introduction the analogic MP227A is a precision isolation amplifier that provides an unparalleled cost-effective combination of linearity, stability, and isolation. it is designed primarily to replace relays and filter elements in multichan- nel data acquisition systems. however, its unique features make it attrac- tive wherever low-level, low frequency signals must be applied in the presence of severe common mode interference. the MP227A offers user-selectable gains from 10 to 1000, input full- scale voltage ranges from 10 mv to 1v, 3-pole (60 db/decade) filtering from 5 hz, extremely good linearity, superb common-mode rejection, and very low drift. all parameters are commensurate with a/d conversion at levels up to 13 bits. the MP227A includes an internal power oscillator and isolated supply so that no external drivers are needed. the isolated power ( 4v nominal) can be used for open thermocouple indication or offsetting strain gauge inputs. features q high common mode rejection 170 db q excellent linearity 0.0075% q selectable input range 10 mv fs to 1v fs q low noise <0.5 ? rms q low drift 3.0 ? rti per month q built-in 3-pole filtering q built-in oscillator/driver applications q thermocouple temperature measurement q weighing systems q strain gauge measurements q remote data acquisition and precision telemetry systems q microvolt and millivolt level measurements q replacement for classical instrumentation amplifier figure 1. MP227A block diagram. precision isolation amplifier replaces relays and filter elements in multichannel dass MP227A isolated power supply & chopper drive choice of auto/zero or auto reference demod power drive +15v amp 22k 1.5 ? floating input guard isolated voltage available shielded metal case two-pole butterworth filter separate power ground chopper input for stability proprietary ultra linear flux cancelling coupler high quality bypass capacitors case at system ground ?v
all specifications guaranteed at 25? unless otherwise noted. analog input gain range 10 to 1000, non-inverting, resistor programmable; optimized for gains of 50 to 500 non-linearity ?.0075% fsr max. at g = 50 to 500 ?.01% fsr max. at g = 1000 ?.05% fsr max. at g = 10 input amplifier type isolated chopper linear differential input voltage range 10 mv to 1v full scale maximum safe differential input voltage 16v rms continuous, without damage common mode isolation voltage 1000 vdc, 750v rms max. common mode rejection ratio at dc, with g =100 & 1000, 166 db min.; 1 k source unbalance at 60 hz, with g = 100 & 1000, 176 db typ., 160 db min.; 1 k source unbalance common mode impedance 10,000 m / / 80 pf differential input impedance at dc, 10 m min.; at ac, low-pass filter of 22 k and 1.5 ? overload input impedance 22 k , at 50/60 hz input bias current 0.5 na typ., 3.0 na max.; bias current increases if open input indicator circuit used offset voltage at g = 10, 1 mv typ., ? mv max.; at g = 1000, 150 ? max.; all referred to input (rti); offset voltage may be determined by interpolation for other gain values voltage noise (0.01 to 5 hz) at g = 10, 1.5 ? rms max.; at g = 100 and 1000, 0.5 ? rms max.; rti bandwidth 1 dc to 5 hz nom.; 6 db down at 5 hz overall filtering 2 3-pole, 60 db/decade roll-off (?0 db at 50 hz) input filter 1-pole rc, 3 db cut-off at 5 hz output filter 2-pole butterworth, 3 db cut-off at 5 hz analog output voltage range 10v full scale output impedance at dc 0.1 maximum load ? ma and 500 pf output protection continuous short circuit to ground output chopper noise (1 mhz bw) 1 mv p-p spike at approximately 10 khz 3 stability gain tempco at g = 10 and 100, ?5 ppm fsr/? max.; at g = 1000, ?5 ppm fsr/? max.; exclusive of external gain setting resistor offset voltage tempco at g = 10, ?.0 ?/? max. at g = 100, 1.7 ?/? max.; at g = 1000, ?.5 ?/? max.; all rti bias current tempco 100 pa/? max., at 25?; doubles every 10? max. power supply sensitivity at g = 1000, ?.0 ?/%; at g = 10, 10 ?/% max.; rti warm-up drift (5 minutes) within 2 ? rti typ. at g = 1000 long term drift 3.0 ? rti/month typ. isolated power supply output voltage ? vdc nom., with respect to input lo current ? ma full load regulation 12%, no load to full load ripple 60 mv p-p at 10 khz input power supply requirements +15v, 3% 3 ma, no load ?5v, 3% 5 ma, no load environmental and mechanical operating temperature range 0? to +70? storage temperature range ?5? to +85? relative humidity 0 to 85% non-condensing up to 40? dimensions 1.2" x 2.8" x 0.5" (30 mm x 70 mm x 12 mm) shielding rfi: 6 sides; emi: 5 sides notes: 1. modifications for bandwidths from dc to 100 hz, or optimized for spe- cific settling times are available on special order. please contact factory. 2. filter nodes are externally accessi- ble to allow modification of charac- teristics. 3. output chopper noise can be re- duced to negligible level by suggest- ed output multiplexer circuit. specifications subject to change with- out notice. MP227A specifications
operation data application the MP227A was designed as an economically com- petitive and functionally superior alternative to the relay multiplexing circuits traditionally used in multichannel data acquisition systems. in a typical thermocouple system, the MP227A replaces three functional blocks for each channel ?the input filter and a dual relay, as well as the common channel high gain amplifier ?and permits high-level, solid-state multiplexing to be used for low cost and high reliability. the MP227A provides significantly better isolation and common-mode rejection than low-level relays and it puts the gain at a point in the system where the band- width is lowest (prior to multiplexing), thereby reducing total system noise. even where multiplexing is not used, the unusual combination of performance and price makes the MP227A attractive for a wide variety of industrial applications. when many MP227As are used in a system, a high- speed, high-level analog multiplexer switches the MP227A outputs to a common analog output bus for subsequent a/d conversion. any high precision isola- tion amplifier/filter used in such a configuration has an inherent error source of sizable magnitude that is often overlooked, ignored, or simply unknown; that is, dumped charge effects. this application note discuss- es the problem, the solution, and the fringe benefits. dumped charge figure 2 shows the apparently straightforward connec- tion of multiple amplifiers/filters and multiplexer to a common a/d converter. each time the multiplexer in figure 2 switches chan- nels, for instance, from channel 1 to channel 2, the channel 1 output appears across c2, the capacitance of the output bus. the output stage of channel 2 must absorb that dumped charge before it can reach a true final value dependent only on its input. the exact mag- nitude of the dumped charge is not important; what is significant is that the channel 2 amplifier may be forced to deliver a peak instantaneous current beyond its design specifications. the dumped charge (q) is defined as, q = idt, where i = c dv/dt in a typical example, the outputs of the two channels could be at the extreme ends of the range. channel 1 output = +10v. channel 2 output = ?0v. this makes the voltage difference (dv). dv = 20 volts. assume that the capacitance of the output bus c2 is about 500 pf, and a reasonable turn-on time for an analog switch is 100 ns, or, c = 500 (10 ?2 ) dt = 100 (10 ? ) solving first for the current and then the dumped charge, gives: 500(10 ?2 ) i = c (dv/dt) = 20 = 100 ma 100 (10 ? ) q = idt = (100)(10?)(100)(10?) = 10,000 pico coulomb under these conditions, ic op amps, such as the pop- ular 741, have been found to have full-scale current excursion lasting as long as a microsecond. if the design factors allow a conventional ic output stage to drive the multiplexer instead of a high preci- sion amplifier with an output/filter stage, no real harm is done by the dumped charge. the amplifier eventually recovers and c2 charges to the new value. the recov- ery time constant is the on resistance of the multiplex- er switch and c2 for r on = 300 c2 = 500 pf t = (300)(500) 10 ?2 = 0.15 ? amp amp amp multiplexer chan 1 chan 2 chan n c2 typ. 500 pf bus capacitance buffered adc figure 2. multiplexing amplifier outputs.
in high resolution systems, ten time constants should be allowed to reach a voltage within 0.005% of final value. therefore, the actual time should be 1.5 ?. the 1.5 ? settling time required in this example is usually less than the settling time of the conventional buffer amplifier at the multiplexer output, and the dumped charge effect can be safely ignored. the dumped charge cannot be ignored, however, when high precision amplifiers employing output filters are required. the problem many isolation and/or instrumentation amplifiers do not include an output filter. on the other hand, the MP227A has an integral two-pole butterworth filter in the output stage. the feedback element of the MP227A is a ca- pacitor, and a sudden voltage step at the amplifier out- put, such as the dumped charge, presents a problem. the dumped charge demands excessive current in too short a time and causes the amplifier to momentarily open-loop. the summing node changes to a large volt- age, inducing current flow in the input resistor and caus- ing an extraneous charge on the feedback capacitor. this error source has produced observed errors as large as 0.05% in typical applications. the solution figure 4 shows the addition of a single-pole filter (r1,c1) at the output of each MP227A and ahead of the multiplexer. c1 of the succeeding channel now ab- sorbs the charge accumulated on c2 from the preced- ing channel. the MP227A no longer sees a step but a well controlled exponential change, well within its ca- pabilities. hence, the output stage in the MP227A does not open-loop, and no spurious charge is placed on the feedback capacitor. the best results are obtained with a time constant be- tween 0.25 and 0.5 ?. this must be short for two rea- sons: 1) a settling time of up to 10rc does not signifi- cantly add to multiplexer settling time; and 2) the re- covery time is sufficiently short for final values that are independent of the duty cycle involved in reading a channel. r1 should be between 50 and 270 ; this value is kept intentionally low to reduce voltage divider error (r1 + ron relative to rin of the follower at the multiplexer output) to an insignificant level. these values of r1 yield values for c1 between 10,000 pf and 1,000 pf which is an acceptable range for c1. in the capacitive voltage divider, formed by c1 and the bus capacitance c2, as c1 decreases in size relative to c2, the initial voltage transferred to c1 by a succeeding channel ap- proaches its final value and leaves a smaller exponen- tial rise portion. r1, c1 must be included for all high reso- lution (>12 bits) applications of the MP227A. isolated power supply & chopper drive choice of auto/zero or auto reference demod power drive +15v a5 a6 amp 22k 1.5 ? input hi l2 auto zero return gain input lo ? out +v out guard +15v 15v ana rtn (case) feedback output pwr gnd ?v multiplexer chan 1 chan 2 chan n c2 c1 c1 c1 r1 r1 r1 MP227A MP227A MP227A buffered adc figure 3. MP227A isolation amplifier functional block diagram. figure 4. MP227A with output filters added.
fringe benefits noise spikes inherent in the design of high perfor- mance isolation amplifiers are attenuated by 10 db or more by the r1 c1 output filter. the superior isolation of the MP227A is provided by transformer coupling. a modulator/demodulator is used in the analog signal path and is driven by an integral chopper/power driver. it is impossible to avoid some stray capacitance between the driver circuitry and the output. careful design and layout of the MP227A has reduced the resulting output noise spikes caused by stray capacitance to 1 mv p-p, which is 0.01% relative to 10v fs, when measured over a bandwidth of 1 mhz. the noise spikes repeat at 20 khz, or twice the nominal 10 khz frequency of the MP227A chopper driver. if the output filter time constant (r1 c1) is equal to 0.5 ?, then: fc = 1/2 rc = 333 khz this low cut-off frequency ensures that the output spikes, over an effective bandwidth in excess of 1 mhz, are attenuated 10 db or more, which is enough to re- duce this error source from .01% to a negligible level. using the MP227A offset adjustment provision is made for offset adjustment on the MP227A precision isolation amplifier by connecting a 25k or 50 k (100 ppm/? or better) multi-turn poten- tiometer (r2) with a 1 m resistor as shown in figure 5. to adjust, momentarily short input hi, input lo, and az rtn to the output ana rtn and set the offset potentiometer for zero output at the output terminal. setting the MP227A gain the gain of the MP227A may be set to any value from 10 to 1000 by connecting an external resistor (rg) be- tween the gain and input lo terminals as shown in figure 5. 10.375 x 10 3 gain = rg an rn55e or better resistor is recommended for tem- perature stability. untrimmed, the absolute gain will be within +2% and ?% of the calculated value. gain trimming the gain may be deliberately fine-trimmed, if desired, by connecting a 500 (100 ppm/? or better) poten- tiometer (r1) between the feedback and output terminals as shown in figure 5. r1 compensates for the tolerance of rg plus the unit-to-unit gain variability (3%) between multiple MP227As. this also allows standardization of the outputs of multiple MP227As to a common full-scale range. for volume production where cost is a factor, the trimpot may be replaced with a fixed resistor selected during final testing. +15 offset (r2) 50 k ?5 1 meg 500 w gain (r1) supply ground signal output +15v in ?5v in a5 feed- back output ana rtn pwr gnd a6 guard input hi l2 +v out ? out input lo auto zero return gain r g twisted shielded pair see note 2 MP227A * *see text for discussion of open circuit detection note 1: positions of terminal points shown above have been rearranged for simplicity. refer to label diagram for actual positions. note 2: for proper guarding, the guard (shield) should be connected to the signal source common or input lo at the module. figure 5. typical external connections ?MP227A
auto-zero return the signal that is amplified by the MP227A is actually the difference between the input lo and the auto- zero (az) voltages. for normal operation, tie the az terminal directly to the input lo terminals. in some applications, it may be convenient to offset the input deliberately by an amount that exceeds the range of the offset trimpot (for example, to obtain expanded scale operation or to cancel out the initial or ?are?out- put of a load cell). to do this, connect the az terminal to a source of voltage equal to the desired offset, with noise performance and stability at least as good as the signal source. observe that both the input hi signal and the az sig- nal (if any) are measured with respect to the input lo terminal. for best linearity, each signal must be within 1v of input lo. open input indication the user-accessible isolated power supply voltages make it possible to use a simple open input indication network. connect a resistor on the order of 180 m to the input hi and either the +4v or ?v isolated power output terminal. this network produces a bleeder cur- rent of approximately 20 na through the input source circuitry. if the source should open, this bleeder current will drive the MP227A output into a saturated state. the speed of this response is a function of the MP227A gain setting and input time constant. multiplexing MP227As the outputs of multiple MP227As may be multiplexed to a common analog line as indicated in figure 4. a single rc filter ahead of each mux is suggested. ordering guide specify MP227A 2.76? (70 mm) 2.40? (61.0 mm) .18? (4.5 mm) .20? (5.1 mm) input hi l2 az rtn gain input lo ? out +v out guard +15v ?5v ana rtn (case) feedback a5 a6 output pwr gnd 0.1? (2.5 mm) 1.18? (30 mm) MP227A (top view) 0.025? dia pins (gold plated) .15? minimum (3.8 mm) 5.1? (13.0 mm) figure 6. MP227A mechanical & pinout.
|
