PDSP16350 PDSP16350 I/Q Splitter/NCO Supersedes version in December 1993 Digital Video & DSP IC Handbook, HB3923-1 The PDSP16350 provides an integrated solution to the need for very accurate, digitised, sine and cosine waveforms. Both these waveforms are produced simultaneously, with 16 bit amplitude accuracy, and are synthesised using a 34 bit phase accumulator. The more significant bits of this provide 16 bits of phase accuracy for the sine and cosine look up tables. With a 20 MHz system clock, waveforms up to 10 MHz can be produced, with 0.001 Hz resolution. If frequency modulation is required with no discontinuities, the phase increment value can be changed linearly on every clock cycle. Alternatively absolute phase jumps can be made to any phase value. The provision of two output multipliers allows the sine and cosine waveforms to be amplitude modulated with a 16 bit value present on the input port. This option can also be used to generate the in-phase and quadrature components from an incoming signal. This I/Q split function is required by systems which employ complex signal processing. DS3711 - 2.3 September 1996 DIN PHASE OFFSET REGISTER PHASE INCR REGISTER SCALING REGISTER ACCUM MUX PHASE ACCUM REGISTER CORDIC PROCESSOR ARRAY FEATURES ■ Direct Digital Synthesiser producing simultaneous sine and cosine values ■ 16 bit phase and amplitude accuracy, giving spur levels down to - 90 dB ■ Synthesised outputs from DC to 10 MHz with accuracies better than 0.001 Hz ■ Amplitude and Phase modulation modes ■ 84 pin PGA or 132 pin QFP SIN COS Fig. 1 Block Diagram APPLICATIONS ASSOCIATED PRODUCTS ■ Numerically controlled oscillator (NCO) PDSP16256/A Programmable FIR Filter ■ Quadrature signal generator PDSP16510A FFT Processor ■ FM, PM, or AM signal modulator PDSP16488A 2D Convolver ■ Sweep Oscillator ■ High density signal constellation applications with simultaneous amplitude and phase modulation ■ VHF reference for UHF generators ■ Signal demodulator 1 PDSP16350 N M JUMP MODE DIN17 DIN19 DIN21 DIN23 VDD DIN26 GND DIN28 DIN30 DIN32 DIN18 DIN20 DIN22 DIN24 DIN25 DIN27 DIN29 DIN31 DIN33 VOUT RES VIN L DIN15 DIN16 SIN15 SIN14 K DIN13 DIN14 SIN13 SIN12 J DIN11 DIN12 SIN11 SIN10 H GND DIN10 SIN9 GND G DIN9 DIN8 SIN8 SIN7 F VDD DIN7 SIN6 VDD E DIN6 DIN5 SIN4 SIN5 D DIN4 DIN3 SIN2 SIN3 C DIN2 DIN1 SIN0 SIN1 B DIN0 A CLOCK 1 COS15 COS13 COS11 COS9 COS7 COS6 COS4 COS2 COS0 GND COS14 COS12 COS10 VDD COS8 GND COS5 COS3 COS1 OES OEC 2 3 4 5 6 7 8 9 10 11 12 13 Fig. 2 A. Pin out - bottom view (84 pin PGA - AC84) 2 CEN PDSP16350 GC SIG GC SIG GC SIG GC SIG 1 N/C 34 N/C 67 GND 100 GND 2 CEN 35 VOUT 68 DIN17 101 VDD 102 GND 3 N/C 36 DIN33 69 N/C 4 SIN0 37 GND 70 DIN16 103 N/C 104 COS15 5 SIN1 38 VDD 71 DIN15 6 SIN2 39 DIN32 72 GND 105 COS14 106 N/C 7 GND 40 N/C 73 VDD 8 VDD 41 DIN31 74 DIN14 107 COS13 108 COS12 9 SIN3 42 DIN30 75 DIN13 10 SIN4 43 N/C 76 DIN12 109 N/C 110 COS11 11 N/C 44 DIN29 77 N/C 12 SIN5 45 DIN28 78 DIN11 111 N/C 112 COS10 13 SIN6 46 N/C 79 DIN10 14 N/C 47 DIN27 80 N/C 113 COS9 114 VDD 15 SIN7 48 GND 81 DIN9 16 SIN8 49 VDD 82 GND 115 GND 116 COS8 17 VDD 50 DIN26 83 VDD 18 GND 51 DIN25 84 DIN8 117 COS7 118 N/C 19 SIN9 52 DIN24 85 DIN7 20 N/C 53 DIN23 86 DIN6 119 COS6 120 COS5 21 SIN10 54 VDD 87 N/C 22 SIN11 55 DIN22 88 DIN5 121 N/C 122 COS4 23 N/C 56 GND 89 N/C 24 SIN12 57 DIN21 90 DIN4 123 N/C 124 COS3 25 SIN13 58 VDD 91 DIN3 26 SIN14 59 DIN20 92 VDD 125 COS2 126 N/C 27 VDD 60 DIN19 93 GND 28 GND 61 GND 94 DIN2 127 COS1 128 VDD 29 SIN15 62 VDD 95 DIN1 30 VIN 63 DIN18 96 N/C 129 GND 130 COS0 31 N/C 64 MODE 97 DIN0 32 N/C 65 JUMP 98 N/C 131 OES 99 CLK 132 OEC 33 RESET 66 VDD Fig.2B Pin out (132 pin ceramic QFP - GC132) 3 PDSP16350 SIGNAL DESCRIPTION DIN33:0 Data bus for the input register. This input register provides a 34 bit, incremental or absolute, phase value, if the mode pin is low. Alternatively if the mode pin is high, it provides either an 18 bit phase increment value, via D17:0, and a 16 bit scale value via D33:18 or a 34 bit phase increment value depending on the JUMP input see below. SIN15:0 16 bit sine output data in fractional two’s complement format. COS15:0 16 bit cosine output data in fractional two’s complement format. CEN Clock enable for the data input register. When low, data will be latched on the rising edge of the clock. When high data will be retained in the input register. MODE Mode control input. When low, data in the input register is interpreted as either a 34 bit phase increment value or a 34 bit absolute phase value. When high, the output multipliers are enabled and will scale the waveforms with the upper 16 bits in the input register. The phase increment is loaded from the the lower 18 bits. The full 34 bit phase increment register can also be loaded using JUMP see below. JUMP With MODE low (Frequency or Phase Modulation) When low JUMP will allow normal phase incrementing to occur. When high, the data on the input pins will be interpreted as a 34 bit absolute phase value to replace the present value in the accumulator. JUMP is internally latched to match the delay through the data input register, and to allow data in the internal pipeline to be correctly processed. CEN must also be low to latch the required data from DIN. When Mode is high (Amplitude Modulation) When low JUMP will allow normal phase incrementing to occur, with the phase increment value taken from the lower 18 data inputs. When high, the data on the input pins will replace the full 34 bits of the phase increment register. CEN must also be low to latch the required data. RES When high will clear the phase accumulator and phase increment registers, after data in the internal pipeline has been correctly processed. CLK Input clock. OES Output enable for SIN 15:0. Outputs are high impedance when OES is high. OEC Output enable for COS15:0. Outputs are high impedance when OEC is high. VIN Valid input flag. A delayed version of this input is available on the VOUT pin, with the delay matching the data processing pipeline delay. This input has no other internal function. VOUT Valid output flag. See above. GND Five ground pins. All must be connected. VCC Four +5V pins. All must be connected. Table 1. Pin Description 4 PDSP16350 DEVICE OPERATION Sine and cosine are simultaneously produced by the Cordic processor, which is addressed by the upper 16 bits of the output from a 34 bit phase accumulator. The accumulator divides the digital phase circle into a number of steps, one step for each state of the accumulator. When the accumulator reaches its maximum value it overflows back to zero and the sequence is repeated. The accumulator is incremented once per incoming clock cycle, by an amount which defines the frequency which is to be generated. The increment required is defined by : Desired O/P Frequency x 2N Increment = Incoming Clock Frequency where N is the number of bits in the accumulator. Since the Nyquist criteria for proper waveform reconstruction must still be obeyed, the maximum output frequency is half the incoming frequency. In practice, when a return is made to the analog world, just meeting the minimum Nyquist requirement would require a ‘brick wall’ low pass filter to remove the alias signals. A more useful ‘rule of thumb’ is to limit the generated waveforms to less than 40% of the clock frequency. The resolution, or tuning sensitivity, of the waveform generator is given by : Incoming Clock Frequency Resolution = Hz 4) DC can be generated since the increment value can be zero. 5) Frequency stability will match the stability of the incoming frequency when the increment is fixed. The residual noise characteristics of an oscillator are very important in modern communication systems. This parameter defines how well the device maintains its set frequency for very short periods (nanoseconds to seconds) of time. Poor figures will significantly affect the system signal to noise ratio and limit the dynamic range. The PDSP16350 will, of course, inherit the residual noise characteristics of the source of the incoming frequency. The output frequency is, however, always less than half the incoming frequency in order to satisfy the Nyquist criterion. This is in contrast to a phase locked loop synthesiser, when a small input frequency controls a high output frequency. The commonly used 20 log N rule states that the phase noise at the output of a synthesiser will be no better than twenty times the log of the ratio of the output frequency to the input frequency. In a phase locked loop synthesiser N is large, in the PDSP16350 it is less than half. Log N is thus less than zero and phase noise improvement is obtained. The output waveforms are produced after a pipeline delay with respect to the DIN inputs. The effects of the JUMP or RES commands are delayed such that all data in the internal pipe will be processed before the discontinuity occurs. New data may be presented to the device on the cycle following the JUMP or RES and a valid result will be obtained after 31 clock cycles. 2N These equations illustrate some very important features of direct digital synthesisers :1) Tuning sensitivity is defined by both the number of bits in the accumulator and the incoming time base frequency. 2) The oscillator tunes linearly over its entire range. 3) The frequency accuracy matches the accuracy of the incoming increment value. 1 2 3 PI PI 4 5 30 31 32 33 34 35 CLK MODE RESET JUMP CEN DATA IN RESULT Device Reset A Apply phase increment B C D First Result Available Fig. 3 Fixed Frequency Timing Diagram 5 PDSP16350 USING THE PDSP16350 Frequency, phase, and amplitude modulation are all possible with the PDSP16350. The former two requirements are satisfied by the ability to change the phase increment value on every clock cycle. The latter needs the addition of two multipliers, which allow both sine and cosine to be modified by an incoming waveform. Fixed Frequency, Constant Amplitude To generate sine and cosine outputs at a fixed frequency, the MODE pin should be tied low, see Fig. 3. The phase increment value required to generate the desired frequency should be clocked into the internal phase increment register. This value is entered via the DIN port with CEN low. If CEN subsequently goes inactive (high), the value need not be maintained on the input pins. The correct phase increment value can be calculated as follows : Desired O/P Frequency x 234 DIN value = Clock Frequency This will give a decimal value which must be converted to a 34 bit binary number. The frequency resolution of the generated waveforms will be : A practical example can be used to illustrate the calculation. With a clock frequency of 10.73864 MHz, and the need to generate an output frequency of 20 kHz, then the above equation tells us we need a DIN value of 31996359. This corresponds to a binary value of: DIN33:0 = 00 0000 0001 1110 1000 0011 1001 1100 0111 The resolution would be 0.0006 Hz. It should be noted that the accuracy of the PDSP16350 cannot be any better than the accuracy of the incoming clock, and these resolutions are based on perfect incoming waveforms. Fixed Frequency, Modulated Amplitude The MODE pin should be high if modulation of the output waveforms is required. In this mode each of the output waveforms is multiplied by the 16 bit, two’s complement, value, present on the most significant 16 bits of the DIN port. The phase increment register is normally loaded with the 18 bit value on the least significant portion of the DIN bus. It is also possible to load the full 34 bits of the phase increment register when greater accuracy is required, this is explained below. When using the full 34 bits it is possible to obtain the same frequency resolution as in the fixed amplitude mode described earlier. When using 18 bit accuracy directly from the DIN bus the correct phase increment value can be calculated as follows : Clock Frequency Desired O/P Frequency Resolution = Hz x 218 DIN value = 234 Clock Frequency With a 20 MHz clock this results in a frequency resolution of 0.001 Hz. This can be improved by reducing the clock frequency, with the Nyquist restraint being the limiting factor. The latter states that the frequency of the generated waveform must be no more than 50% of the input clock. In practice 40% is a better limit to use, as previously discussed. 1 2 3 4 5 A B C D The frequency resolution is correspondingly reduced and given by : Clock Frequency Resolution = Hz 218 30 31 32 33 34 35 CLK MODE RESET JUMP DATA IN RESULT Device Reset A Apply first data C First Result Available Fig. 4 Amplitude Modulation (18bit frequency accuracy) 6 B D PDSP16350 It can also be used to generate the in-phase and quadraFig. 4 shows the operation of the device when loading the ture components of an analog waveform, which has been phase increment directly from the DIN bus. First the device digitized and which is to be processed using complex techmust be reset then data is presented on each clock cycle. The niques. Such a quadrature heterodyning system, alternatively amplitude modulation value is presented on the most signifiknown as an IQ splitter, is shown in Fig. 5. cant 16 bits while the phase increment is presented on the The output from an A/D converter drives the D33:18 inputs least significant 18 bits. The first valid result is obtained after of the PDSP16350. If all sixteen inputs are not required, the 31 cycles. (In this mode the least significant 16 bits of the unused least significant bits should be tied to ground, and the phase increment register remain low). more significant inputs connected to the A/D converter. MultiFig.6 shows the operation of the device when using the full plying an input signal with a local oscillator in this manner 34 bits of the phase increment register. First the device must produces both sum and difference components. The former be reset, then the full 34 bits of the phase increment register can be removed by using the PDSP16256 Programmable FIR are loaded from the DIN bus by taking signal JUMP high Filter. before the rising edge of the clock. Following this new data can be presented on each cycle of the clock. The amplitude modulation value is presented on the most significant 16 bits while the phase increment is presented on the least significant 18 bits. The least significant 16 bits of the phase increment register remain fixed at the value loaded using JUMP. The first Analog valid result is obtained after 31 Data cycles. When using JUMP to load the phase increment regisPhase A/D EPROM ter, normal operation cannot be Increment Converter maintained. This is because the Value amplitude modulation value normally presented on the most D17:0 D33:18 significant 16 bits of the DIN bus I are replaced by part of the new phase increment value. The AM mode is useful in O/P PDSP PDSP16256 16 bit DATA 16350 systems requiring frequency Cordic Q Sin / Cos sweeps. By varying the ampliGenerator tudes at different frequencies, it is possible to compensate for the analog gain characteristics of amplifiers further along in the system. Fig. 5 IQ Split Function 1 2 3 4 5 PI A B 32 33 34 35 36 37 CLK MODE RESET JUMP DATA IN RESULT Device Reset A Apply phase Apply first data increment B C D First Result Available Fig. 6 Amplitude Modulation (34bit frequency accuracy) 7 PDSP16350 the transmitter, and the other for the receiver. The phase increment value is supplied by the counter block which simply increments at a rate determined by dividing down the time base clock. The synthesised frequency thus increases during the sweep period. A number of the more significant phase increment bits are used to supply the addresses to a PROM. The output of this PROM is used to amplitude modulate the sine and cosine waveforms. In this manner it is possible to compensate, at the source, for any poor frequency versus gain characteristics of analog circuits further along in the system. The digital outputs directly drive two D/A converters. Once in the analog world, it is necessary to remove the alias frequencies with low pass filters. The phase linearity and pass band ripple characteristics of these filters are very important, if the correct phase relationships are to be maintained between the two waveforms. Modulated Frequency The output frequency can be modulated very simply, see Fig 8. Since the phase increment value can be loaded as a complete word every cycle, there is no need to provide internal double buffering to prevent spurious frequencies being generated during the load operation. Binary Frequency Shift Keyed (BFSK) modulation can easily be implemented by externally multiplexing between two phase increment values representing the two frequencies to be used. The value to be used can be instantaneously changed, thus maintaining phase coherence, whilst the bit to be transmitted changes from a mark to a space. Frequency hopping could also be simply effected by clocking a new random number into the DIN port once every thousand cycles, for instance. The output will reflect any change in the frequency after 31 system clock cycles. If the phase increment value on the DIN port is changed on each clock cycle, then the output frequency will change without introducing any discontinuities. Thus, a linear frequency sweep can be achieved by incrementing the Gain value on the DIN port by a SWEEP Compensate GENERATOR fixed amount each cycle. AlROM ternatively, a logarithmic sweep could be implemented by ‘walking’ a one across the DIN port. Shifting the input D17:0 D33:18 one place to the left every hundred cycles, for example, would double the frequency every time. PDSP 16 bit Chirp generation for FM 16350 Cordic CW Radar systems is a typiSin / Cos cal example of the need for Generator linear frequency sweeps. This application requires the generation of quadrature chirp waveforms and is illustrated in simplified form by Fig. 7. One waveform is needed for D/A SIN D/A COS Fig. 7 Quadrature Chirp Generator 1 2 3 4 5 A B C D 30 31 32 33 34 35 CLK MODE RESET JUMP DATA IN RESULT Device Reset A Apply First Data First Result Available Fig. 8 Frequency Modulation Timing Diagram 8 B C D PDSP16350 Relative phase jumps may be made with or without amplitude modulation. For example, if a jump of 180 degrees is required, this can be done with a value of : Alternatively, if no amplitude modulation is needed, an absolute jump to a phase value can be made, see Fig. 9. This can be done by activating the JUMP input during one cycle and also presenting the new phase value at the same time. For example, if a jump to 270 degrees is required : DIN33:0 = 10 0000 0000 0000 0000 0000 0000 0000 0000 DIN33:0 = 11 0000 0000 0000 0000 0000 0000 0000 0000 This is loaded into the phase increment register for one cycle, then the normal increment value is re-loaded in the following cycle. The RES (reset) input can alternatively be used if a jump to 0 degrees is needed. This avoids using the DIN inputs and can be used with or without amplitude modulation. The reset function is internally synchronised to the input clock. Modulated Phase 1 2 3 4 5 A B C D 30 31 32 33 34 35 CLK MODE RESET JUMP DATA IN RESULT Device Reset A Apply First Data Absolute Phase Jump First Result Available B C D Absolute phase jump result Fig. 9 Phase Modulation Timing Diagram 9 PDSP16350 NOTES 1. Exceeding these ratings may cause permanent damage. Functional operation under these conditions is not implied. 2. Maximum dissipation or 1 second should not be exceeded, only one output to be tested at any one time. 3. Exposure to absolute maximum ratings for extended periods may affect device reliablity. 4. Vcc = Max, Outputs Unloaded, Clock Freq = Max. 5. CMOS levels are defined as VIH = VDD - 0.5v VIL = +0.5v 6. Current is defined as positive into the device. 7. The øJC data assumes that heat is extracted from the top face of the package. ABSOLUTE MAXIMUM RATINGS (Note 1) Supply voltage Vcc -0.5V to 7.0V Input voltage VIN -0.5V to Vcc + 0.5V Output voltage VOUT -0.5V to Vcc + 0.5V Clamp diode current per pin IK (see note 2) 18mA Static discharge voltage (HMB) 500V Storage temperature TS -65°C to 150°C Ambient temperature with power applied TAMB Military -55°C to +125°C Industrial -40°C to 85°C Junction temperature 150°C Package power dissipation 3500mW Thermal resistances Junction to Case øJC 5°C/W ELECTRICAL CHARACTERISTICS Operating Conditions (unless otherwise stated) Commercial: TAMB = 0°C to +70°C TJ(MAX) = 95°C VCC = 5.0V±5% Ground = 0V Industrial: TAMB = -40°C to +85°C TJ(MAX) = 110°C VCC = 5.0V±10% Ground = 0V Military: TAMB = -55°C to +125°C TJ(MAX) = 150°C VCC = 5.0V±10% Ground = 0V Static Characteristics Characteristic Symbol Min. Output high voltage Output low voltage Input high voltage Input low voltage Input leakage current Input capacitance Output leakage current Output S/C current VOH VOL VIH VIL IIN CIN IOZ ISC Value Typ. 2.4 3.0 -10 Units 0.4 0.8 +10 10 -50 40 Conditions Max. +50 250 V V V V µA pF µA mA IOH = 4mA IOL = -4mA GND < VIN < VCC GND < VOUT < VCC VCC = Max Switching Characteristics Characteristic D33:0 signal setup to clock rising edge D33:0 signal hold after clock rising edge CEN setup to clock rising edge CEN hold after clock rising edge JUMP, RES setup to clock rising edge JUMP hold after clock rising edge RES hold after clock rising edge Clock rising edge to output valid Clock freq Clock High Time Clock Low Time OES,OEC low to data valid OES,OEC high to data high impedance Pipeline delay VIN to VOUT Vcc Current (CMOS inputs) Vcc Current (TTL inputs) 10 Industrial Min. 15 4 20 0 10 6 8 5 DC 15 20 31 - Typ. Max. 30 20 20 20 31 430 460 Military Min. 15 4 20 0 10 6 8 5 DC 15 20 31 - Typ. Units Max. ns ns ns ns ns ns ns 30 ns 20 MHz ns ns 20 ns 20 ns 31 CLKs 450 mA 500 mA Conditions 30pF 30pF 30pF See Note 4 See Note 4 PDSP16350 ORDERING INFORMATION Industrial (-40°C to +85°C) PDSP16350 / B0 / AC (20MHz - PGA) PDSP16350 / B0/ GC (20MHz - QFP) Military (-55°C to +125°C) PDSP16350 / A0 / AC (20MHz - PGA) PDSP16350 / A0/ GC (20MHz - QFP) 11 PDSP16350 HEADQUARTERS OPERATIONS MITEL SEMICONDUCTOR Cheney Manor, Swindon, Wiltshire SN2 2QW, United Kingdom. Tel: (01793) 518000 Fax: (01793) 518411 MITEL SEMICONDUCTOR 1500 Green Hills Road, Scotts Valley, California 95066-4922 United States of America. Tel (408) 438 2900 Fax: (408) 438 5576/6231 Internet: http://www.gpsemi.com CUSTOMER SERVICE CENTRES ● FRANCE & BENELUX Les Ulis Cedex Tel: (1) 69 18 90 00 Fax : (1) 64 46 06 07 ● GERMANY Munich Tel: (089) 419508-20 Fax : (089) 419508-55 ● ITALY Milan Tel: (02) 6607151 Fax: (02) 66040993 ● JAPAN Tokyo Tel: (03) 5276-5501 Fax: (03) 5276-5510 ● KOREA Seoul Tel: (2) 5668141 Fax: (2) 5697933 ● NORTH AMERICA Scotts Valley, USA Tel: (408) 438 2900 Fax: (408) 438 5576/6231 ● SOUTH EAST ASIA Singapore Tel:(65) 3827708 Fax: (65) 3828872 ● SWEDEN Stockholm Tel: 46 8 702 97 70 Fax: 46 8 640 47 36 ● TAIWAN, ROC Taipei Tel: 886 2 25461260 Fax: 886 2 27190260 ● UK, EIRE, DENMARK, FINLAND & NORWAY Swindon Tel: (01793) 726666 Fax : (01793) 518582 These are supported by Agents and Distributors in major countries world-wide. © Mitel Corporation 1998 Publication No. DS3711 Issue No. 2.3 September 1996 TECHNICAL DOCUMENTATION – NOT FOR RESALE. PRINTED IN UNITED KINGDOM This publication is issued to provide information only which (unless agreed by the Company in writing) may not be used, applied or reproduced for any purpose nor form part of any order or contract nor to be regarded as a representation relating to the products or services concerned. No warranty or guarantee express or implied is made regarding the capability, performance or suitability of any product or service. The Company reserves the right to alter without prior notice the specification, design or price of any product or service. Information concerning possible methods of use is provided as a guide only and does not constitute any guarantee that such methods of use will be satisfactory in a specific piece of equipment. It is the user's responsibility to fully determine the performance and suitability of any equipment using such information and to ensure that any publication or data used is up to date and has not been superseded. These products are not suitable for use in any medical products whose failure to perform may result in significant injury or death to the user. All products and materials are sold and services provided subject to the Company's conditions of sale, which are available on request. All brand names and product names used in this publication are trademarks, registered trademarks or trade names of their respective owners. 12