IMD Parameter Spurious Free Dynamic Range Limits 60.4 68.0 76.3 Units (Limits) dBFS dBFS dBFS (min) f1IN = 246MHz, VIN = -15dBFS f2IN = 250MHz, VIN = -15dBFS (f1IN + f2IN = -9dBFS) -77.5 -3 dBFS reference, -50dBFS target ±1.3 Channel - Channel Offset Match 10 MHz -1dBFS driven ±0.2 %FS Channel - Channel Gain Match 10 MHz -1dBFS driven ±0.4 %FS 249MHz -3dBFS driven channel, 50Ω termination measured channel 60(+42) (Note 16) dBc Intermodulation Distortion Dynamic Gain Error dBFS ±2 dB INTERCHANNEL CHARACTERISTICS Crosstalk (AGC fixed at 0dB gain, with AGC operating the crosstalk will improve at the output) CIC OUTPUT CHARACTERISTICS SNR Signal-to-Noise Ratio SINAD Signal-to-Noise and Distortion SFDR Spurious Free Dynamic Range CIC Decimation = 8, NCO = 11.1MHz (248.9MHz @65MSPS aliases to 11.1 MHz), fIN = 249MHz at -3dBFS, signal observed at F1 In Debug tap 68.6 66.1 dBFS 68.5 66.1 dBFS 75.1 70.3 dBc DDC OUTPUT CHARACTERISTICS SNR Signal-to-Noise Ratio SINAD Signal-to-Noise and Distortion SFDR Spurious Free Dynamic Range SNR Signal-to-Noise Ratio SINAD Signal-to-Noise and Distortion SFDR Spurious Free Dynamic Range SNR Signal-to-Noise Ratio SINAD Signal-to-Noise and Distortion SFDR Spurious Free Dynamic Range GSM Filter set, 200kHz 82 (+42) (Note channel BW, NCO = 11.1MHz 16) (248.9MHz @52MSPS aliases 73 to 11.1 MHz), fS = 52MSPS, 90 fIN = 249MHz at -9dBFS GSM Filter set, 200kHz 76 (+42) (Note channel BW, NCO = 11.1MHz 16) (248.9MHz @52MSPS aliases 74 to 11.1 MHz), fS = 52MSPS, 90 fIN = 249MHz at -3dBFS GSM Filter set, 200kHz 79 (+42) (Note channel BW, NCO = 11.1MHz 16) (248.9MHz @65MSPS aliases 71 to 11.1 MHz), fS = 65MSPS, 81 fIN = 249MHz at -9dBFS 7 dBFS dBc dBc dBFS dBc dBc dBFS dBc dBc www.national.com LM97593 Conditions Symbol LM97593 Symbol Parameter SNR Signal-to-Noise Ratio SINAD Signal-to-Noise and Distortion SFDR Spurious Free Dynamic Range Conditions Typical (Note 10) Limits GSM Filter set, 200kHz 74 (+42) (Note channel BW, NCO = 11.1MHz 16) (248.9MHz @65MSPS aliases 71 to 11.1 MHz), fS = 65MSPS, 80 fIN = 249MHz at -3dBFS Units (Limits) dBFS dBc dBc DC and Logic Electrical Characteristics Unless otherwise specified, the following specifications apply: AGND = DGND = DRGND = D18GND = 0V, VA = VD = VDR = +3.3V, VD18 = +1.8V, Internal VREF = +1.0V, fCLK = 65 MHz, VCM = VCOM, tR = tF = TBD ns, CL = 5 pF/pin. CIC Decimation = 48, F2 Decimation = 2. Typical values are for TA = 25°C. Boldface limits apply for TMIN ≤ TA ≤ TMAX. All other limits apply for TA = 25° C. Symbol Parameter Conditions Typical (Note 10) Limits Units (Limits) VIL Voltage input low 0.7 V (max) VIH Voltage input high 2.3 V (min) IOZ Input current 20 µA VOL Voltage output low (IOL = 7mA) 0.4 V (max) VOH Voltage output high (IOH = -7mA) 2.4 V (min) CIN Input capacitance 5.0 pF 121 mA (max) POWER SUPPLY CHARACTERISTICS IA ADC Analog Supply Current 65MSPS 96 IA ADC Analog Supply Current 52MSPS 84 ID ADC Digital Supply Current 65MSPS 24 ID ADC Digital Supply Current 52MSPS 20 IDR Digital Output Supply Current (Note 15) 65MSPS 14 IDR Digital Output Supply Current (Note 15) 52MSPS 10 ID18 Digital Core Supply Current 65MSPS 67 ID18 Digital Core Supply Current 52MSPS 53 PD65 Total Power Dissipation GSM Set, 65MSPS 560 PD52 Total Power Dissipation GSM Set, 52MSPS 485 PSRR www.national.com Power Supply Rejection Ratio Rejection of Full-Scale Error with VA = 3.0V vs. 3.6V 8 mA 28 mA (max) mA 18 mA (max) mA 78 mA (max) mA 793 mW (max) mW dB Unless otherwise specified, the following specifications apply: AGND = DGND = DRGND = D18GND = 0V, VA = VD = VDR = +3.3V (±10%), VD18 = +1.8V (±10%), Internal VREF = +1.0V, fCLK = 65 MHz, VCM = VCOM, tR = tF = 1 ns, CL = 5 pF/pin. CIC Decimation = 48, F2 Decimation = 2. Typical values are for TA = 25°C. Boldface limits apply for TMIN ≤ TA ≤ TMAX. All other limits apply for TA = 25°C. (Note 13) Parameter (CL=50pF) Symbol Min Typical (Note 10) Max Units Clock Input FCK Clock (CK) Frequency (Figure 6) 20 65 MHz tCKDC CK duty cycle, DCS off (Figure 6) 40 60 % tRF CK rise and fall times (VIL to VIH) (Figure 6) 2 ns NCO Tuning Resolution 0.02 Hz NCO Phase Resolution 0.005 o Control Interface tMRA MR Active Time (Figure 4) 4 CK periods tMRIC MR Inactive to first Control Port Access (Figure 4) 10 CK periods tMRSU MR Setup Time to CK (Figure 4) 6 ns tMRH MR Hold Time from CK (Figure 4) 2 ns tSISU SI Setup Time to CK (Figure 5) 6 ns tSIH SI Hold Time from CK (Figure 5) 2 ns tSIW SI Pulse Width (Figure 5) 4 CK periods DVGA Interface tSTIW A|BSTROBE Inactive Pulse Width (Figure 7) tGSTB A|BGAIN setup before A|BSTROBE (Figure 7) CK periods 2 ns 6 Parallel Output Interface tOENV POUT_EN Active to POUT[15:0] Valid (Figure 9) 12 ns tOENT POUT_EN Inactive to POUT[15:0] Tri-State (Figure 9) 10 ns tSELV PSEL[2:0] to POUT[15:0] Valid (Figure 10) 13 ns tPOV RDY to POUT[15:0] New Value Valid (Note 5) (Figure 11) 7 ns tDBG SCK to POUT[15:0], RDY, SFS, AOUT, BOUT Valid (Figure 12) 4 ns ns Serial Interface tSFSV SCK to SFS Valid (Note 3) (Figure 13) -2 1.6 3.5 tOV SCK to A|BOUT Valid (Note 4) (Figure 13) -2 1.7 3.5 tRDYW RDY Pulse Width (Figure 13) tDCMSU PSEL[2:0] Setup Time to SCK_IN (Figure 8) tDCMH PSEL[2:0] Hold Time from SCK_IN (Figure 8) tRDYV SCK to RDY valid (Figure 13) -3 1.8 ns 2 CK periods 3 1.4 ns 0.5 -0.9 ns 4 ns JTAG Interface tJPCO Propagation Delay TCK to TDO (Figure 14) 25 ns tJSCO Propagation Delay TCK to Data Out (Figure 14) 35 ns tJPDZ Disable Time TCK to TDO (Figure 14) 25 ns tJSDZ Disable Time TCK to Data Out (Figure 14) tJPEN Enable Time TCK to TDO (Figure 14) tJSEN tJSSU 35 ns 0 25 ns Enable Time TCK to Data Out (Figure 14) 0 35 ns Setup Time Data to TCK (Figure 14) 10 ns tJPSU Setup Time TDI, TMS to TCK (Figure 14) 10 ns tJSH Hold Time Data to TCK (Figure 14) 45 ns tJPH Hold Time TCK to TDI, TMS (Figure 14) 45 ns tJCH TCK Pulse Width High (Figure 14) 50 ns 9 www.national.com LM97593 AC Electrical Characteristics LM97593 Parameter (CL=50pF) Symbol tJCL TCK Pulse Width Low (Figure 14) JTAGFMAX TCK Maximum Frequency (Figure 14) Min Typical (Note 10) Max Units 10 MHz ns 40 Microprocessor Interface tCSU Control Setup before the controlling signal goes low (Figure 15) 5 ns tCHD Control hold after the controlling signal goes high (Figure 15) 5 ns tCSPW Controlling strobe pulse width (Write) (Figure 15) 30 tCDLY Control output delay controlling signal low to D (Read) (Figure 15) 30 ns tCZ Control tri-state delay after controlling signal high (Figure 15) 20 ns ns Note 1: Absolute Maximum Ratings indicate limits beyond which damage to the device may occur. Operating Ratings indicate conditions for which the device is guaranteed to be functional, but do not guarantee specific performance limits. For guaranteed specifications and test conditions, see the Electrical Characteristics. The guaranteed specifications apply only for the test conditions listed. Some performance characteristics may degrade when the device is not operated under the listed test conditions. Operation of the device beyond the maximum Operating Ratings is not recommended. Note 2: All voltages are measured with respect to GND = AGND = DGND = DRGND = 0V, unless otherwise specified. Note 3: When the input voltage at any pin exceeds the power supplies (that is, VIN < AGND, or VIN > VA), the current at that pin should be limited to ±25 mA. The ±50 mA maximum package input current rating limits the number of pins that can safely exceed the power supplies with an input current of ±25 mA to two. Note 4: The maximum allowable power dissipation is dictated by TJ,max, the junction-to-ambient thermal resistance, (θJA), and the ambient temperature, (TA), and can be calculated using the formula PD,max = (TJ,max - TA )/θJA. The values for maximum power dissipation listed above will be reached only when the device is operated in a severe fault condition (e.g. when input or output pins are driven beyond the power supply voltages, or the power supply polarity is reversed). Such conditions should always be avoided. Note 5: Human Body Model is 100 pF discharged through a 1.5 kΩ resistor. Machine Model is 220 pF discharged through 0 Ω. Note 6: Reflow temperature profiles are different for lead-free and non-lead-free packages. Note 7: The inputs are protected as shown below. Input voltage magnitudes above VA or below GND will not damage this device, provided current is limited per (Note 3). 30008770 Note 8: To guarantee accuracy, it is required that |VA–VD| ≤ 100 mV and separate bypass capacitors are used at each power supply pin. Note 9: With the test condition for VREF = +1.0V (2VP-P differential input), the 12-Bit LSB is 488 µV. Note 10: Typical figures are at TA = 25°C and represent most likely parametric norms at the time of product characterization. The typical specifications are not guaranteed. Test Limits are guaranteed to National's AOQL (Average Outgoing Quality Level). Note 11: Integral Non Linearity is defined as the deviation of the analog value, expressed in LSBs, from the straight line that passes through positive and negative full-scale. Note 12: The input capacitance is the sum of the package/pin capacitance and the sample and hold circuit capacitance. Note 13: Timing specifications are tested at TTL logic levels, VIL = 0.4V for a falling edge and VIH = 2.4V for a rising edge. Note 14: Optimum performance will be obtained by keeping the reference input in the 0.8V to 1.2V range. The LM4051CIM3-ADJ (SOT23 package) is recommended for external reference applications. Note 15: IDR is the current consumed by the switching of the output drivers and is primarily determined by load capacitance on the output pins, the supply voltage, VDR, and the rate at which the outputs are switching (which is signal dependent). IDR=VDR(C0 x f0 + C1 x f1 +....C11 x f11) where VDR is the output driver power supply voltage, Cn is total capacitance on the output pin, and fn is the average frequency at which that pin is toggling. Note 16: (+x) indicates the additional dynamic range provided by the AGC. The DVGA in front of the LM97593 provides 42 dB of gain adjustment. www.national.com 10 LM97593 DDC Timing Diagrams 30008705 FIGURE 4. LM97593 Master Reset Timing 30008706 FIGURE 5. LM97593 Synchronization Input (SI) Timing 30008707 FIGURE 6. LM97593 Clock Timing 30008708 FIGURE 7. LM97593 DVGA Interface Timing 30008709 FIGURE 8. LM97593 Dual Chip Mode Timing 11 www.national.com LM97593 30008710 FIGURE 9. LM97593 Parallel Output Enable Timing 30008711 FIGURE 10. LM97593 Parallel Output Select Timing 30008712 FIGURE 11. LM97593 Parallel Output Data Ready Timing 30008713 FIGURE 12. LM97593 Debug Mode Timing www.national.com 12 LM97593 30008714 FIGURE 13. LM97593 Serial Port Timing 30008715 FIGURE 14. LM97593 JTAG Port Timing 13 www.national.com LM97593 30008716 FIGURE 15. LM97593 Control I/O Timing www.national.com 14 Unless otherwise specified, the following specifications apply for AGND = DGND = D18GND = DRGND = 0V, VA = VD = VDR = +3.3V, VD18 = +1.8V, PD = 0V, Internal VREF = +1.0V, fCLK = 65 MHz, fIN = 12 MHz, AIN = 0dBFS, CL = 10 pF/pin, Duty Cycle Stabilizer On. Boldface limits apply for TJ = TMIN to TMAX: all other limits TJ = 25°C DNL INL 300087100 300087101 15 www.national.com LM97593 ADC Typical Performance Characteristics DNL, INL LM97593 ADC Typical Performance Characteristics Unless otherwise specified, the following specifications apply for AGND = DGND = D18GND = DRGND = 0V, VA = VD = VDR = +3.3V, VD18 = +1.8V, PD = 0V, Internal VREF = +1.0V, fCLK = 65 MHz, fIN = 249 MHz, AIN = -9dBFS, CL = 10 pF/pin, Duty Cycle Stabilizer On. Boldface limits apply for TJ = TMIN to TMAX: all other limits TJ = 25°C SNR, SINAD, SFDR vs. VSUPPLY Distortion vs. VSUPPLY 300087102 300087103 Gain Tracking Error vs. VSUPPLY SNR, SINAD, SFDR vs. Temperature 300087105 300087104 Distortion vs. Temperature Gain Tracking Error vs. Temperature 300087106 www.national.com 300087107 16 LM97593 SNR, SINAD, SFDR vs. fIN (fCLK = 52MSPS, AIN = -9dBFS) Distortion vs. fIN (fCLK = 52MSPS, AIN = -9dBFS) 300087110 300087111 SNR, SINAD, SFDR vs. fIN (fCLK = 52MSPS, AIN = -3dBFS) Distortion vs. fIN (fCLK = 52MSPS, AIN = -3dBFS) 300087112 300087113 SNR, SINAD, SFDR vs. fIN (fCLK = 65MSPS, AIN = -9dBFS) Distortion vs. fIN (fCLK = 65MSPS, AIN = -9dBFS) 300087114 300087115 17 www.national.com LM97593 Distortion vs. fIN (fCLK = 65MSPS, AIN = -3dBFS) Distortion vs. fIN (fCLK = 65MSPS, AIN = -3dBFS) 300087116 300087117 Spectral Response @ 20MHz Input (fCLK = 52MSPS, AIN = -9dBFS) Spectral Response @ 20MHz Input (fCLK = 52MSPS, AIN = -3dBFS) 300087118 300087119 Spectral Response @ 20MHz Input (fCLK = 65MSPS, AIN = -9dBFS) Spectral Response @ 20MHz Input (fCLK = 65MSPS, AIN = -3dBFS) 300087120 www.national.com 300087121 18 LM97593 Spectral Response @ 250MHz Input (fCLK = 52MSPS, AIN = -9dBFS) Spectral Response @ 250MHz Input (fCLK = 52MSPS, AIN = -3dBFS) 300087122 300087123 Spectral Response @ 250MHz Input (fCLK = 65MSPS, AIN = -9dBFS) Spectral Response @ 250MHz Input (fCLK = 65MSPS, AIN = -3dBFS) 300087124 300087125 IMD: f1IN = 246MHz, f2IN = 250MHz (fCLK = 65MSPS, A1IN,2IN = -15dBFS) 300087149 19 www.national.com LM97593 CIC Output Typical Performance Characteristics Unless otherwise specified, the following specifications apply for AGND = DGND = D18GND = DRGND = 0V, VA = VD = VDR = +3.3V, VD18 = +1.8V, PD = 0V, Internal VREF = +1.0V, fCLK = 65 MHz, fIN = 249 MHz, AIN = -9dBFS, CL = 10 pF/pin, Duty Cycle Stabilizer On. Boldface limits apply for TJ = TMIN to TMAX: all other limits TJ = 25°C Spectral Response @ 20MHz Input (fCLK = 52MSPS, AIN = -3dBFS) Spectral Response @ 20MHz Input (fCLK = 52MSPS, AIN = -9dBFS) 300087126 300087127 Spectral Response @ 20MHz Input (fCLK = 65MSPS, AIN = -9dBFS) Spectral Response @ 20MHz Input (fCLK = 65MSPS, AIN = -3dBFS) 300087128 300087129 Spectral Response @ 250MHz Input (fCLK = 52MSPS, AIN = -9dBFS) Spectral Response @ 250MHz Input (fCLK = 52MSPS, AIN = -3dBFS) 300087130 www.national.com 300087131 20 LM97593 Spectral Response @ 250MHz Input (fCLK = 65MSPS, AIN = -9dBFS) Spectral Response @ 250MHz Input (fCLK = 65MSPS, AIN = -3dBFS) 300087132 300087133 21 www.national.com LM97593 DDC Output Typical Performance Characteristics Unless otherwise specified, the following specifications apply for AGND = DGND = D18GND = DRGND = 0V, VA = VD = VDR = +3.3V, VD18 = +1.8V, PD = 0V, CIC decimation = 8, Internal VREF = +1.0V, fCLK = 65 MHz, fIN = 249 MHz, AIN = -9dBFS, CL = 10 pF/pin, Duty Cycle Stabilizer On. Boldface limits apply for TJ = TMIN to TMAX: all other limits TJ = 25°C Spectral Response @ 20MHz Input (fCLK = 52MSPS, AIN = -3dBFS) Spectral Response @ 20MHz Input (fCLK = 52MSPS, AIN = -9dBFS) 300087134 300087135 Spectral Response @ 20MHz Input (fCLK = 65MSPS, AIN = -9dBFS) Spectral Response @ 20MHz Input (fCLK = 65MSPS, AIN = -3dBFS) 300087136 300087137 Spectral Response @ 250MHz Input (fCLK = 52MSPS, AIN = -9dBFS) Spectral Response @ 250MHz Input (fCLK = 52MSPS, AIN = -3dBFS) 300087138 www.national.com 300087139 22 LM97593 Spectral Response @ 250MHz Input (fCLK = 65MSPS, AIN = -9dBFS) Spectral Response @ 250MHz Input (fCLK = 65MSPS, AIN = -3dBFS) 300087140 300087141 23 www.national.com LM97593 Functional Description 30008717 FIGURE 16. LM97593 Dual ADC / Digital Tuner / AGC Block Diagram with Control Register Associations The LM97593 contains two identical 12-bit ADCs driving the digital down-conversion (DDC) circuitry shown in the block diagram in Figure 16. filters. The final filter outputs can be converted to a 12-bit floating point format or standard two’s complement format. The output data is available at both serial and parallel ports. The LM97593’s DDC maintains over 100 dB of spurious free dynamic range and over 100 dB of out-of-band rejection. This allows considerable latitude in channel filter partitioning between the analog and digital domains. The frequencies, phase offsets, and phase dither of the two sine/cosine numerically controlled oscillators (NCOs) can be independently specified. Two sets of coefficient memories and a crossbar switch allow shared or independent filter coefficients and bandwidth for each channel. Both channels share the same decimation ratio and input/output formats. Each channel has its own AGC circuit for use with narrowband radio channels where most of the channel filtering precedes the ADC. The AGC closes the loop around the DVGA, compressing the dynamic range of the signal into the ADC. AGC gain compensation in the LM97593 removes the DVGA gain steps at the output. The time alignment of this gain compensation circuit can be adjusted. The AGC can be configured to operate continuously or set to a fixed gain. The two AGC circuits operate independently but share the same programmed parameters and control signals. The chip receives configuration and control information over a microprocessor-compatible bus consisting of an 8-bit data I/O port, an 8-bit address port, a chip enable strobe, a read strobe, and a write strobe. The chip’s control registers (8 bits each) are memory mapped into the 8-bit address space of the ADC The ADCs operate off of a +3.3V supply and use a pipeline architecture with error correction circuitry to help ensure maximum performance. The differential analog input signal is digitized to 12 bits. The user has the choice of using an internal 1.0 Volt or an external reference. Any external reference is buffered on-chip to ease the task of driving that pin. The clock frequency is rated up to 65 MHz. The analog input for both channels is acquired at the rising edge of the clock. The digital data for a given sample is delayed by the pipeline for 7 clock cycles before it reaches the input to the DDC circuit. A logic high on the power down (PD) pin reduces the converter power consumption to 50 mW. The DDC power can be further reduced by gating off the clock as described in section 7.0 Power Management. DDC Each independent DDC channel down converts the sub-sampled IF to baseband, decimates the signal rate by a programmable factor ranging from 32 to 16384, provides channel filtering, and outputs quadrature symbols. A crossbar switch enables either of the two inputs or a test register to be routed to either DDC channel. Flexible channel filtering is provided by the two programmable decimating FIR www.national.com 24 close to the reference input pin. There is no need to bypass the VREF pin when the internal reference is used. 1.3 Signal Inputs The signal inputs are VIN A+ and VINA− for one ADC and VINB+ and VINB− for the other ADC . The input signal, VIN, is defined as VIN A = (VINA+) – (VINA−) ADC Application Information (Eq. 1) for the "A" converter and VIN B = (VINB+) – (VINB−) 1.0 ADC OPERATING CONDITIONS We recommend that the following conditions be observed for operation: 3.0V ≤ VA ≤ 3.6V VD = VA = VDR VD18 = 1.8V 10 MHz ≤ fCLK ≤ 65 MHz 1.0 V internal reference VCM = 1.5V (from VCOMA and VCOMB) (Eq. 2) for the "B" converter. Figure 17 shows the expected input signal range. Note that the common mode input voltage, VCM, should be in the range of 1.0V to 2.0V. The peaks of the individual input signals should never exceed 2.6V. The ADC performs best with a differential input signal with each input centered around a common mode voltage, VCM. The peak-to-peak voltage swing at each analog input pin should not exceed the value of the reference voltage or the output data will be clipped. The two input signals should be exactly 180° out of phase from each other and of the same amplitude. For single frequency inputs, angular errors result in a reduction of the effective full scale input. For complex waveforms, however, angular errors will result in distortion. 1.1 Analog Inputs There is one reference input pin, VREF, which is used to select an internal reference, or to supply an external reference. The ADC has two analog signal input pairs, VIN A+ and VIN A- for one converter and VIN B+ and VIN B- for the other converter. Each pair of pins forms a differential input pair. 1.2 Reference Pins The ADC is designed to operate with an internal 1.0V reference or an external 1.0V reference, but performs well with external reference voltages in the range of 0.8V to 1.2V. Lower reference voltages will decrease the signal-to-noise ratio (SNR) of the ADC. Increasing the reference voltage (and the input signal swing) beyond 1.2V may degrade THD for a fullscale input, especially at higher input frequencies. It is important that all grounds associated with the reference voltage and the analog input signal make connection to the ground plane at a single, quiet point to minimize the effects of noise currents in the ground path. The six Reference Bypass Pins (VRPA, VCOMA, VRNA, VRPB, VCOMB and VRNB) are made available for bypass purposes. All these pins should each be bypassed to ground with a 0.1 µF capacitor. A 10 µF capacitor should be placed between the VRPA and VRNA pins and between the VRPB and VRNB pins, as shown in Figure 45. This configuration is necessary to avoid reference oscillation, which could result in reduced SFDR and/or SNR. Smaller capacitor values than those specified will allow faster recovery from the power down mode, but may result in degraded noise performance. Loading any of these pins other than VCOMA and VCOMB may result in performance degradation. The nominal voltages for the reference bypass pins are as follows: VCOM = 1.5 V VRP = VCOM + VREF / 2 VRN = VCOM − VREF / 2 User choice of an on-chip or external reference voltage is provided. The internal 1.0 Volt reference is in use when the the VREF pin is connected to VA. If a voltage in the range of 0.8V to 1.2V is applied to the VREF pin, that is used for the voltage reference. When an external reference is used, the VREF pin should be bypassed to ground with a 0.1 µF capacitor 30008750 FIGURE 17. Expected Input Signal Range For single frequency sine waves the full scale error in LSBs can be described as approximately EFS = 4096 ( 1 - sin (90° + dev)) (Eq. 3) Where dev is the angular difference in degrees between the two signals having a 180° relative phase relationship to each other (see Figure 18). Drive the analog inputs with a source impedance less than 100Ω. 30008751 FIGURE 18. Angular Errors Between the Two Input Signals Will Reduce the Output Level or Cause Distortion For differential operation, each analog input pin of the differential pair should have a peak-to-peak voltage equal to the 25 www.national.com LM97593 control port. Page select bits allow access to the overlaid A and B set of FIR coefficients. JTAG boundary scan and on-chip diagnostic circuits are provided to simplify system debug and test. The LM97593 supports 3.3V I/O even though the core logic voltage is 1.8V. The LM97593 outputs swing to the 3.3V rail so they can be directly connected to 5V TTL inputs if desired. LM97593 reference voltage, VREF, be 180 degrees out of phase with each other and be centered around VCM. line. Refer to Application Note AN-905 for information on setting characteristic impedance. It is highly desirable that the the source driving the ADC CLK pin only drive that pin. However, if that source is used to drive other things, each driven pin should be a.c. terminated with a series RC to ground such that the resistor value is equal to the characteristic impedance of the clock line and the capacitor value is 1.3.1 Single-Ended Operation Performance with differential input signals is better than with single-ended signals. For this reason, single-ended operation is not recommended. However, if single ended-operation is required and the resulting performance degradation is acceptable, one of the analog inputs should be connected to the d.c. mid point voltage of the driven input. The peak-to-peak input signal at the driven input pin should be twice the reference voltage to maximize SNR and SINAD performance (Figure 17b). For example, set VREF to 1.0V, bias VIN− to 1.5V and drive VIN+ with a signal range of 0.5V to 2.5V. Because very large input signal swings can degrade distortion performance, better performance with a single-ended input can be obtained by reducing the reference voltage when maintaining a full-range output. (Eq. 4) where tPD is the signal propagation time down the clock line, "L" is the line length and ZO is the characteristic impedance of the clock line. This termination should be as close as possible to the ADC clock pin but beyond it as seen from the clock source. Typical tPD is about 150 ps/inch (60 ps/cm) on FR-4 board material. The units of "L" and tPD should be the same (inches or centimeters). The duty cycle of the clock signal can affect the performance of the A/D Converter. Because achieving a precise duty cycle is difficult, the LM97593 has a Duty Cycle Stabilizer which can be enabled using the REFSEL/DCS pin. It is designed to maintain performance over a clock duty cycle range of 30% to 70% at 65 MSPS. 1.3.2 Driving the Analog Inputs The VIN+ and the VIN− inputs of the ADC consist of an analog switch followed by a switched-capacitor amplifier. As the internal sampling switch opens and closes, current pulses occur at the analog input pins, resulting in voltage spikes at the signal input pins. As the driving source attempts to counteract these voltage spikes, it may add noise to the signal at the ADC analog input. C1, C2, and C3 as shown in Figure 45 improve the ADC performance by filtering these voltage spikes. These components should be placed close to the ADC inputs because the input pins of the ADC are the most sensitive part of the system and this is the last opportunity to filter that input. For Nyquist applications the RC pole should be at the ADC sample rate. The ADC input capacitance in the sample mode should be considered when setting the RC pole. For wideband undersampling applications, the RC pole should be set at about 1.5 to 2 times the maximum input frequency to maintain a linear delay response. The values of the RC shown in Figure 45 are suitable for applications with input frequencies up to approximately 70MHz. 2.2 REFSEL/DCS This pin is used in conjunction with VREF (pin 21) to select the reference source and turn the Duty Cycle Stabilizer (DCS) on or off. When REFSEL/DCS is LOW and VREF is HIGH, the internal 1.0V reference is selected and DCS is On. When REFSEL/DCS is HIGH, an external reference voltage in the range of 0.8V to 1.2V should be applied to the VREF input. DCS is On. With REFSEL/DCS pin connected to VCOMA or VCOMB, the internal 1.0V reference is selected and DCS is Off. When enabled, duty cycle stabilization can compensate for clock inputs with duty cycles ranging from 30% to 70% and generate a stable internal clock, improving the performance of the part. 1.3.3 Input Common Mode Voltage The input common mode voltage, VCM, should be in the range of 1.0V to 2.0V and be a value such that the peak excursions of the analog signal do not go more negative than ground or more positive than 2.6V. See Section 1.2. TABLE 1. VREF, REFSEL/DCS Pin Functions 2.0 DIGITAL INPUTS Digital TTL/CMOS compatible inputs consist of CK, REFSEL/ DCS. 2.1 CLK The CLK signal controls the timing of the sampling process. Drive the clock input with a stable, low jitter clock signal in the range of 10 MHz to 65 MHz. The higher the input frequency, the more critical it is to have a low jitter clock. The trace carrying the clock signal should be as short as possible and should not cross any other signal line, analog or digital, not even at 90°. The CLK signal also drives an internal state machine. If the CLK is interrupted, or its frequency too low, the charge on internal capacitors can dissipate to the point where the accuracy of the output data will degrade. This is what limits the lowest sample rate. The clock line should be terminated at its source in the characteristic impedance of that line. Take care to maintain a constant clock line impedance throughout the length of the www.national.com REFSEL/ DCS (pin 8) VREF (pin 21) Reference DCS Logic Low Logic High Internal 1.0 V ON Logic High 0.8 to 1.2V External VCOMA or VCOMB Logic High Internal 1.0V OFF VCOMA or VCOMB 0.8 to 1.2V External ON OFF 2.3 PD The PD pin, when high, holds the ADC in a power-down mode to conserve power when the converter is not being used. The output data pins are undefined and the data in the pipeline is corrupted while in the power down mode. The Power Down Mode Exit Cycle time is determined by the value of the components on pins 15, 16, 17, 22, 23 and 24. These capacitors lose their charge in the Power Down mode and must be recharged by on-chip circuitry before conversions can be accurate. Smaller capacitor values allow slightly faster recovery from the power down mode, but can result in a reduction in SNR, SINAD and ENOB performance. 26 3.0 CONTROL INTERFACE The LM97593 is configured by writing control information into 237 control registers within the chip. The contents of these control registers and how to use them are described in section 9.1 Control Register Addresses and Defaults. The registers are written to or read from using the D[7:0], A[7:0], CE, RD and WR pins. This interface is designed to allow the LM97593 to appear to an external processor as a memory mapped peripheral. See Figure 15 for details. The control interface is asynchronous with respect to the system clock, CK. This allows the registers to be written or read at any time. In some cases this might cause an invalid operation since the interface is not internally synchronized. In order to assure correct operation, SI must be asserted after the control registers are written. The D[7:0], A[7:0], WR, RD and CE pins should not be driven above the positive supply voltage. 4.0 DOWN CONVERTERS A detailed block diagram of each DDC channel is shown in Figure 19. Each down converter uses a complex NCO and mixer to quadrature downconvert a signal to baseband. The “FLOAT TO FIXED CONVERTER” treats the 15-bit mixer output as a mantissa and the AGC output, EXP, as a 3-bit exponent. It performs a bit shift on the data based on the value of EXP. This bit shifting is used to expand the compressed dynamic range resulting from the DVGA operation. The DVGA gain is adjusted in 6dB steps which are equivalent to each digital bit shift. Digitally compensating for the DVGA gain steps in the LM97593 causes the DDC output to be linear with respect to the DVGA input. The AGC operation will be completely transparent at the LM97593 output. The exponent (EXP) can be forced to its maximum value by setting the EXP_INH bit. If x in(n) is the DDC input, the signal after the “FLOAT TO FIXED CONVERTER” is 3.1 Master Reset A master reset pin, MR, is provided to initialize the LM97593 to a known condition and should be strobed after power up. This signal will clear all sample data and all user programmed data (filter coefficients and AGC settings). All outputs will be disabled (tri-stated). ASTROBE and BSTROBE will be asserted to initialize the DVGA values. Section 9.1 Control Register Addresses and Defaults describes the control register default values. x3(n) = xin(n)*cos(ωn)*2EXP (Eq. 5) for the I component. Changing the ‘cos’ to ‘sin’ in this equation will provide the Q component. The “FLOAT TO FIXED CONVERTER” circuit expands the dynamic range compression performed by the DVGA. Signals from this point onward extend across the full dynamic range of the signals applied to the DVGA input. This allows the AGC to operate continuously through a burst without producing artifacts in the signal due to the settling response of the decimation filters after a 6dB DVGA gain adjustment. For example, if the DVGA input signal were to increase causing the ADC output level to cross the AGC threshold level, the gain of the DVGA would change by -6dB. The 6dB step is allowed to propagate through the ADC and mixers and is compensated out just before the filtering. The accuracy of 3.2 Synchronizing Multiple LM97593 Chips A system containing two or more LM97593 chips will need to be synchronized if coherent operation is desired. To synchronize multiple LM97593 chips, connect all of the sync input pins together so they can be driven by a common sync strobe. Synchronization occurs on the first rising edge of CK after SI goes high. When SI is asserted all sample data is immediately cleared, the numerically controlled oscillator (NCO) phase offset is initialized, the NCO dither generators are reset, and the CIC decimation ratio is initialized. Only the configuration data loaded into the microprocessor interface remains unaffected. SI may be held low as long as desired after a minimum of 4 CK periods. 30008718 FIGURE 19. LM97593 Down Converter, Channel A (Channel B is identical) 27 www.national.com LM97593 3.3 Input Source The input crossbar switch allows either VINA, VINB, or a test register to be routed to the channel A or channel B AGC/ DDC. The AGC outputs, AGAIN and BGAIN, are not switched. If VINA and VINB are exchanged the AGC loop will be open and the AGC will not function properly. Selecting the test register as the input source allows the AGC or DDC operation to be verified with a known input. See section 8.0 Test and Diagnostics for further discussion. DDC Application Information LM97593 formance for GSM systems. The user can also design and download their own final filter to customize the channel’s spectral response. Typical uses of programmable filter F2 include matched (root-raised cosine) filtering, or filtering to generate oversampled outputs with greater out of band rejection. The 63 tap symmetrical filter is downloaded into the chip as 32 words, 16 bits each. Saturation to plus or minus full scale is performed at the output of F1 and F2 to clip the signal rather than allow it to roll over. The LM97593 provides two sets of coefficient memory for both F1 and F2. These coefficient memories can be independently routed to channel A, channel B, or both channel A and B with a crossbar switch. The coefficients can be switched on the fly but some time will be required before valid output data is available. the compensation is dependent on timing and the accuracy of the DVGA gain step. The LM97593 allows the timing of the gain compensation to be adjusted in the EXT_DELAY register; see the end of section 6.0 AGC for more information. The AGC requires 21 bits (14-bit internal bus output + 7-bit shift) to represent the full linear dynamic range of the signal. The output word must be set to either 24-bit or 32-bit to take advantage of the entire dynamic range available. The LM97593 can also be configured to output a floating point format with up to 138dB of numerical resolution using only 12 output bits. The “SHIFT UP” circuit will be discussed in the section 4.2 Four Stage CIC Filter. A 4-stage cascaded-integrator-comb (CIC) filter and a twostage decimate by 4 or 8 finite impulse response (FIR) filter are used to lowpass filter and isolate the desired signal. The CIC filter reduces the sample rate by a programmable factor ranging from 8 to 2048 (decimation ratio). The CIC outputs are followed by a gain stage and then followed by a two-stage decimate by 4 or 8 filter. The gain circuit allows the user to boost the gain of weak signals by up to 42 dB in 6 dB steps. It also rounds the signal to 21 bits and saturates at plus or minus full scale. The first stage of the two stage filter is a 21-tap, symmetric decimate by 2 FIR filter (F1) with programmable 16 bit tap weights. The coefficients of the first 11 taps are downloaded to the chip as 16 bit words. Since the filter is a symmetric configuration only the first 11 coefficients must be loaded. Section 4.4 First Programmable FIR Filter provides a generic set of coefficients that compensate for the rolloff of the CIC filter and provide a passband flat to 0.01dB with 70 dB of out of band rejection. A second coefficient set is provided that has a narrower output passband and greater out-of-band rejection. The second set of coefficients is ideal for systems such as GSM where far-image rejection is more important than adjacent channel rejection. The second stage is a 63 tap decimate by 2 or 4 programmable FIR filter (F2) also with 16 bit tap weights. Filter coefficients for a flat response from -0.4FS to +0.4FS of the 4.1 The Numerically Controlled Oscillator The tuning frequency of each down converter is specified as a 32 bit word (.02Hz resolution at CK=52MHz) and the phase offset is specified as a 16 bit word (.005o). These two parameters are applied to the Numerically Controlled Oscillator (NCO) circuit to generate sine and cosine signals used by the digital mixer. The NCOs can be synchronized with NCOs on other chips via the sync pin SI. This allows multiple down converter outputs to be coherently combined, each with a unique phase and amplitude. The tuning frequency is set by loading the FREQ register according to the formula FREQ = 232F/FCK, where F is the desired tuning frequency and FCK is the chip’s clock rate. FREQ is a 2’s complement word. The range for F is from -FCK/2 to +FCK(1-2-31)/2. If a sub-sampled signal is in an even Nyquist zone the sampling process causes the order of the I and Q components to be reversed. Should this occur simply invert the polarity of the tuning frequency F. Complex NCO Output Phase Dither Enabled Complex NCO Output Phase Dither Disabled 30008799 FIGURE 21. Example of NCO spurs due to phase truncation (After Phase Dithering) 30008719 FIGURE 20. Example of NCO spurs due to phase truncation (Before Phase Dithering) The 2’s complement format represents full-scale negative as 10000000 and full-scale positive as 01111111 for an 8-bit example. The 16 bit phase offset is set by loading the PHASE register according to the formula PHASE = 216P/2π, where P is the desired phase in radians ranging between 0 and 2π. PHASE output sample rate with 80dB of out of band rejection are provided in Section 4.5 Second Programmable FIR Filter. A second set of F2 coefficients is also provided to enhance perwww.national.com 28 Complex NCO Output Phase Dither Disabled 30008720 FIGURE 22. NCO Spurs due to Phase Quantization Complex NCO Output Phase Dither Enabled 4.2 Four Stage CIC Filter The mixer outputs are decimated by a factor of N in a four stage CIC filter. N is programmable to any integer between 8 and 2048. Decimation is programmed in the DEC register where DEC = N - 1. The programmable decimation allows the chip’s usable output bandwidth to range from about ±1.27kHz to ±650kHz when the input data rate (which is equal to the chip’s clock rate, FCK) is 52 MHz. For the maximum sample rate of 65MHz, the LM97593’s output bandwidth will range from about ±1.58kHz to ±812kHz. A block diagram of the CIC filter is shown in Figure 24. The CIC filter is primarily used to decimate the high-rate incoming data while providing a rough lowpass characteristic. The lowpass filter will have a sin(x)/x response (similar to the AGC’s CIC shown in Figure 39) where the first null is at FS/N. 30008721 FIGURE 23. Worst Case Amplitude Spur, NCO at FS/8 The CIC filter has a gain equal to N4 (filter decimation^4) which must be compensated for in the “SHIFT UP” circuit shown in Figure 24 as well as Figure 19. This circuit has a gain equal to 2(SCALE-44), where SCALE ranges from 0 to 40. This circuit divides the input signal by 244 providing 30008722 FIGURE 24. Four-stage decimate by N CIC filter maximum headroom through the CIC filter. For optimal noise performance the SCALE value is set to increase this level until the CIC filter is just below the point of distortion. A value is normally calculated and loaded for SCALE such that 29 www.national.com LM97593 is an unsigned 16-bit number. P ranges from 0 to 2π(1-2-16). Phase dithering can be enabled to reduce the spurious signals created by the NCO due to phase truncation. This truncation is unavoidable since the frequency resolution is much finer than the phase resolution. With dither enabled, spurs due to phase truncation are below -100 dBc for all frequencies and phase offsets. Each NCO has its own dither source and the initial state of one is maximally offset with respect to the other so that they are effectively uncorrelated. The phase dither sources are on by default. They are independently controlled by the DITH_A and DITH_B bits. The amplitude resolution of the ROM creates a worst-case spur amplitude of -101dBc rendering amplitude dither unnecessary. The spectrum plots in Figure 20 and Figure 21show the effectiveness of phase dither in reducing NCO spurs due to phase truncation for a worst-case example (just below FS/8). With dither off, the spur is at -86.4dBFS. With dither on, the spur is below -125dBFS, disappearing into the noise floor. This spur is spread into the noise floor which results in an SNR of -83.6dBFS. The channel filter’s processing gain will further improve the SNR. Figure 22 shows the spur levels as the tuning frequency is scanned over a narrow portion of the frequency range. The spurs are again a result of phase quantization but their locations move about as the frequency scan progresses. As before, the peak spur level drops when dithering is enabled. When dither is enabled and the fundamental frequency is exactly at FS/8, the worst-case spur due to amplitude quantization can be observed at -101dBc in Figure 23. LM97593 GAINSHIFTUP*GAINCIC ≤ 1. The actual gain of the CIC filter will only be unity for power-of-two decimation values. In other cases the gain will be somewhat less than unity. 4.3 Channel Gain The gain of each channel can be boosted up to 42 dB by shifting the output of the CIC filter left by 0 to 7 bits prior to rounding it to 21 bits. For channel A, the gain of this stage is: GAIN = 2GAIN_A , where GAIN_A ranges from 0 to 7. Overflow due to the GAIN circuit is saturated (clipped) at plus or minus full scale. Each channel can be given its own GAIN setting. 4.4 First Programmable FIR Filter (F1) The CIC/GAIN outputs are followed by two stages of filtering. The first stage is a 21 tap decimate-by-2 symmetric FIR filter with programmable coefficients. Typically, this filter compensates for a slight droop induced by the CIC filter while removing undesired alias images above Nyquist. In addition, it often provides stopband assistance to F2 when deep stop bands are required. The filter coefficients are 16-bit 2’s complement numbers. Unity gain will be achieved through the filter if the sum of the 21 coefficients is equal to 216. If the sum is not 216, then F1 will introduce a gain equal to (sum of coefficients)/ 216. The 21 coefficients are identified as coefficients h1(n), n = 0, ..., 20 where h1(10) is the center tap. The coefficients are symmetric, so only the first 11 are loaded into the chip. Two example sets of coefficients are provided here. The first set of coefficients, referred to as the standard set (STD), compensates for the droop of the CIC filter providing a passband which is flat (0.01 dB ripple) over 95% of the final output bandwidth with 70dB of out-of-band rejection (see Figure 25). The filter has a gain of 0.999 and is symmetric with the following 11 unique taps (1|21, 2|20, ..., 10|12, 11): 29, -85, -308, -56, 1068, 1405, -2056, -6009, 1303, 21121, 32703 30008724 FIGURE 26. F1 GSM frequency response 4.5 Second Programmable FIR Filter (F2) The second stage decimate by two or four filter also uses externally downloaded filter coefficients. F2 determines the final channel filter response. The filter coefficients are 16-bit 2’s complement numbers. Unity gain will be achieved through the filter if the sum of the 63 coefficients is equal to 216. If the sum is not 216, then the F2 will introduce a gain equal to (sum of coefficients)/216. The 63 coefficients are identified as h2(n), n = 0, ..., 62 where h2(31) is the center tap. The coefficients are symmetric, so only the first 32 are loaded into the chip. An example filter (STD F2 coefficients, see Figure 27) with 80dB out-of-band rejection, gain of 1.00, and 0.03 dB peak to peak passband ripple is created by this set of 32 unique coefficients: -14, -20, 19, 73, 43, -70, -82, 84, 171, -49, -269, -34, 374, 192, -449, -430, 460,751, -357, -1144, 81, 1581, 443, -2026, -1337, 2437, 2886, -2770, -6127, 2987, 20544, 29647 A second set of F2 coefficients (GSM set, see Figure 28) suitable for meeting the stringent wideband GSM requirements with a gain of 0.999 are: -536, -986, 42, 962, 869, 225, 141, 93, -280, -708, -774, -579, -384, -79, 536, 1056, 1152, 1067, 789, 32, -935, -1668, -2104, -2137, -1444, 71, 2130, 4450, 6884, 9053, 10413, 10832 The filter coefficients of both filters can be used to tailor the spectral response to the user’s needs. For example, the first can be loaded with the standard set to provide a flat 30008723 FIGURE 25. F1 STD frequency response The second set of coefficients (GSM set) are intended for applications that need deeper stop bands or need oversampled outputs. These requirements are common in cellular systems where out of band rejection requirements can exceed 100dB (see Figure 26). They are useful for wideband radio architectures where the channelization is done after the ADC. These filter coefficients introduce a gain of 0.984 and are: -49, -340, -1008, -1617, -1269, 425, 3027, 6030, 9115, 11620, 12606 www.national.com 30 LM97593 30008725 30008727 FIGURE 27. F2 STD frequency response FIGURE 29. CIC, F1, & F2 STD frequency response 30008726 30008728 FIGURE 28. F2 GSM frequency response FIGURE 30. CIC, F1, & F2 STD Passband Flatness response through to the second filter. The latter can then be programmed as a Nyquist (typically a root-raised-cosine) filter for matched filtering of digital data. The complete channel filter response for standard coefficients is shown in Figure 29. Passband flatness is shown in Figure 30. The complete filter response for GSM coefficients is shown in Figure 31. GSM Passband flatness is shown in Figure 32. The mask shown in Figure 31 is derived from the ETSI GSM 5.05 specifications for a normal Basestation Transceiver (BTS). For interferers, 9dB was added to the carrier to interference (C/I) ratios. For blockers, 9dB was added to the difference between the blocker level and 3dB above the reference sensitivity level. 4.6 Channel Bandwidth vs. Sample Rate When the LM97593 is used for GSM systems, a bandwidth of about 200kHz is desired. With a sample rate of 52MHz, the total decimation of 192 provides the desired 270.833kHz output sample rate. This output sample rate in combination with the FIR filter coefficients create the desired channel bandwidth. If the sample rate is increased to 65MHz, the decimation must also be increased to 65MHz/270.833kHz or 240. This new decimation rate will maintain the same output 30008729 FIGURE 31. CIC, F1, & F2 GSM frequency response bandwidth. The output bandwidth may only be changed in relation to the output sample rate by creating a new set of FIR filter coefficients. As the filter bandwidth 31 www.national.com LM97593 wave inputs the 1/2 should be set to 1 to prevent signal distortion. 4.8 Data Latency and Group Delay The LM97593 latency calculation assumes that the FIR filter latency will be equal to the time required for data to propagate through one half of the taps. The CIC filter provides 4N equivalent taps where N is the CIC decimation ratio. F1 and F2 provide 21 and 63 taps respectively. When these filters are reflected back to the input rate, the effective taps are increased by decimation. This results in a total of 151N taps. The total latency is found by dividing the number of taps by 2 and adding pipeline delays. When the F2 decimation is 2 the latency is 80N. When the F2 decimation is 4 the latency is 82N. The LM97593 filters are linear phase filters so the group delay remains constant. 30008730 5.0 OUTPUT MODES After processing by the DDC, the data is then formatted for output. All output data is two’s complement. The serial outputs power up in a tri-state condition and must be enabled when the chip is configured. Parallel outputs are enabled by the POUT_EN pin. Output formats include truncation to 8 or 32 bits, rounding to 16 or 24 bits, and a 12-bit floating point format (4-bit exponent, 8-bit mantissa, 138dB numeric range). This function is performed in the OUTPUT CIRCUIT shown in Figure 33. FIGURE 32. CIC, F1, & F2 GSM Passband Flatness decreases relative to the output sample rate, the CIC droop compensation performed by F1 may no longer be required. 4.7 Overall Channel Gain The overall gain of the chip is a function of the amount of decimation (N), the settings of the “SHIFT UP” circuit (SCALE), the GAIN setting, the sum of the F1 coefficients, and the sum of the F2 coefficients. The overall gain is shown below in Equation 2. (Eq. 6) Where: (Eq. 7) and: (Eq. 8) 30008731 FIGURE 33. LM97593 output circuit It is assumed that the DDC output words are treated as fractional 2’s complement words. The numerators of GF1 and GF2 equal the sums of the impulse response coefficients of F1 and F2, respectively. For the STD and GSM sets, GF1 and GF2 are nearly equal to unity. Observe that the AGAIN term in (Eq. 6) is cancelled by the DVGA operation so that the entire gain of the DRCS is independent of the DVGA setting when EXP_INH=0. The 1/2 appearing in (Eq. 6) is the result of the 6dB conversion loss in the mixer. For full-scale square www.national.com The channel outputs are accessible through serial output pins and a 16-bit parallel output port. The RDY pin is provided to notify the user that a new output sample period (OSP) has begun. OSP refers to the interval between output samples at the decimated output rate. For example, if the input rate (and clock rate) is 52 MHz and the overall decimation factor is 192 (N=48, F2 decimation=2) the OSP will be 3.69 microseconds which corresponds to an output sample 32 LM97593 30008732 FIGURE 34. Serial output formats. Refer to Figure 13 for detailed timing information rate of 270.833kHz. An OSP starts when a sample is ready and stops when the next one is ready. output period. This SFS pulse will be coincident with RDY and only a single SCK period wide. The TDM modes are summarized in Table 2. 5.1 Serial Outputs The LM97593 provides a serial clock (SCK), a frame strobe (SFS) and two data lines (AOUT and BOUT) to output serial data. The MUX_MODE control register specifies whether the two channel outputs are transmitted on two separate serial pins, or multiplexed onto one pin in a time division multiplexed (TDM) format. Separate output pins are not provided for the I and Q halves of complex data. The I and Q outputs are always multiplexed onto the same serial pin. The I-component is output first, followed by the Q-component. By setting the PACKED mode bit to ‘1’ a complex pair may be treated as a single double-wide word. The RDY signal is used to identify the first word of a complex pair of the TDM formatted output when the SFS_MODE bit is set to ‘0’. Setting SFS_MODE to ‘1’ causes the LM97593 to output a single SFS pulse for each TABLE 2. TDM Modes SFS_MODE MUX_MODE 0 1 SERIAL OUTPUTS AOUT BOUT 0 OUTA OUTB 1 OUTA, OUTB LOW 0 OUTA OUTB 1 OUTA, OUTB LOW The serial outputs use the format shown in Figure 34. Figure 34(a) shows the standard output mode (the PACKED mode bit is low). The chip clocks the frame and data out of the chip 33 www.national.com LM97593 30008733 FIGURE 35. Serial Daisy-Chain Mode on the rising edge of SCK (or falling edge if the SCK_POL bit in the input control register is set high). Data should be captured on the falling edge of SCK (rising if SCK_POL=1). The chip sends the I data first by setting SFS high (or low if SFS_POL in the input control register is set high) for one clock cycle, and then transmitting the data, MSB first, on as many SCK cycles as are necessary. Without a pause, the Q data is transferred next as shown in Figure 34(a). If the PACKED control bit is high, then the I and Q components are sent as a double length word with only one SFS strobe as shown in Figure 34(b). If both channels are multiplexed out the same serial pin, then the subsequent I/Q channel words will be transmitted immediately following the first I/Q pair as shown in Figure 34(c). Figure 34(c) also shows how SFS_MODE=1 allows the SFS signal to be used to identify the A and B channels in the TDM serial transmission. The serial output rate is programmed by the RATE register to CK divided by 1, 2, 4, 8, 16, or 32. The serial interface will not work properly if the programmed rate of SCK is insufficient to clock out all the bits in one OSP. gardless of the state of MUX_MODE, and the data is sent as mI/eI/eQ/ mQ which allows the two exponents to form an 8bit word. This is shown in Figure 34(d). For all formats, once the defined length of the word is complete, SCK stops toggling. 5.4 Parallel Outputs Output data from the channels can also be taken from a 16bit parallel port. A 3-bit word applied to the POUT_SEL[2:0] pins determines which 16-bit segment is multiplexed to the parallel port. Table 3 defines this mapping. To allow for bussing of multiple chips, the parallel port is tri-stated unless POUT_EN is low. The RDY signal indicates the start of an OSP and that new data is ready at the parallel output. The user has one OSP to cycle through whichever registers are needed. The RATE register must be set so that each OSP is at least 5 SCK periods. 5.4.1 Parallel Port Output Numeric Formats The I/Q samples can be rounded to 16 or 24 bits or the full 32 bit word can be read. By setting the word size to 32 bits it is possible to read out the top 16-bits and only observe the top 8 bits if desired. Additionally, the output samples can be formatted as floating point numbers with an 8-bit mantissa and a 4 bit exponent. For the fixed-point formats, the valid bits are justified into the MSBs of the registers of Table 3 and 5.2 Serial Port Daisy-Chain Mode Two LM97593s can be connected in series so that a single DSP serial port can receive four DDC output channels. This mode is enabled by setting the SDC_EN bit to ‘1’ on the serial daisy-chan (SDC) master. The SDC master is the LM97593 which is connected to the DSP while the SDC slave’s serial output drives the master. The SDC master’s RATE register must be set so that its SCK rate is twice that of the SDC slave, the SDC master must have MUX_MODE=1, the SDC slave must have MUX_MODE=0 and PACKED=1, and both chips must come out of a MR or SI event within four CK periods of each other. In this configuration, the master’s serial output data is shifted out to the DSP and then the slave’s serial data is shifted out. All the serial output data will be muxed onto the master’s AOUT pin as shown in Figure 35. TABLE 3. Register Selection for Parallel Output 5.3 Serial Port Output Number Formats Several numeric formats are selectable using the FORMAT control register. The I/Q samples can be rounded to 16 or 24 bits, or truncated to 8 bits. The packed mode works as described above for these fixed point formats. A floating point format with 138dB of dynamic range in 12 bits is also provided. The mantissa (m) is 8 bits and the exponent (e) is 4 bits. The MSB of each segment is transmitted first. When the packed mode is selected, the I/Q samples are packed re- www.national.com POUT_SEL Normal Register Contents Floating Point Register Contents 0 IA upper 16-bits 0000/eIA/mIA 1 IA lower 16-bits 0x0000 2 QA upper 16-bits 0000/eQA/mQA 3 QA lower 16-bits 0x0000 4 IB upper 16-bits 0000/eIB/mIB 5 B lower 16-bits 0x0000 6 QB upper 16-bits 0000/eQB/mQB 7 QB lower 16-bits 0x0000 all other bits are set to zero. For the floating point format, the valid bits are placed in the upper 16-bits of the appropriate channel register using the format 0000/eI/mI for the I samples. 34 30008734 FIGURE 36. Output Gain Scaling vs. Input Signal In order to use the AGC, the DRCS Control Panel software may be used to calculate the programmable parameters. To generate these parameters, only the desired setpoint, deadband+ hysteresis, and loop time constant need to be supplied. All subsequent calculations are performed by the software. Complete details of the AGC operation are provided in an appendix. 30008735 FIGURE 37. AGC Setup AGC setpoint and deadband are illustrated in Figure 37. The loop time constant is a measure of how fast the loop will track 30008758 FIGURE 38. Function controlled by EXT_DELAY register 35 www.national.com LM97593 a changing signal. Values down to approximately 1.0 microsecond will be stable with the second order LC noise filter. Since the DVGA operates with 6dB steps the deadband should always be greater than 6dB to prevent oscillation. An increased deadband value will reduce the amount of AGC operation. A decreased deadband value will increase the amount of AGC operation but will hold the ADC output closer to the setpoint. The threshold should be set so that transients do not cause sustained overrange at the ADC inputs. The threshold setting can also be used to set the ADC input near its optimal performance level. The AGC will free run when AGC_HOLD_IC is set to ‘0’. It may be set to a fixed gain by setting AGC_HOLD_IC to ‘1’ after programming the desired gain in the AGC_IC_A and AGC_IC_B registers. Allowing the AGC to free run should be appropriate for most applications. Programming the AGC_COMB_ORD register allows the AGC power detector bandwidth to be reduced if desired. This will tend to improve the power detector’s ability to reject the signal carrier frequency and reduce overall AGC activity. Figure 39 shows the power detector response. The analog gain change from the DVGA must be compensated by the "Float To Fixed" converter after the appropriate delay. This delay can be adjusted by the EXT_DELAY register value to make sure the analog gain change is properly compensated in the digital domain. Figure 38 shows the internal clock latency paths related to the DVGA and "Float To Fixed" timing conpensation. In this diagram registers are represented by z-N where N is the sample delay in ADC clock periods. Following the path from the output of the AGC integrator through the DVGA, bandpass filter, ADC and internal register delays adds up to 6 clocks prior to the "Float To Fixed" converter excluding the bandpass filter and ADC. Following the path from the AGC integrator to the "Float To Fixed" is also 6 clocks when EXT_DELAY = 0. The value programmed in EXT_DELAY should be set to the pipeline latency of the ADC plus the latency of the bandpass filter (typically one clock). If ASTROBE and BSTROBE are not used then subtract one from the resulting total latency. The LM97593 includes an integrated ADC with a pipeline latency of 7 clocks. Adding one additional clock period for the bandpass filter requires EXT_DELAY = 7 when the DVGA ASTROBE and BSTROBE signals are used, otherwise, program EXT_DELAY = 6. In most cases ASTROBE and BSTROBE signals are not used so EXT_DELAY is typically set to 6. More accurate time alignment may improve the equalizer / demodulator performance for EDGE modulated signals and other signals with a large AM component. 6.0 AGC The LM97593 AGC processor monitors the output level of the ADC and servos it to the desired setpoint. The ADC input is controlled by the DVGA to maintain the proper setpoint level. DVGA operation results in a compression of the signal through the ADC. The DVGA signal compression is reversed in the LM97593 to provide > 120dB of linear dynamic range. This is illustrated in Figure 36. LM97593 interface of the chip to be verified. Also, the AGC loop can be opened by setting AGC_HOLD_IC high and setting the gain of the DVGA by programming the appropriate value into the AGC_IC_A/B register. 7.0 POWER MANAGEMENT The LM97593 can be placed in a low power (static) state by stopping the input clock and setting the PD pin high. To prevent this from placing the LM97593 into unexpected states, the SI pin of the LM97593 should be asserted prior to disabling the input clock and held asserted until the input clock has returned to a stable condition. 8.3 Debug Access Port Real-time access to the following signals is provided by configuring the control interface debug register: • NCO sine and cosine outputs • data after round following mixers • data before F1 and F2 • data after CIC filter within the AGC The access points are multiplexed to a 20-bit parallel output port which is created from signal pins POUT[15:0], AOUT, BOUT, SFS, and RDY according to the table below: 8.0 TESTABILITY 8.1 JTAG Boundary Scan The LM97593 supports IEEE 1149.1 compliant JTAG Boundary Scan for the I/O's. The following pins are used: TRST TMS TDI TDO TCK (test reset) (test mode select) (test data in) (test data out) (test clock) Normal Mode Pin POUT[15:0] RDY SFS AOUT BOUT The following JTAG instructions are supported: Instruction BYPASS EXTEST IDCODE SAMPLE/PRELOAD HIGHZ Description Connects TDI directly to TDO Enables the test access port controller to drive the outputs Connects the 32-bit ID register to TDO Allows the test access port to sample the device inputs and preload test output data Tri-states the outputs Debug Mode Pin DEBUG[19:4] DEBUG[3] DEBUG[2] DEBUG[1] DEBUG[0] SCK will be set to the proper strobe rate for each debug tap point. POUT_EN and PSEL[2:0] have no effect in Debug Mode. The outputs are turned on when the Debug Mode bit is set. Normal serial outputs are also disabled. 9.0 CONTROL REGISTERS The chip is configured and controlled through the use of 8-bit control registers. These registers are accessed for reading or writing using the control bus pins (CE, RD, WR, A[7:0], and D [7:0]) described in section 3.0 Control Interface. The two sets of FIR coefficients are overlaid at the same memory address. Use the PAGE_SEL registers to access the second set of coefficients. The register names and descriptions are listed below in section 9.1 Control Register Addresses and Defaults. A quick reference table is provided in the Condensed LM97593 Address Map. The JTAG Boundary Scan can be used to verify printed circuit board continuity at the system level. 8.2 Test Register The user is able to program a value into TEST_REG and substitute this for the normal channel inputs from the AIN/ BIN pins by selecting it with the crossbar. With the NCO frequency set to zero this allows the DDCs and the output 9.1 Control Register Addresses and Defaults Register Name Width Type Defaulta Addr Bit Description 7:0 2:0 CIC decimation control. N=DEC+1. Valid range is from 7 to 2047. Format is an unsigned integer. This affects both channels. 4 DEC 11b R/W 7 0(LSBs) 1(MSBs) DEC_BY_4 1b R/W 0 1 Controls the decimation factor in F2. 0=Decimate by 2. 1=Decimate by 4. This affects both channels. SCALE 6b R/W 0 2 5:0 CIC SCALE parameter. Format is an unsigned integer representing the number of left bit shifts to perform on the data prior to the CIC filter. Valid range is from 0 to 40. This affects both channels. GAIN_A 3b R/W 0 3 2:0 Value of left bit shift prior to F1 for channel A. GAIN_B 3b R/W 0 4 2:0 Value of left bit shift prior to F1 for channel B. RATE 1B R/W 1 5 7:0 Determines rate of serial output clock. The output rate is FCK/ (RATE+1). Unsigned integer values of 0, 1, 3, 7, 15, and 31 are allowed. SOUT_EN 1b R/W 0 6 0 Enables the serial output pins AOUT, BOUT, SCK, and SFS. 0=Tristate. 1=Enabled. SCK_POL 1b R/W 0 6 1 Determines polarity of the SCK output. 0=AOUT, BOUT, and SFS change on the rising edge of SCK (capture on falling edge). 1=They change on the falling edge of SCK. www.national.com 36 Width Type Defaulta Addr Bit Description SFS_POL 1b R/W 0 6 2 Determines polarity of the SFS output. 0=Active High. 1=Active Low. RDY_POL 1b R/W 0 6 3 Determines polarity of the RDY output. 0=Active High. 1=Active Low. MUX_MODE 1b R/W 0 6 4 Determines the mode of the serial outputs. 0=Each channel is output on its respective pin, 1=Both channels are multiplexed and output on AOUT. See also . PACKED Register Name FORMAT FREQ_A PHASE_A FREQ_B 1b R/W 0 6 5 Controls when SFS goes active. 0=SFS pulses prior to the start of the I and the Q words. 1=SFS pulses only once prior to the start of each I/Q sample pair (i.e. the pair is treated as a doublesized word) The I word precedes the Q word. See . Width Type Defaulta Addr Bit Description 7:6 Determines output number format. 0=Truncate serial output to 8 bits. Parallel output is truncated to 32 bits. 1=Round both serial and parallel to 16-bits. All other bits are set to 0. 2=Round both serial and parallel to 24-bits. All other bits are set to 0. 3=Output floating point. 8-bit mantissa, 4-bit exponent. All other bits are set to 0. 7:0 Frequency word for channel A. Format is a 32-bit, 2’s complement number spread across 4 registers. The LSBs are in the lower registers. The NCO frequency F is F/FCK=FREQ_A/ 232. 7:0 Phase word for channel A. Format is a 16-bit, unsigned magnitude number spread across 2 registers. The LSBs are in the lower registers. The NCO phase PHI is PHI=2*pi*PHASE_A/ 216. 7:0 Frequency word for channel B. Format is a 32-bit, 2’s complement number spread across 4 registers. The LSBs are in the lower registers. The NCO frequency F is F/FCK=FREQ_B/ 232. 2b 4B 2B 4B R/W R/W R/W R/W 0 0 0 0 6 7-10 11-12 13-16 2B R/W 0 17-18 7:0 Phase word for channel B. Format is a 16-bit, unsigned magnitude number spread across 2 registers. The LSBs are in the lower registers. The NCO phase PHI is PHI=2*pi*PHASE_B/ 216. A_SOURCE 2 R/W 0 19 1:0 0=Select AIN as channel input source. 1=Select BIN. 2=3=Select TEST_REG as channel input source. B_SOURCE 2 R/W 1 19 3:2 0=Select AIN as channel input source. 1=Select BIN. 2=3=Select TEST_REG as channel input source. EXP_INH 1b R/W 0 20 0 0=Allow exponent to pass into FLOAT TO FIXED converter. 1=Force exponent in DDC channel to a 7 (maximum digital gain). This affects both channels. Reserved 1b R/W 1 20 1 AGC_FORCE on the CLC5902. Do not use. Reserved 1b R/W 0 20 2 AGC_RESET_EN on the CLC5902. Do not use. AGC_HOLD_IC 1b R/W 0 20 3 0=Normal closed-loop operation. 1=Hold integrator at initial condition. This affects both channels. AGC_LOOP_GAIN 2b R/W 0 20 4:5 Bit shift value for AGC loop. Valid range is from 0 to 3. This affects both channels. Reserved 2B R/W 0 21-22 7:0 AGC_COUNT on the CLC5902. Do not use. AGC_IC_A 1B R/W 0 23 7:0 AGC fixed gain for channel A. Format is an 8-bit, unsigned magnitude number. The channel A DVGA gain will be set to the inverted three MSBs. PHASE_B 37 www.national.com LM97593 Register Name LM97593 Register Name Width Type Defaulta Addr Bit Description AGC_IC_B 1B R/W 0 24 7:0 AGC fixed gain for channel B. Format is an 8-bit, unsigned magnitude number. The channel B DVGA gain will be set to the inverted three MSBs. AGC_RB_A 1B R 0 25 7:0 AGC integrator readback value for channel A. Format is an 8bit, unsigned magnitude number. The user can read the magnitude MSBs of the channel A integrator from this register. AGC_RB_B 1B R 0 26 7:0 AGC integrator readback value for channel B. Format is an 8bit, unsigned magnitude number. The user can read the magnitude MSBs of the channel B integrator from this register. TEST_REG 14b R/W 0 27(LSBs) 28(MSBs) 7:0 5:0 Test input source. Instead of processing values from the A|BIN pins, the value from this location is used instead. Format is 14-bit 2s complement number spread across 2 registers. Reserved 1B - - 29 7:0 For future use. Width Type Defaulta Addr Bit Reserved 1B - - 30 7:0 DEBUG_EN 1b R/W 0 31 0 Register Name Description For future use 0=Normal. 1=Enables access to the internal probe points. Selects internal node tap for debug. 0 selects F1 output for BI, 20 bits 1 selects F1 output for BQ, 20 bits 2 selects F1 output for AQ, 20 bits 3 selects F1 output for AI, 20 bits 4 selects F1 input for BI, 20 bits 5 selects F1 input for BQ, 20 bits 6 selects F1 input for AI, 20 bits 7 selects F1 input for AQ, 20 bits 8 selects NCO A, cosine output. 17 bits, 3 LSBs are 0. 9 selects NCO A, sine output, 17 bits, 3 LSBs are 0. 10 selects NCO B, cosine output, 17 bits, 3 LSBs are 0. 11 selects NCO B, sine output, 17 bits, 3 LSBs are 0. 12 selects NCO AI, rounded output, 15 bits, 5 LSBs are 0. 13 selects NCO AQ, rounded output, 15 bits, 5 LSBs are 0. 14 selects NCO BI, rounded output, 15 bits, 5 LSBs are 0. 15 selects NCO BQ, rounded output, 15 bits, 5 LSBs are 0. 16 selects AGC CIC filter output. 9 MSBs from ch A, next 9 bits from ch B, 2 LSBs are 0. 17-31 Reserved. DEBUG_TAP 5b R/W 0 31 5:1 DITH_A 1b R/W 1 31 6 0=Disable NCO dither source for channel A. 1=Enable. DITH_B 1b R/W 1 31 7 0=Disable NCO dither source for channel B. 1=Enable. 32B R/W 0 128-159 7:0 RAM space that defines key AGC loop parameters. Format is 32 separate 8-bit, 2’s complement numbers. This is common to both channels. 7:0 Coefficients for F1. Format is 11 separate 16-bit, 2’s complement numbers, each one spread across 2 registers. The LSBs are in the lower registers. For example, coefficient h0[7:0] is in address 160, h0[15:8] is in address 161, h1[7:0] is in address 162, h1[15:8] is in address 163. PAGE_SEL_F1=1 maps these addresses to coefficient memory B. 7:0 Coefficients for F2. Format is 32 separate 16-bit, 2’s complement numbers, each one spread across 2 registers. The LSBs are in the lower registers. For example, coefficient h0[7:0] is in address 182, h0[15:8] is in address 183, h1[7:0] is in address 184, h1[15:8] is in address 185. PAGE_SEL_F2=1 maps these addresses to coefficient memory B. AGC_TABLE F1_COEFF F2_COEFF www.national.com 22B 64B R/W R/W 0 0 160-181 182-245 38 Width Type Defaulta Addr Bit Description COEF_SEL_F1A 1b R/W 0 246 0 Channel A F1 coefficient select register. 0=memory A, 1=memory B. COEF_SEL_F1B 1b R/W 0 246 1 Channel B F1 coefficient select register. 0=memory A, 1=memory B. PAGE_SEL_F1 1b R/W 0 246 2 F1 coefficient page select register. 0=memory A, 1=memory B. COEF_SEL_F2A 1b R/W 0 247 0 Channel A F2 coefficient select register. 0=memory A, 1=memory B. COEF_SEL_F2B 1b R/W 0 247 1 Channel B F2 coefficient select register. 0=memory A, 1=memory B. PAGE_SEL_F2 1b R/W 0 247 2 F2 coefficient page select register. 0=memory A, 1=memory B. SFS_MODE 1b R/W 0 248 0 0=SFS asserted at the start of each output word when PACKED=1 or each I/Q pair when PACKED=0, 1=SFS asserted at the start of each output sample period. Width Type Defaulta Addr Bit Description SDC_EN 1b R/W 0 248 1 AGC_COMB_ORD 2b R/W 0 249 1:0 Enable reduced bandwidth AGC power detector. 0=2nd-order decimate-by-eight CIC, 1=1-tap comb added to CIC, 2=4-tap comb added to CIC. EXT_DELAY 5b R/W 0 249 6:2 Number of CK period delays needed to align the DVGA gain step with the digital gain compensation step. Set this register to 7 if ASTROBE and BSTROBE are not used. Otherwise set to 8. Register Name 0=normal serial mode, 1=serial daisy-chain master mode. a. These are the default values set by a master reset (MR). Sync in (SI) will not affect any of these values. 9.2 Condensed LM97593 Address Map Register Name Addr Addr Hex Bit7 Bit6 Bit5 DEC 0 0x00 Dec7 Dec6 Dec5 DEC_BY_4 1 0x01 SCALE 2 0x02 GAIN_A 3 GAIN_B 4 RATE 5 0x05 Rate7 Rate6 Rate5 SERIAL_CTRL 6 0x06 FMT1 FMT0 Packed FREQ_A 7 0x07 FA7 FA6 FA5 FA4 FA3 FA2 FA1 FA0 8 0x08 FA15 FA14 FA13 FA12 FA11 FA10 FA9 FA8 Bit4 Dec4 Bit3 Dec3 Bit2 Bit1 Bit0 Dec2 Dec1 Dec0 Dec10 Dec9 Dec8 Sc2 Sc1 Sc0 0x03 GA2 GA1 GA0 0x04 GB2 GB1 GB0 Rate2 Rate1 Rate0 DecBy4 Sc5 Sc4 Rate4 Sc3 Rate3 MuxMode RDY_POL SFS_POL SCK_POL SOUT_EN 9 0x09 FA23 FA22 FA21 FA20 FA19 FA18 FA17 FA16 10 0x0A FA31 FA30 FA29 FA28 FA27 FA26 FA25 FA24 11 0x0B PA7 PA6 PA5 PA4 PA3 PA2 PA1 PA0 12 0x0C PA15 PA14 PA13 PA12 PA11 PA10 PA9 PA8 13 0x0D FB7 FB6 FB5 FB4 FB3 FB2 FB1 FB0 14 0x0E FB15 FB14 FB13 FB12 FB11 FB10 FB9 FB8 15 0x0F FB23 FB22 FB21 FB20 FB19 FB18 FB17 FB16 16 0x10 FB31 FB30 FB29 FB28 FB27 FB26 FB25 FB24 17 0x11 PB7 PB6 PB5 PB4 PB3 PB2 PB1 PB0 18 0x12 PB15 PB14 PB13 PB12 PB11 PB10 PB9 PB8 Source 19 0x13 BS1 BS0 AS1 AS0 AGC_CTRL 20 0x14 AgcHldlC Reserved Reserved Explnh AGC_COUNT 21 0x15 Reserved Reserved Reserved Reserved Reserved Reserved Reserved Reserved 22 0x16 Reserved Reserved Reserved Reserved Reserved Reserved Reserved Reserved PHASE_A FREQ_B PHASE_B AgcLG1 39 AgcLG0 www.national.com LM97593 Register Name LM97593 Register Name Addr Addr Hex Bit7 Bit6 Bit5 AGC_IC_A 23 0x17 AgclcA7 AgclcA6 AgclcA5 AgclcA4 AgclcA3 AgclcA2 AgclcA1 AgclcA0 AGC_IC_B 24 0x18 AgclcB7 AgclcB6 AgclcB5 AgclcB4 AgclcB3 AgclcB2 AgclcB1 AgclcB0 AGC_RB_A 25 0x19 AgcRbA7 AgcRbA6 AgcRbA5 AgcRbA4 AgcRbA3 AgcRbA2 AgcRbA1 AgcRbA0 AGC_RB_B 26 0x1A AGCRbB7 AGCRbB6 AGCRbB5 AGCRbB4 AGCRbB3 AGCRbB2 AGCRbB1 AGCRbB0 TEST_REG 27 0x1B 28 0x1C DEBUG 31 0x1F AGC_TABLE 128 0x80 159 0x9F Test7 DITH_B Test6 DITH_A Bit4 Bit3 Bit2 Bit1 Bit0 Test5 Test4 Test3 Test2 Test1 Test0 Test13 Test12 Test11 Test10 Test9 Test8 TapSel3 TapSel2 TapSel1 TapSel0 DebugEnable TapSel4 The AGC Table loads from the low address to the high address in this order: "1st location, 2nd location..." Register Name Addr Addr Hex F1_COEFF 160 0xA0 181 0xB5 182 0xB6 245 0xF5 F1_CTRL 246 0xF6 PgSelF1 CfSelF1B CfSelF1A F2_CTRL 247 0xF7 PgSelF2 CfSelF2B CfSelF2A SERIAL_CTRL2 248 0xF8 SdcEn SfsMode AGC_CTRL2 0xF9 F2COEFF www.national.com 249 Bit7 Bit6 Bit5 Bit4 Bit3 Bit2 Bit1 Bit0 The FIR Coefficients load from the low address to the high address in this order: "1st location low byte, 1st location high byte, 2nd location..."" The Page Select bits determine which set of coefficient memory is written. ExtDelay4 ExtDelay3 ExtDelay2 ExtDelay1 ExtDelay0 CombOrd1 40 CombOrd0 A block diagram of the AGC is shown in Figure 40. The DVGA interface comprises four pins for each of the channels. The first three pins of this interface are a 3-bit binary word that controls the DVGA gain in 6dB steps (AGAIN). The final pin is ASTROBE which allows the AGAIN bits to be latched into the DVGA’s register. A key feature of the ASTROBE, illustrated Figure 41, is that it toggles only if the data on AGAIN has changed from the previous cycle. Not shown is that ASTROBE and BSTROBE are independent. For example, ASTROBE only toggles when AGAIN has changed. BSTROBE will not toggle because AGAIN has changed. This is done to minimize unnecessary digital noise on the sensitive analog path through the DVGA. ASTROBE and BSTROBE are asserted during MR and SI to properly initialize the DVGAs. The absolute value circuit and the 2-stage, decimate-by-8 CIC filter comprise the power detection part of the AGC. The power detector bandwidth is set by the CIC filter to FCK/8. The absolute value circuit doubles the effective input frequency. This has the effect of reducing the power detector bandwidth from FCK/8 to FCK/16. For a full-scale sinusoidal input, the absolute value circuit output is a dc value of 511*(2/π). Because the absolute value 30008736 FIGURE 39. Power detector filter response, 52 MHz 30008737 FIGURE 40. LM97593 AGC circuit, Channel A 30008738 FIGURE 41. Timing diagram for AGC/DVGA interface, Channel A. Refer to Figure 7 for detailed timing information. 41 www.national.com LM97593 circuit also generates undesired even harmonic terms, the CIC filter (response shown in Figure 39) is required to remove these harmonics. The first response null occurs at AGC Theory of Operation LM97593 FCK/8, where FCK is the clock frequency, and the response magnitude is at least 25dB below the dc value from FCK/10 to 9FCK/10. Because the 2nd harmonic from the absolute value circuit is about 10dB below the dc this means that the ripple in the detected level is about 0.7dB or less for input frequencies between FCK/20 to 19FCK/20. Setting the AGC_COMB_ORD register to either 1 or 2 will narrow the power detector’s bandwidth as shown in Figure 39. The “FIXED TO FLOAT CONVERTER” takes the fixed point 9-bit output from the CIC filter and converts it to a “floating point” number. This conversion is done so that the 32 values in the RAM can be uniformly assigned (dB scale) to detected power levels (54 dB range). This provides a resolution of 1.7dB between detected power levels. The truth table for this converter is given in Table 4. The upper three bits of the output represent the exponent (e) and the lower 2 are the mantissa (m). The exponent is determined by the position of the leading ‘1’ out of the CIC filter. An output of ‘001XX’ corresponds to a leading ‘1’ in bit 2 (LSB is bit 0). The exponent increases by one each time the leading ‘1’ advances in bit position. The mantissa bits are the two bits that follow the leading ‘1’. If we define E as the decimal value of the exponent bits and M as the decimal value of the mantissa bits, the output of the CIC filter, POUT, corresponding to a given “FIXED TO FLOAT CONVERTER” output is: that the DVGA gain control polarity is positive as is the case for the CLC5526. The gain around the entire loop must be negative. Observe in Equation 6 that the control gain is dependent on operating point G. If we instead compute the control gain with log conversion then: (Eq. 11) which is no longer operating-point dependent. The log function is constructed by computing the CIC filter output associated with each address (Equation 5) and converting these to dB. Full scale (dc signal) is 20log(511)=54dB. The reference subtraction is constructed by subtracting the desired loop servo point (in dB) from the table values computed in the previous paragraph. For example, if it is desired that the DVGA servo the ADC input level (sinusoidal signal) to -6dBFS, the number to subtract from the data is: (Eq. 12) The table data will then cross through zero at the address corresponding to this reference level. A deadband wider than 6dB should then be constructed symmetrically about this point. This prevents the loop from hunting due to the 6dB gain steps of the DVGA. Any deadband in excess of 6dB appears as hysteresis in the servo point of the loop as illustrated in Figure 37. The deadband is constructed by loading zeros into those addresses on either side of the one which corresponds to the reference level. The last function of the RAM table is that of error amplification. All the operations preceding this one gave a table slope SRAM = 1. This must now be adjusted in order to control the time constant of the loop given by: (Eq. 9) The max() and min() operators account for row 1 of Table 4 which is a special case because M=POUT. Equation 5 associates each address of the RAM with a CIC filter output. TABLE 4. Fixed to Float Converter Truth Table INPUT OUTPUT (eeemm) 0-3 000XX 4-7 001XX 8-15 010XX 16-31 011XX 32-63 100XX 64-127 101XX 128-255 110XX 256-511 111XX (Eq. 13) The term GL in this equation is the loop gain: (Eq. 14) The design equations are obtained by solving Equation 13 for GL and Equation 14 for SRAM. AGC_LOOP_GAIN is a control register value that determines the number of bits to shift the output of the RAM down by. This allows some of the loop gain to be moved out of the RAM so that the full output range of the table is utilized but not exceeded. The valid range for AGC_LOOP_GAIN is from 0 to 3 which corresponds to a 0 to 3 bit shift to the left. An example set of numbers to implement a loop having a reference of 6dB below full scale, a deadband of 8dB, and a loop gain of 0.108 is: -102 -102 -88 -80 -75 -70 -66 -63 -61 -56 -53 -50 -47 -42 -39 -36 -33 -29 -25 -22 -19 -15 -11 -0 0 0 0 0 0 13 17 20 As shown in Figure 40, the 32X8 RAM look-up table implements the functions of log converter, reference subtraction, error amplifier, and deadband. The user must build each of these functions by constructing a set of 8-bit, 2’s complement numbers to be loaded into the RAM. Each of these functions and how to construct them are discussed in the following paragraphs. A log conversion is done in order to keep the loop gain independent of operating point. To see why this is beneficial, the control gain of the DVGA computed without log conversion is: (Eq. 10) where G is the decimal equivalent of GAIN and Go accounts for the DVGA gain in excess of unity. This equation assumes www.national.com 42 LM97593 30008740 FIGURE 43. Example of programmed RAM contents 30008739 FIGURE 42. Example of programmed RAM contents TABLE 5. 15-bit Mixer Output Alignment into the 22-bit SHIFT-UP Based On EXP AGAINa EXPb Inputc 21d 20 19 18 17 16 15 14 ... 8 7 6 5 4 3 2 1 0 000 = -12dB 111 = +0dB -12dB 14 13 12 11 10 9 8 7 ... 1 0 L L L L L L L 001 = -6dB 110 = -6dB -12dB 14 14 13 12 11 10 9 8 ... 2 1 0 L L L L L L 010 = 0dB 101 = -12dB -12dB 14 14 14 13 12 11 10 9 ... 3 2 1 0 L L L L L 011 = +6dB 100 = -18dB -12dB 14 14 14 14 13 12 11 10 ... 4 3 2 1 0 L L L L 100 = +12dB 011 = -24dB -12dB 14 14 14 14 14 13 12 11 ... 5 4 3 2 1 0 L L L 101 = +18dB 010 = -30dB -12dB 14 14 14 14 14 14 13 12 ... 6 5 4 3 2 1 0 L L 110 = +24dB 001 = -36dB -12dB 14 14 14 14 14 14 14 13 ... 7 6 5 4 3 2 1 0 L 111 = +30dB 000 = -42dB -12dB 14 14 14 14 14 14 14 14 ... 8 7 6 5 4 3 2 1 0 a. AGAIN sets the DVGA or analog gain value. b. EXP sets the "FIXED TO FLOAT CONVERTER" or digital gain value. c. 22-bit input to SHIFT-UP block in Figure 19 horizontally, linearized SHIFT-UP value vertically. d. The numbers in the center of the table represent the mixer output bits. 'L' represents a logic low. These values are shown with respect to the table addresses in Figure 42, and the CIC filter output POUT in Figure 43. For a 52MHz clock rate and AGC_LOOP_GAIN=2, these values result in a loop time constant of 1.5µs. The error signal from the loop gain “SHIFT DOWN” circuit is gated into the loop integrator. A MUX within the integrator feedback allows the integrator to be initialized to the value loaded into AGC_IC_A (channel B can be set independently). The top eight bits of the integrator output can also be read back over the microprocessor interface from the AGC_RB_A (or AGC_RB_B) register. The top 3 bits become AGAIN and are output along with the ASTROBE signal on the DVGA interface pins. The valid range of AGAIN is from 0 to 7 which corresponds to a valid range of 0 to 211-1 for the 11-bit, 2’s complement integrator output from which AGAIN is derived. This is illustrated in Figure 44. The integrator saturates at these limits to prevent overshoots as the integrator attempts to enter the valid range. The AGAIN value is inverted (EXP) and used to adjust the gain of the incoming signal to provide a linear output dynamic range. The relationship between the DVGA analog gain (AGAIN) and the “FIXED TO FLOAT CONVERTER” digital gain (EXP) is shown in Table 5. The DVGA’s compression of the incoming signal in the analog domain vs. the subsequent expansion in the digital domain is shown in Figure 36. The AGC may be forced to free run by setting AGC_HOLD_IC low. Writing an initial condition to AGC_IC_A|B and then setting AGC_HOLD_IC high will force the AGC to a fixed gain. The three MSBs of the value written to AGC_IC_A|B are inverted and output to drive the DVGA. Allowing the AGC to free run should be appropriate for most applications. If the INH_EXP bit is not set, the DVGA gain word (EXP) is routed to the “FLOAT TO FIXED CONVERTER” in the DDCs prior to the programmable decimation filter. The EXP signals are delayed to account for the propagation delay of the DVGA interface and the ADC12DL080 ADC. 43 www.national.com LM97593 General Applications Information 10.0 OUTPUTS Be very careful when driving a high capacitance bus. The more capacitance the output drivers must charge for each conversion, the more instantaneous digital current flows through VDR and DRGND. These large charging current spikes can cause on-chip ground noise and couple into the analog circuitry, degrading dynamic performance. Adequate bypassing, limiting output capacitance and careful attention to the ground plane will reduce this problem. To minimize noise due to output switching, minimize the load currents at the digital outputs. Only one driven input should be connected to each output pin. Additionally, inserting series resistors of about 100Ω at the digital outputs, close to the ADC pins, will isolate the outputs from trace and other circuit capacitances and limit the output currents, which could otherwise result in performance degradation. See Figure 45. 30008741 FIGURE 44. AGC integrator output limits www.national.com 44 LM97593 30008757 FIGURE 45. Application Circuit 45 www.national.com LM97593 Best performance at high frequencies and at high resolution is obtained with a straight signal path. That is, the signal path through all components should form a straight line wherever possible. Be especially careful with the layout of inductors. Mutual inductance can change the characteristics of the circuit in which they are used. Inductors should not be placed side by side, even with just a small part of their bodies beside each other. The analog input should be isolated from noisy signal traces to avoid coupling of spurious signals into the input. Any external component (e.g., a filter capacitor) connected between the converter's input pins and ground or to the reference input pin and ground should be connected to a very clean point in the ground plane. All analog circuitry (input amplifiers, filters, reference components, etc.) should be placed in the analog area of the board. All digital circuitry and I/O lines should be placed in the digital area of the board. The LM97593 should be between these two areas. Furthermore, all components in the reference circuitry and the input signal chain that are connected to ground should be connected together with short traces and enter the ground plane at a single, quiet point. All ground connections should have a low inductance path to ground. 11.0 POWER SUPPLY CONSIDERATIONS The power supply pins should be bypassed with a 10 µF capacitor and with a 0.1 µF ceramic chip capacitor near each power pin. Leadless chip capacitors are preferred because they have low series inductance. The LM97593 is sensitive to power supply noise. Accordingly, the noise on the analog supply pins should be kept below 100 mVP-P. No pin should ever have a voltage on it that is in excess of the supply voltages, not even on a transient basis. Be especially careful of this during power turn on and turn off. The VDR pin provides power for the output drivers and should be operated from a supply equivalent to VD. 12.0 LAYOUT AND GROUNDING A proper grounding and proper routing of all signals are essential to ensure accurate conversion. Maintaining separate analog and digital areas of the board, with the LM97593 between these areas, is required to achieve specified performance. The ground return for the data outputs (DRGND) carries the ground current for the output drivers. The output current can exhibit high transients that could add noise to the conversion process. To prevent this from happening, the DRGND pins should NOT be connected to system ground in close proximity to any of the LM97593's other ground pins. Capacitive coupling between the typically noisy digital circuitry and the sensitive analog circuitry can lead to poor performance. The solution is to keep the analog circuitry separated from the digital circuitry, and to keep the clock line as short as possible. Digital circuits create substantial supply and ground current transients. The logic noise thus generated could have significant impact upon system noise performance. The best logic family to use in systems with A/D converters is one which employs non-saturating transistor designs, or has low noise characteristics, such as the 74LS, 74HC(T) and 74AC(T)Q families. The worst noise generators are logic families that create the largest supply current transients during clock or signal edges, like the 74F and the 74AC(T) families. The effects of the noise generated from the LM97593 output switching can be minimized through the use of 100Ω resistors in series with each data output line. Locate these resistors as close to the LM97593 output pins as possible. Since digital switching transients are composed largely of high frequency components, total ground plane copper weight will have little effect upon the logic-generated noise. This is because of the skin effect. Total surface area is more important than is total ground plane volume. Generally, analog and digital lines should cross each other at 90° to avoid crosstalk. To maximize accuracy in high speed, high resolution systems, however, avoid crossing analog and digital lines altogether. It is important to keep clock lines as short as possible and isolated from ALL other lines, including other digital lines. Even the generally accepted 90° crossing should be avoided with the clock line as even a little coupling can cause problems at high frequencies. This is because other lines can introduce jitter into the clock line, which can lead to degradation of SNR. Also, the high speed clock can introduce noise into the analog chain. www.national.com 13.0 DYNAMIC PERFORMANCE To achieve the best dynamic performance, the clock source driving the CLK input must be free of jitter. Isolate the ADC clock from any digital circuitry with buffers, as with the clock tree shown in Figure 46. The gates used in the clock tree must be capable of operating at frequencies much higher than those used if added jitter is to be prevented. Best performance will be obtained with a differential input drive, compared with a single-ended drive, as discussed in Sections 1.3.1 and 1.3.2. It is good practice to keep the clock line as short as possible and to keep it well away from any other signals. Other signals can introduce jitter into the clock signal, which can lead to reduced SNR performance, and the clock can introduce noise into other lines. Even lines with 90° crossings have capacitive coupling, so try to avoid even these 90° crossings of the clock line. 30008756 FIGURE 46. Isolating the ADC Clock from other Circuitry with a Clock Tree 46 15.1 Driving the inputs (analog or digital) beyond the power supply rails. For proper operation, all inputs should not go more than 100 mV beyond the supply rails (more than 100 mV below the ground pins or 100 mV above the supply pins). Exceeding these limits on even a transient basis may cause faulty or erratic operation. It is not uncommon for high speed digital components (e.g., 74F and 74AC devices) to exhibit overshoot or undershoot that goes above the power supply or below ground. A resistor of about 47Ω to 100Ω in series with any offending digital input, close to the signal source, will eliminate the problem. Do not allow input voltages to exceed the supply voltage, even on a transient basis. Not even during power up or power down. Be careful not to overdrive the inputs of the LM97593 with a device that is powered from supplies outside the range of the LM97593 supply. Such practice may lead to conversion inaccuracies and even to device damage. 15.2 Attempting to drive a high capacitance digital data bus. The more capacitance the output drivers must charge for each conversion, the more instantaneous digital current flows through VDR and DRGND. These large charging current spikes can couple into the analog circuitry, degrading dynamic performance. Adequate bypassing and maintaining separate analog and digital areas on the pc board will reduce this problem. The digital data outputs should be buffered (with 74ACQ541, for example) if they will drive a large capacitive load. Dynamic performance can also be improved by adding series resistors at each digital output, close to the LM97593, which reduces the energy coupled back into the part's output pins by limiting the output current. A reasonable value for these resistors is 100Ω. 15.3 Using an inadequate amplifier to drive the analog input. As explained in Section 1.3, the capacitance seen at the input alternates between 8 pF and 7 pF, depending upon the phase of the clock. This dynamic load is more difficult to drive than is a fixed capacitance. If the amplifier exhibits overshoot, ringing, or any evidence of instability, even at a very low level, it will degrade performance. A small series resistor at each amplifier output and a capacitor at the analog inputs will improve performance. 0.8V ≤ VREF ≤ 1.2V Operating outside of these limits could lead to performance degradation. 15.5 Inadequate network on Reference Bypass pins (VRPA, VRNA, VCOMA, VRPB, VRNB and VCOMB). As mentioned in Section 1.2, these pins should be bypassed with 0.1 µF capacitors to ground at VRMA and VRMB and with a 10 µF between pins VRPA and VRNA and between VRPB and VRNB for best performance. 15.6 Using a clock source with excessive jitter, using excessively long clock signal trace, or having other signals coupled to the clock signal trace. This will cause the sampling interval to vary, causing excessive output noise and a reduction in SNR and SINAD performance. Evaluation Hardware Evaluation boards are available to facilitate designs based on the LM97593: 16.1 LM97593EB The LM97593 evaluation board provides a complete narrowband receiver from IF to digital symbols. 16.2 SOFTWARE Control panel software for the LM97593 supports complete device configuration on both evaluation boards. Integrated capture software manages the capture of data and its storage in a file on a PC. Matlab script files support data analysis: FFT, DNL, and INL plotting. This software and additional application information is available on the Basestation CDROM. 47 www.national.com LM97593 Also, it is important that the signals at the two inputs have exactly the same amplitude and be exactly 180º out of phase with each other. Board layout, especially equality of the length of the two traces to the input pins, will affect the effective phase between these two signals. Remember that an operational amplifier operated in the non-inverting configuration will exhibit more time delay than will the same device operating in the inverting configuration. 15.4 Operating with the reference pins outside of the specified range. As mentioned in Section 1.2, VREF should be in the range of Common Application Pitfalls LM97593 Physical Dimensions inches (millimeters) unless otherwise noted LM97593VH PQFP Package Dimensions www.national.com 48 LM97593 Notes 49 www.national.com LM97593 Dual ADC / Digital Tuner / AGC Notes For more National Semiconductor product information and proven design tools, visit the following Web sites at: Products Design Support Amplifiers www.national.com/amplifiers WEBENCH www.national.com/webench Audio www.national.com/audio Analog University www.national.com/AU Clock Conditioners www.national.com/timing App Notes www.national.com/appnotes Data Converters www.national.com/adc Distributors www.national.com/contacts Displays www.national.com/displays Green Compliance www.national.com/quality/green Ethernet www.national.com/ethernet Packaging www.national.com/packaging Interface www.national.com/interface Quality and Reliability www.national.com/quality LVDS www.national.com/lvds Reference Designs www.national.com/refdesigns Power Management www.national.com/power Feedback www.national.com/feedback Switching Regulators www.national.com/switchers LDOs www.national.com/ldo LED Lighting www.national.com/led PowerWise www.national.com/powerwise Serial Digital Interface (SDI) www.national.com/sdi Temperature Sensors www.national.com/tempsensors Wireless (PLL/VCO) www.national.com/wireless THE CONTENTS OF THIS DOCUMENT ARE PROVIDED IN CONNECTION WITH NATIONAL SEMICONDUCTOR CORPORATION (“NATIONAL”) PRODUCTS. NATIONAL MAKES NO REPRESENTATIONS OR WARRANTIES WITH RESPECT TO THE ACCURACY OR COMPLETENESS OF THE CONTENTS OF THIS PUBLICATION AND RESERVES THE RIGHT TO MAKE CHANGES TO SPECIFICATIONS AND PRODUCT DESCRIPTIONS AT ANY TIME WITHOUT NOTICE. NO LICENSE, WHETHER EXPRESS, IMPLIED, ARISING BY ESTOPPEL OR OTHERWISE, TO ANY INTELLECTUAL PROPERTY RIGHTS IS GRANTED BY THIS DOCUMENT. TESTING AND OTHER QUALITY CONTROLS ARE USED TO THE EXTENT NATIONAL DEEMS NECESSARY TO SUPPORT NATIONAL’S PRODUCT WARRANTY. EXCEPT WHERE MANDATED BY GOVERNMENT REQUIREMENTS, TESTING OF ALL PARAMETERS OF EACH PRODUCT IS NOT NECESSARILY PERFORMED. NATIONAL ASSUMES NO LIABILITY FOR APPLICATIONS ASSISTANCE OR BUYER PRODUCT DESIGN. BUYERS ARE RESPONSIBLE FOR THEIR PRODUCTS AND APPLICATIONS USING NATIONAL COMPONENTS. PRIOR TO USING OR DISTRIBUTING ANY PRODUCTS THAT INCLUDE NATIONAL COMPONENTS, BUYERS SHOULD PROVIDE ADEQUATE DESIGN, TESTING AND OPERATING SAFEGUARDS. EXCEPT AS PROVIDED IN NATIONAL’S TERMS AND CONDITIONS OF SALE FOR SUCH PRODUCTS, NATIONAL ASSUMES NO LIABILITY WHATSOEVER, AND NATIONAL DISCLAIMS ANY EXPRESS OR IMPLIED WARRANTY RELATING TO THE SALE AND/OR USE OF NATIONAL PRODUCTS INCLUDING LIABILITY OR WARRANTIES RELATING TO FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT. LIFE SUPPORT POLICY NATIONAL’S PRODUCTS ARE NOT AUTHORIZED FOR USE AS CRITICAL COMPONENTS IN LIFE SUPPORT DEVICES OR SYSTEMS WITHOUT THE EXPRESS PRIOR WRITTEN APPROVAL OF THE CHIEF EXECUTIVE OFFICER AND GENERAL COUNSEL OF NATIONAL SEMICONDUCTOR CORPORATION. As used herein: Life support devices or systems are devices which (a) are intended for surgical implant into the body, or (b) support or sustain life and whose failure to perform when properly used in accordance with instructions for use provided in the labeling can be reasonably expected to result in a significant injury to the user. A critical component is any component in a life support device or system whose failure to perform can be reasonably expected to cause the failure of the life support device or system or to affect its safety or effectiveness. National Semiconductor and the National Semiconductor logo are registered trademarks of National Semiconductor Corporation. All other brand or product names may be trademarks or registered trademarks of their respective holders. Copyright© 2008 National Semiconductor Corporation For the most current product information visit us at www.national.com National Semiconductor Americas Technical Support Center Email: [email protected] Tel: 1-800-272-9959 www.national.com National Semiconductor Europe Technical Support Center Email: [email protected] German Tel: +49 (0) 180 5010 771 English Tel: +44 (0) 870 850 4288 National Semiconductor Asia Pacific Technical Support Center Email: [email protected] National Semiconductor Japan Technical Support Center Email: [email protected]