ADVANCED LINEAR DEVICES, INC. ALD500RAU/ALD500RA/ALD500R PRECISION INTEGRATING ANALOG PROCESSOR WITH PRECISION VOLTAGE REFERENCE APPLICATIONS BENEFITS • • • • • • • • • • • Low cost, simple functionality Wide dynamic signal range Very high noise immunity Automatic compensation and cancellation of error sources • Easy to use to acquire bipolar signals • Up to 19 bit (18 bit + sign bit) single conversion or 21 bit (20 bit + sign bit) multiple conversion and noise performance • Inherently linear and stable with temperature and component variations 4 1/2 digits to 5 1/2 digits plus sign measurements Precision analog signal processor Precision sensor interface High accuracy DC measurement functions Portable battery operated instruments Computer peripheral PCMCIA FEATURES • • • • • • • • • • • • • • • • Resolution up to 18 bits plus sign bit and over-range bit Accuracy independent of input source impedances Accurate on-chip voltage reference Tempco as low as 10 ppm/°C guaranteed Chip select - power down mode High input impedance of 1012 Ω Inherently filters and integrates any external noise spikes Differential analog input Wide bipolar analog input voltage range ±3.5V Automatic zero offset compensation Low linearity error - as low as 0.001% typical Fast zero-crossing comparator - 1µs Low power dissipation - 6mW typical Automatic internal polarity detection Low input current - 2pA typical Optional digital control from a microcontroller, an ASIC, or a dedicated digital circuit • Flexible conversion speed vs. resolution trade-off PIN CONFIGURATION ALD500R IB 1 20 CS CINT 2 19 V+ V- 3 18 DGND CAZ 4 17 COUT B BUF 5 16 AGND 6 15 A C-REF 7 14 V+IN C+REF 8 13 V-IN N/C 9 12 V+REF N/C 10 11 V-REF QE, PE, SE PACKAGE * N/C pin is connected internally. Connect to V-. Ordering Information Resolution Endpoint Linearity Voltage Reference Accuracy/Tempco Package Type 20L PDIP 16 bit 20L SOIC 20L QSOP Operating Temperature 20LCDIP 0.015% 0.5% 50ppm/C ALD500R-50PE ALD500R-50SE ALD500R-50QE 0°C to 70°C 17 bit 0.01% 0.3% 20ppm/C ALD500RA-20PE ALD500RA-20SE ALD500RA-20QE 0°C to 70°C 18 bit 0.005% ALD500RAU-20PE ALD500RAU-20SE ALD500RAU-20QE 17 bit 0.01% 0.2% 10ppm/C ALD500RA-10PE ALD500RA-10SE ALD500RA-10QE 18 bit 0.005% ALD500RAU-10PE ALD500RAU-10SE ALD500RAU-10QE 17 bit 0.01% 0.3% 20ppm/C ALD500RA-20PEI ALD500RA-20SEI ALD500RA-20QEI 18 bit 0.005% ALD500RAU-20PEI ALD500RAU-20SEI ALD500RAU-20QEI 18 bit 0.005% 0.3% 20ppm/C 0°C to 70°C -40°C to +85°C ALD500RAU-20DE -55°C to +125°C * Contact factory for customized voltage reference voltage levels, accuracy and tempco specifications. Rev. 1.01 © 1999 Advanced Linear Devices, Inc., 415 Tasman Drive, Sunnyvale, California 94089-1706 Tel: (408) 747-1155, Fax: (408) 747-1286 http://www.aldinc.com FIGURE 1. ALD500R Functional Block Diagram V-REF (11) V+REF (12) RREF = 100KΩ CB = 0.1µF RREF C+REF CINT RINT CREF CAZ C-REF (7) (8) CAZ (4) BUF (5) CINT (2) REFINT SWR SWR Buffer + SWIN V+IN Integrator - (14) SW-R SW+R + + Comp1 - Level Shift Comp2 + COUT (17) SWAz SW+R AGND (6) SWIN SW-R SWS SWAZ Polarity Detection DGND (18) SWG V-IN (13) REF Control VSS VDD N/C N/C (3) (19) (10) (9) Bias IB (1) CB Analog Switch Control Signals CS (20) Phase Decoding Logic A B (15) (16) Control Logic GENERAL DESCRIPTION GENERAL THEORY OF OPERATION The ALD500RAU/ALD500RA/ALD500R are integrating dual slope analog processors, designed to operate on ±5V power supplies for building precision analog-to-digital converters. The ALD500RAU/ALD500RA/ALD500R feature specifications suitable for 18 bit/17 bit/16 bit resolution conversion, respectively. Together with three capacitors, two resistors, and a digital controller, a precision Analog to Digital converter with auto zero can be implemented. The digital controller can be implemented by an external microcontroller, under either hardware (fixed logic) or software control. For ultra high resolution applications, up to 23 bit conversion can be implemented with an appropriate digital controller and software. Dual-Slope Conversion Principles of Operation The ALD500R series of analog processors accept differential inputs and the external digital controller first counts the number of pulses at a fixed clock rate that a capacitor requires to integrate against an unknown analog input voltage, then counts the number of pulses required to deintegrate the capacitor against a known internal reference voltage. This unknown analog voltage can then be converted by the microcontroller to a digital word, which is translated into a high resolution number, representing an accurate reading. This reading, when ratioed against the reference voltage, yields an accurate, absolute voltage measurement reading. The ALD500R analog processors consist of on-chip digital control circuitry to accept control inputs, integrating buffer amplifiers, analog switches, and voltage comparators and a highly accurate, ultra-stable voltage reference. It functions in four operating modes, or phases, namely auto zero, integrate, deintegrate, and integrator zero phases. At the end of a conversion, the comparator output goes from high to low when the integrator crosses zero during deintegration. ALD500R analog processors also provide direct logic interface to CMOS logic families. 2 The basic principle of dual-slope integrating analog to digital converter is simple and straightforward. A capacitor, CINT , is charged with the integrator from a starting voltage, VX, for a fixed period of time at a rate determined by the value of an unknown input voltage, which is the subject of measurement. Then the capacitor is discharged at a fixed rate, based on an external reference voltage, back to VX where the discharge time, or deintegration time, is measured precisely. Both the integration time and deintegration time are measured by a digital counter controlled by a crystal oscillator. It can be demonstrated that the unknown input voltage is determined by the ratio of the deintegration time and integration time, and is directly proportional to the magnitude of the external reference voltage. The major advantages of a dual-slope converter are: a. Accuracy is not dependent on absolute values of integration time tINT and deintegration time tDINT, but is dependent on their relative ratios. Long-term clock frequency variations will not affect the accuracy. A standard crystal controlled clock running digital counters is adequate to generate very high accuracies. b. Accuracy is not dependent on the absolute values of RINT and C INT, as long as the component values do not vary through a conversion cycle, which typically lasts less than 1 second. c. Offset voltage values of the analog components, such as VX, are cancelled out and do not affect accuracy. d. Accuracy of the system depends mainly on the accuracy and the stability of the voltage reference value. Advanced Linear Devices ALD500RAU/ALD500RA/ALD500R e. Very high resolution, high accuracy measurements can be achieved simply and at very low cost. An inherent benefit of the dual slope converter system is noise immunity. The input noise spikes are integrated (averaged to near zero) during the integration periods. Integrating ADCs are immune to the large conversion errors that plague successive approximation converters and other high resolution converters and perform very well in high-noise environments. The slow conversion speed of the integrating converter provides inherent noise rejection with at least a 20dB/decade attenuation rate. Interference signals with frequencies at integral multiples of the integration period are, theoretically, completely removed. Integrating converters often establish the integration period to reject 50/60Hz line frequency interference signals. The relationship of the integrate and deintegrate (charge and discharge) of the integrating capacitor values are shown below: (1) VX = VINT - (VREF . tDINT / RINT . CINT) (deintegrate cycle) (2) Combining equations 1 and 2 results in: VIN / VREF = -tDINT / tINT Actual data conversion is accomplished in two phases: Input Signal Integration Phase and Reference Voltage Deintegration Phase. The integrator output is initialized to 0V prior to the start of Input Signal Integration Phase. During Input Signal Integration Phase, internal analog switches connect VIN to the buffer input where it is maintained for a fixed integration time period (tINT). This fixed integration period is generally determined by a digital counter controlled by a crystal oscillator. The application of VIN causes the integrator output to depart 0V at a rate determined by VIN and a direction determined by the polarity of VIN. The Reference Voltage Deintegration Phase is initiated immediately after tINT , within 1 clock cycle. During ReferenceVoltage Deintegration Phase, internal analog switches connect a reference voltage having a polarity opposite that of VIN to the integrator input. Simultaneously the same digital counter controlled by the same crystal oscillator used above is used to start counting clock pulses. The Reference Voltage Deintegration Phase is maintained until the comparator output inside the dual slope analog processor changes state, indicating the integrator has returned to 0V. At that point the digital counter is stopped. The Deintegration time period (tDINT ), as measured by the digital counter, is directly proportional to the magnitude of the applied input voltage. VINT = VX - (VIN . tINT / RINT . CINT ) (integrate cycle) reference voltage is integrated VREF = Reference Voltage CINT = Integrating Capacitor value RINT = Integrating Resistor value (3) where: Vx = An offset voltage used as starting voltage VINT = Voltage change across CINT during tINT and during tDINT (equal in magnitude) VIN = Average, or an integrated, value of input voltage to be measured during tINT (Constant VIN ) tINT = Fixed time period over which unknown voltage is integrated tDINT = Unknown time period over which a known After the digital counter value has been read, the digital counter, the integrator, and the auto zero capacitor are all reset to zero through an Integrator Zero Phase and an Auto Zero Phase so that the next conversion can begin again. In practice, this process is usually automated so that analog-todigital conversion is continuously updated. The digital control is handled by a microprocessor or a dedicated logic controller. The output, in the form of a binary serial word, is read by a microprocessor or a display adapter when desired. CINT RINT INTEGRATOR - ANALOG INPUT (VIN) VINT - COMPARATOR + S1 VOLTAGE REFERENCE REF SWITCHES COUT + POLARITY DETECTION SWITCH DRIVER PHASE CONTROL CONTROL LOGIC POLARITY CONTROL INTEGRATOR OUTPUT A VIN ≈ VFULL SCALE VINT = 4.1V MAX VIN ≈ 1/2VFULL SCALE Vx ≈ 0 tDINT tINT B MICROCONTROLLER (CONTROL LOGIC + COUNTER) tDINT Figure 2. Basic Dual-Slope Converter ALD500RAU/ALD500RA/ALD500R Advanced Linear Devices 3 ABSOLUTE MAXIMUM RATINGS Supply voltage, V+ Differential input voltage range Power dissipation Operating temperature range PE, SE package Operating temperature range QE package Storage temperature range Lead temperature, 10 seconds 13.2V -0.3V to V+ +0.3V 600 mW 0°C to +70°C -55°C to +125°C -65°C to +150°C +260°C OPERATING ELECTRICAL CHARACTERISTICS TA = 25°C V+ = +5V V- = -5V (V supply ± 5V) unless otherwise specified; CAZ = CREF = 0.47µf 500RAU Parameter Symbol Resolution Typ 500RA Max 15 Zero-Scale Error ZSE End Point Linearity ENL Best Case Straight Line Linearity NL Zero-Scale Temperature Coefficient TCZS Full-Scale Symmetry Error (Rollover Error) SYE Full-Scale Temperature Coefficient TCFS Input Current IIN Common-Mode Voltage Range Min Min Typ Min 30 0.001 Typ Max Unit Test Conditions µV Notes 1, 7 0.005 0.008 % % 0°C to 70°C 60 0.0025 0.003 0.003 0.005 0.005 0.007 0.003 0.010 0.015 0.005 0.015 0.020 % Notes 1, 2 0°C to +70°C 0.0025 0.003 0.005 0.003 0.008 % Notes 1, 2 0.004 CMVR VSS+1.5 500R Max 0.008 0.015 0.3 0.6 0.3 0.7 0.3 0.7 0.15 0.3 0.15 0.35 0.15 0.35 0°C to +70°C µV/°C 0°C to +70°C ppm/°C Notes 1, 7 0°C to 70°C 0.005 0.008 0.01 % 0.008 0.010 0.012 % 1.3 1.3 1.3 2 2 2 ppm/°C 0°C to +70°C Note 7 pA VIN = 0V VDD-1.5 V SS+1.5 VDD-1.5 VSS+1.5 VDD-1.5 V Integrator Output Swing VINT VSS+0.9 VDD-0.9 V SS+0.9 VDD-0.9 VSS+0.9 VDD-0.9 V Analog Input Signal Range VIN VSS+1.5 VDD-1.5 V SS+1.5 VDD-1.5 VSS+1.5 VDD-1.5 V Voltage Reference Range VREF VDD-1 V 4 VSS+1 VDD-1 VSS+1 VDD-1 VSS+ 1 Advanced Linear Devices AGND = 0V ALD500RAU/ALD500RA/ALD500R DC & AC ELECTRICAL CHARACTERISTICS TA = 25°C V supply = ±5.0V unless otherwise specified; CAZ = CREF = 0.47µf 500RAU Parameter Symbol Min Typ 500RA Max 0.6 Min Typ 1.0 500R Max 0.6 Min Typ Unit Test Conditions 1.0 mA V+ = 5V , A =1,B=1 10 mW V supply= ±5V 4.5 5.5 V Note 4 -4.5 -5.5 V Note 4 V ISOURCE = 400µA V ISINK = 1.1mA 1.0 IS Power Dissipation PD Positive Supply Range V+S 4.5 5.5 4.5 5.5 Negative Supply Range V-S -4.5 -5.5 -4.5 -5.5 Comparator Logic 1, Output High VOH Comparator Logic 0, Output Low VOL Logic 1, Input High Voltage VIH Logic 0, Input Low Voltage VIL Logic Input Current IL 0.01 0.01 0.01 Comparator Delay tD 1 1 1 10 10 4 4 4 0.4 3.5 Max 0.6 Supply Current 0.4 3.5 0.4 3.5 1 V 1 1 V µA µsec Note 5 Figure 3. ALD500R TIMING DIAGRAM 1 Conversion Cycle ~ ~ ~ ~ ~ ~ 1.8432 MHz Clock 123,093 Clock Pulses ~ ~ 0.5416 µs 123,093 Clock Pulses ~ ~ ~ ~ 66.667 msec. A INPUT ~ ~ 66.667 msec. COUT Positive Input Signal NOT VALID COUT Negative Input Signal NOT VALID START CONVERSION CYCLE ~ ~ ~ ~ B INPUT Auto Zero Phase Input Signal Integration Phase Reference Voltage Deintegration Phase Integrator Zero Phase Clock data in or clock data out of counters within the the microcontroller or fixed logic controller, as needed. Fixed number of clock pulses by design. Variable number of clock pulses. Fixed period of approx.1 msec. At VIN MAX, max. number of clock pulses ~ 246,185 = Auto Zero Phase Stop counter upon detection of comparator output going from high to low state. REPEAT CONVERSION CYCLE START INTEGRATION CYCLE START DEINTEGRATION CYCLE ALD500RAU/ALD500RA/ALD500R Advanced Linear Devices START INTEGRATOR ZERO CYCLE 5 DC ELECTRICAL CHARACTERISTICS TA = 25°C V supply = +5.0V unless otherwise specified; CAZ = CREF = 0.47µf, RREF = 100KΩ (1% metal film) Parameter Symbol Min 0.998 500RAU-10 500RAU-20 500RA-10 500RA-20 Typ Max Min Typ 1.000 1.002 0.997 1.000 Voltage Reference VR Supply Voltage Rejection Ratio VRSR 95 Temperature Coefficient VRTC 3 Long Term Drift ∆VREF / ∆t Warm Up Time 500RAU-50 Max 1.003 0.995 95 10 Unit 1.000 1.005 95 7 20 -0.08 -0.08 10 10 10 V dB 15 -0.08 50 ppm/C° ppm/ 1000hrs Note 7 min. Note 7 Positive Supply Operating Voltage Range V+ 4.5 5.5 4.5 5.5 4.5 5.5 V Negative Supply Operating Voltage Range V- -5.5 -4.5 -5.5 -4.5 -5.5 -4.5 V Power Down Supply Current IPD 25 25 Test Conditions 500RA-50 500R-50 Min Typ Max µA 25 Note 7 NOTES: 1. Integrate time ≥ 66 msec., Auto Zero time ≥ 66 msec., VINT = 4V, VIN = 2.0V Full Scale Resolution = VINT /integrate time/clock period 2. End point linearity at ±1/4, ±1/2, ±3/4 Full Scale after Full Scale adjustment. 3. Rollover Error also depends on CINT, CREF , CAZ characteristics. ~ 4. Contact factory for other power supply operating voltage ranges, including Vsupply = ±3V or Vsupply = ±2.5V. 5. Recommended selection of clock periods of one of the following: t clk = 0.27µsec, 0.54µsec, or 1.09µsec which corresponds to clock frequencies of 3.6864 MHz, 1.8432 MHz, 0.9216 MHz respectively. 6. RREF is 100K Ω 1% metal film, 50 ppm/C. 7. Sample tested parameter. 6 Advanced Linear Devices ALD500RAU/ALD500RA/ALD500R PIN DESCRIPTION Pin No. Symbol Description 1 IB Bias circuit pin. Connect a 0.1µF capacitor from this pin to V- to minimize noise. 2 CINT Integrator capacitor connection. 3 V- Negative power supply. 4 CAZ The Auto-zero capacitor connection. 5 BUF The Integrator resistor buffer connection. 6 AGND This pin is analog ground. 7 C-REF Negative reference capacitor connection. 8 C+REF Positive reference capacitor connection. 9 N/C Internally connected. Connect to V- for normal operation. 10 N/C Internally connected. Connect to V- for normal operation. 11 V-REF External voltage reference (-) connection. High impedance load (≥100MΩ) only. 12 V+ REF External voltage reference (+) connection. High impedance load (≥100MΩ) only. 13 V-IN Negative analog input. 14 V+ IN Positive analog input. 15 A Converter phase control MSB Input. 16 B Converter phase control LSB Input. 17 COUT Comparator output. COUT is HIGH during the Integration phase when a positive input voltage is being integrated and is LOW when a negative input voltage is being integrated. A HIGH-to-LOW transition on COUT signals the processor that the Deintegrate phase is completed. COUT is undefined during the Auto-Zero phase. It should be monitored to time the Integrator Zero phase. 18 DGND Digital ground. 19 V+ Positive power supply. 20 CS Chip select - power down pin. Logic 1 = power on. Logic 0 = power down. Table 1. Conversion Phase and Control Logic Internal Analog Switch Functions Switch Functions Conversion Phase Input Connect Reference Input Polarity SWIN SW+R or SW- R Auto Zero Reference Sample VIN=AGND System Offset SWAZ SWR SWG SWS Control Logic Auto Zero A = 0, B = 1 Open Open Closed Closed Closed Open Input Signal Integration A = 1, B = 0 Closed Open Open Open Open Open Reference Voltage Deintegration A = 1, B = 1 Open Closed* Open Open Closed Open Integrator Output Zero A = 0, B = 0 Open Open Open Closed Closed Closed *SW+R would be closed for a positive input signal. SW-R would be closed for a negative input signal. ALD500RAU/ALD500RA/ALD500R Advanced Linear Devices 7 ALD500RAU/ALD500RA/ALD500R CONVERSION CYCLE The ALD500RAU/ALD500RA/ALD500R conversion cycle takes place in four distinct phases, the Auto Zero Phase, the Input Signal Integration Phase, the Reference Voltage Deintegration Phase, and the Integrator Zero Phase. A typical measurement cycle uses all four phases in an order sequence as mentioned above. The internal analog switch status for each of these phases is summarized in Table 1. time and depends on system parameters and component value selections. The total number of clock pulses or clock counts, during integration phase determine the resolution of the conversion. For high resolution applications, this total number of clock pulses should be maximized. The basic unit of resolution is in µV/count. Before the end of this phase, comparator output is sampled by the microcontroller. This phase is terminated by changing logic inputs AB from 10 to 11. Reference Voltage Deintegration Phase (DINT Phase) The following is a detailed description of each one of the four phases of the conversion cycle. Auto Zero Phase (AZ Phase) The analog-to-digital conversion cycle begins with the Auto Zero Phase, when the digital controller applies low logic level to input A and high logic level to input B of the analog processor. During this phase, the reference voltage is stored on reference capacitor CREF, comparator offset voltage and the sum of the buffer and integrator offset voltages are stored on auto zero capacitor CAZ. During the Auto Zero Phase, the comparator output is characterized by an indeterminate waveform. During the Auto Zero Phase, the external input signal is disconnected from the internal circuitry of the ALD500RAU/ ALD500RA/ALD500R by opening the two SW IN analog switches and connecting the internal input nodes internally to analog ground. A feedback loop, closed around the integrator and comparator, charges the C AZ capacitor with a voltage to compensate for buffer amplifier, integrator and comparator offset voltages. This is the system initialization phase, when a conversion is ready to be initiated at system turn-on. In practice the converter can be operated in continuous conversion mode, where AZ phase must be long enough for the circuit conditions to settle out any system errors. Typically this phase is set to be equal to tINT. Input Signal Integration Phase (INT Phase) During the Input Signal Integration Phase (INT), the ALD500RAU/ALD500RA/ALD500R integrates the differential voltage across the (V +IN) and (V-IN) inputs. The differential voltage must be within the device's common-mode voltage range CMVR. The integrator charges CINT for a fixed period of time, or counts a fixed number of clock pulses, at a rate determined by the magnitude of the input voltage. During this phase, the analog inputs see only the high impedance of the noninverting operational amplifier input of the buffer. The integrator responds only to the voltage difference between the analog input terminals, thus providing true differential analog inputs. The input signal polarity is determined by software control at the end of this phase: COUT = 1 for positive input polarity; COUT = 0 for negative input polarity. The value is, in effect, the sign bit for the overall conversion result. At the end of the Input Signal Integration Phase, Reference Voltage Deintegration Phase begins. The previously charged reference capacitor is connected with the proper polarity to ramp the integrator output back to zero. The ALD500RAU/ ALD500RA/ALD500R analog processors automatically selects the proper logic state to cause the integrator to ramp back toward zero at a rate proportional to the reference voltage stored on the reference capacitor. The time required to return to zero is measured by the counter in the digital processor using the same crystal oscillator. The phase is terminated by the comparator output after the comparator senses when the integrator output crosses zero. The counter contents are then transferred to the register. The resulting time measurement is proportional to the magnitude of the applied input voltage. The duration of this phase is precisely measured from the transition of AB from 10 to 11 to the falling edge of the comparator output, usually with a crystal controlled digital counter chain. The comparator delay contributes some error in this phase. The typical comparator delay is 1µsec . The comparator delay and overshoot will result in error timing, which translates into error voltages. This error can be zeroed and minimized during Integrator Output Zero Phase and corrected in software, to within ±1 count of the crystal clock (which is equivalent to within ± 1 LSB, when 1 clock pulse = 1 LSB). Integrator Zero Phase ( INTZ Phase) This phase guarantees the integrator output is at 0V when the Auto Zero phase is entered, and that only system offset voltages are compensated. This phase is used at the end of the reference voltage deintegration and is used for applications with high resolutions. If this phase is not used, the value of the Auto-Zero capacitor (CAZ) must be much greater than the value of the integration capacitor (CINT) to reduce the effects of charge-sharing. The Integrator Zero phase should be programmed to operate until the Output of the Comparator returns "HIGH". A typical Integrator Zero Phase lasts 1msec. The comparator delay and the controller's response latency may result in Overshoot causing charge buildup on the integrator at the end of a conversion. This charge must be removed or performance will degrade. The Integrator Output Zero phase should be activated (AB = 00) until COUT goes high. At this point, the integrator output is near zero. Auto Zero Phase should be entered (AB = 01) and the ALD500RAU/ ALD500RA/ALD500R is held in this state until the next conversion cycle. The duration of this phase is selected by design to be a fixed 8 Advanced Linear Devices ALD500RAU/ALD500RA/ALD500R Differential Inputs (V+IN ,V-IN) The ALD500RAU/ALD500RA/ALD500R operates with differential voltages within the input amplifier common-mode voltage range. The amplifier common-mode range extends from 1.5V below positive supply to 1.5V above negative supply. Within this common-mode voltage range, commonmode rejection is typically 95dB. The integrator output also follows the common-mode voltage. When large common-mode voltages with near full-scale differential input voltages are applied, the input signal drives the integrator output to near the supply rails where the integrator output is near saturation. Under such conditions, linearity of the converter may be adversely affected as the integrator swing can be reduced. The integrator output must not be allowed to saturate. Typically, the integrator output can swing to within 0.9V of either supply rails without loss of linearity. Analog Ground Analog Ground is V-IN during Auto Zero Phase and Reference Voltage Deintegration Phase. If V-IN is different from analog ground, a common-mode voltage exists at the inputs. This common mode signal is rejected by the high common mode rejection ratio of the converter. In most applications, V-IN is set at a fixed known voltage (i.e., power supply ground). All other ground connections should be connected to digital ground in order to minimize noise at the inputs. Differential Reference (V+REF, V-REF) The reference voltage can be anywhere from 1V of the power supply voltage rails of the converter. Roll-over error is caused by the reference capacitor losing or gaining charge due to the stray capacitance on its nodes. The difference in reference for (+) or (-) input voltages will cause a roll-over error. This error can be minimized by using a large reference capacitor in comparison to the stray capacitance. Phase Control Inputs (A, B) The A and B logic inputs select the ALD500RAU/ALD500RA/ ALD500R operating phase. The A and B inputs are normally driven by a microprocessor I/O port or external logic, using CMOS logic levels. For logic control functions of A and B logic inputs, see Table 1. Comparator Output (COUT) By monitoring the comparator output during the Input Signal Integration Phase, which is a fixed signal integrate time period, the input signal polarity can be determined by the microcontroller controlling the conversion. The comparator output is HIGH for positive signals and LOW for negative signals during the Input Signal Integration Phase. The state of the comparator should be checked by the microcontroller at the end of the Input Signal Integration Phase, just before transition to the Reference Voltage Deintegration Phase. For very low level input signals noise may cause the comparator output state to toggle between positive and negative states. For the ALD500RAU/ALD500RA/ALD500R, this noise has been minimized to typically within one count. At the start of the Reference Voltage Deintegration Phase, comparator output is set to HIGH state. During the Reference Voltage Deintegration Phase, the microcontroller must monitor the comparator output to make a HIGH-to-LOW transition as the integrator output ramp crosses zero relative to analog ground. This transition indicates that the conversion is complete. The microcontroller then stops and records the pulse count. The internal comparator delay is 1µsec, typically. The comparator output is undefined during the Auto Zero Phase. Positive Input Signal (VIN) 0V Negative Input Signal (VIN) ANALOG INPUT INTEGRATE ANALOG INPUT INTEGRATE INTEGRATOR OUTPUT (VINT) EXTERNAL INPUT POLARITY DETECTION COMPARATOR OUTPUT (COUT) REFERENCE DEINTEGRATE REFERENCE DEINTEGRATE INTEGRATOR OUTPUT (VINT) ZERO CROSSING ZERO CROSSING EXTERNAL INPUT POLARITY DETECTION COMPARATOR OUTPUT (COUT) Figure 4. Comparator Output ALD500RAU/ALD500RA/ALD500R Advanced Linear Devices 9 where: APPLICATIONS AND DESIGN NOTES Determination and Selection of System Variables The procedure outlined below allows the user to determine the values for the following ALD500RAU/ALD500RA/ALD500R system design variables: (1) (2) (3) (4) (5) (6) Determine Input Voltage Range Clock Frequency and Resolution Selection Input Integration Phase Timing Integrator Timing Components (RINT, CINT) Auto Zero and Reference Capacitors Voltage Reference System Timing Figure 3 and Figure 4 show the overall timing for a typical system in which ALD500RAU/ALD500RA/ALD500R is interfaced to a microcontroller. The microcontroller drives the A, B inputs with I/O lines and monitors the comparator output, COUT , using an I/O line or dedicated timer-capture control pin. It may be necessary to monitor the state of the comparator output in addition to having it control a timer directly during the Reference Deintegration Phase. There are four critical timing events: sampling the input polarity; capturing the deintegration time; minimizing overshoot and properly executing the Integrator Output Zero Phase. VIN MAX = Maximum input voltage desired (full count voltage) RINT = Integrating Resistor value For minimum noise and maximum linearity, RINT should be in the range of between 50kΩ to 150kΩ . Integrating Capacitor (CINT) The integrating capacitor should be selected to maximize integrator output voltage swing VINT, for a given integration time, without output level saturation. For +/-5V supplies, recommended VINT range is between +/- 3 Volt to +/-4 Volt. Using the 20µA buffer maximum output current, the value of the integrating capacitor is calculated as follows: CINT = (tINT ) . (20 x 10-6) / VINT where: tINT = VINT = Input Integration Phase Period Maximum integrator output voltage swing It is critical that the integrating capacitor must have a very low dielectric absorption, as charge loss or gain during conversion directly converts into an error voltage. Polypropylene capacitors are recommended while Polyester and Polybicarbonate capacitors may also be used in less critical applications. Reference (CREF) and Auto Zero (CAZ) Capacitors Selecting Input Integration Time For maximum 50/60 cycle noise rejection, Input Integration Time must be picked as a multiple of the period of line frequency. For example, tINT times of 33msec, 66msec and 100 msec maximize 60Hz line rejection, and 20msec, 40 msec, 80msec, and 100 msec maximize 50Hz line rejection. Note that tINT of 100 msec maximizes both 60 Hz and 50Hz line rejection. The duration of the Reference Deintegrate Phase (DINT) is a function of the amount of voltage charge stored on the integrator capacitor during INT phase, and the value of VREF. The D INT phase must be initiated immediately following INT phase and terminated when an integrator output zero-crossing is detected. In general, the maximum number of counts chosen for DINT phase is twice to three times that of INT phase with V REF chosen as a maximum voltage relative to VIN . For example, VREF = VIN(max)/2 would be a good reference voltage. Integrating Resistor (RINT ) The desired full-scale input voltage and amplifier output current capability determine the value of RINT. The buffer and integrator amplifiers each have a full-scale current of 20µA. The value of RINT is therefore directly calculated as follows: = VIN MAX / 20 µA Calculate VREF The reference deintegration voltage is calculated using: VREF = (VINT) . (CINT ) . (RINT) / 2(tINT) INT and DINT Phase Timing RINT C REF and CAZ must be low leakage capacitors (e.g. polypropylene types). The slower the conversion rate, the larger the value CREF must be. Recommended capacitor values for CREF and C AZ are equal to CINT. Larger values for CAZ and CREF may also be used to limit roll-over errors. The ALD500RAU/ALD500RA/ALD500R requires an external RREF in order to operate properly. This R REF should be a 1% metal film 100KΩ resistor, 50 ppm/C. Any other loading must be high impedance (≥100MΩ). Converter Noise The converter noise is the total algebraic sum of the integrator noise and the comparator noise. This value is typically 14 µV peak to peak. The higher the value of the reference voltage, the lower the converter noise. Such sources of noise errors can be reduced by increased integration times, which effectively filter out any such noise. If the integration time periods are selected as multiples of 50/60Hz frequencies, then 50/60Hz noise is also rejected, or averaged out. The signal-to-noise ratio is related to the integration time (tINT ) and the integration time constant (RINT) (CINT) as follows: S/N (dB) = 20 Log ((VINT / 14 x 10-6) . tINT /(RINT . CINT)) This converter noise can also be reduced by using multiple samples and mathematically averaged. For example, taking 16 samples and averaging the readings result in a mathematical (by software) filtering of noise to less than 4µV. 10 Advanced Linear Devices ALD500RAU/ALD500RA/ALD500R EQUATIONS AND DERIVATIONS Dual Slope Analog Processor equations and derivations are as follows: tINT .t 1 V (1) VIN(t)dt = REF DINT . RINT CINT 0 RINT . CINT For VIN(t) = VIN (constant): ∫ CINT = (2) (2a) tINT . IB VINT VIN IB = (3) VINMAX IBMAX (4) = 4 x 16.6667 msec. 2. Pick tINT = 4x 60Hz = 66.6667ms 3. Pick clock period = 1.08507 µs and number of counts over tINT = 0.0666667 = 61440 1.08507x10-6 4. Pick VINMAX value, e.g., VINMAX = 2.0 V I BMAX = 20µA (5a) 6. Pick CREF and CAZ ≥ CINT: CREF ~=CAZ ~= 0.33 µF 7. Pick tDINT = 2 x tINT = 133.3333 msec VIN MAX . tINT tDINT MAX (5b) 8. Calculate VREF = CINT . VINT IBMAX = VINMAX RINT VINTMAX . CINT . RINT tDINT MAX V (6) -6 3 = 4 x 0.33 x 10 x 100 x 10 133.3333 x 10-3 (7) ~ 1.00V = IB and 2.0 = 100 kΩ 20x10-6 5. Applying equation (3) to calculate CINT: Rearranging equations (3) and (4): tINT = RINT = CINT = (0.0666667)(20x10-6)/4 where VINT = 4.0V ~= 0.33 µF From equation (2a), VIN . tINT VREF = tDINT OR VREF = 1. Pick resolution = 16 bit. = 0.0666667 sec. tDINT tINT At VINMAX, the current IB is also at a maximum level, for a given RINT value: RINT = Design Example 1: 1 .t V tINT . VIN = REF DINT RINT . CINT RINT . CINT 1 ... VIN = VREF . DESIGN EXAMPLES We now apply these equations in the following design examples. V At VINT = VINT MAX, equation (6) becomes: . tINT = CINT VINTMAX IBMAX (6a) 1. Select resolution of 17 bit. Total number of counts during tINT is131,072. Combining (6a) and (7): . . ... tINT = CINT VINTMAX RINT VINMAX (8) In equation (5b), substituting equation (8) for tINT: CINT . VINTMAX . RINT VREF = = VIN MAX . VIN MAX tDINT MAX Design Example 2: (9) 2. We can pick tINT of 16.6667 msec. x 5 = 83.3333 msec. or alternately, pick t INT equal 16.6667 msec. x 6 = 100.00 msec. (for 60 Hz rejection) which is t INT = 20.00 msec. x 5 = 100.00 msec. (for 50 Hz rejection) CINT . VINTMAX . RINT Therefore, using t INT = 100 msec. would achieve both 50 Hz and 60 Hz cycle noise rejection. For this example, the following calculations would assume t INT of 100 msec. Now select period equal to 0.5425 µsec. (clock frequency of 1.8432 MHz) tDINT MAX For tDINT MAX = 2 x tINT, equation (9) becomes: VREF = CINT . VINTMAX . RINT (10) 2tINT ALD500RAU/ALD500RA/ALD500R Advanced Linear Devices 11 3. Pick VINMAX = ±2V Design Example 4: Objective: 5 1/2 digit + sign +over-range measurement. For I BMAX = 20µA, applying equation (4), RINT = 1. Pick tINT = 133.333 msec. for 60Hz noise rejection. (16.6667 msec. x 8 cycles) Frequency = 1.8432 MHz clock period = 0.5425 µsec. 2 = 100 K Ω 20x10-6 4. Calculate, using equation (3) for CINT: CINT = (0.1) x (20 x 10-6/4) ~ = 0.5 µF (assume V INTMAX = 4V) During Input Integrate Phase, total count = 133.333 x 10-3 0.5425 x 10-6 Use CINT 0.47µF as the closest practical value. = 245776 5. Pick CREF and CAZ = 0.47 µF For VINT = 4.0V, the basic resolution is 6. Pick tDINT = 2 x tINT = 200 msec. 4 or 16.276 µV/count 245776 7. Calculate the value for VREF, from equation (10): For VINMAX = 2.00V, the input resolution is 16.276 x VINMAX = 8.138 µV/count VINTMAX VREF = CINT . VINTMAX . RINT tDINT MAX = 0.5 x 10-6 x 4 x 100 x 103 2. Pick VIN range = ± 2V 200 x 10-3 2 = 100 KΩ For IB = 20 µA, RINT = 20 x 10-6 = 1.00V 3. Calculate CINT = (0.133333) x (20 x 10-6)/4 ~= 0.67 µF Design Example 3: 4. Pick CREF = CAZ = 0.67 µF 1. Pick resolution of 18 bit. Total number of counts during tINT is 262,144. 2. Pick tINT = 16.66667 msec. x 10 cycles 5. Select tDINT = 2 x tINT = 266.667 msec. 6. Calculate VREF as shown in Design Example 1, substituting the appropriate values: = 0.1666667 sec. This t INT allows clock period of 0.5425 µsec. and still achieve 18 bits resolution. VREF = CINT . VINTMAX . RINT tDINT MAX ~ = 1.005V 3. Again, as shown from previous example, pick VINMAX = ±2V For I BMAX = 20 µA, RINT = 2 = 100 KΩ 20x10-6 4. Next, we calculate CINT: C INT = (0.1666667) x (20 x 10-6)/4 ~ 0.83 µF (VINTMAX = 4.0V) = In this case, use CINT = 1.0 µF to keep V INT < 4.0V 5. Pick CREF and CAZ = 1.0 µF 6. Select tDINT = 2 x tINT = 333.333 msec. 7. Calculate VREF as shown in the previous examples and VREF = 1.00V 12 Advanced Linear Devices ALD500RAU/ALD500RA/ALD500R