THS4521-HT: Op-Amp giving negative differential voltage

Question

Can someone help to explain where I get wrong with this schematics? I use a differential op-amp THS4521 and want to measure the differential signal coming from a pressure sensor (transducer).

The sensor is powered from 5V and it has a range of 0-1000bar. Without connecting the op-amp at 0 bar it gives differential voltage (between P+ and P-) of 0.14mV and 1000bar it is 14mV.

Therefore I used this op-amp to amplify the voltage and measure it with an ADC.

The ADC has no internal gain, therefore it must be like this.

Problem is that when the sensor is connected to the op-amp it gets an offset of ca 3mV.

After reaching 25bar the values on the output of op-amp go below 0mV. Since I cannot measure negative differential voltage with the ADC, I have a problem.

I could swap P+ and P- but then I cannot read anything below 25bar.

Sensor is powered from 5V source, the op-amp from 3.3V. Rest is on the attached picture.

Does anyone has experience with this op-amp and could tell me some hints?

I hae got some inputs like apply some voltage to Vocm to get rid of the offset. I tried to apply 2.5V as I had a reference voltage available near the op-amp. I pulled Vocm up with 10k resistor. Unfortunately, the circuit still does not function as it should. As a result, offset is still there, only it occurs in different place, at 37 bar. Vocm, instead 2.5V is going down to 2.4226.

Sensor input (voltage between P+ and P-) is however not going negative, as it was before. So I observe a strange behaviour that the input voltage is rising with rising pressure but output is going negative above 37bar. In the end of testing I have disconnected the 2.5V, leaving Vocm floating, so similar to initial case, only without two 100nF capacitors coupling to ground. Then the voltage on that pin was 1.72V, almost the same as on output ADP+ and ADP-.

Was the 10k pull-up resistor too big? Was the applied voltage too high? I am out of ideas so far. Does anyone here have a suggestion what can I try next?

Thanks in advance.

Comment:

I cant change the Op-amp right now but i am considering it to change in the next design.I am thinking INA128 Op-Amp from texas would the job since we need high temp component. Is there any way to improve with existing design.

Answer 1

The DC input offset voltage on the THS4521 is typically +/-0.2 mV and can be as high as +/-2 mV at ambient temperatures. Across the whole range it can be as high as +/-4 mV.

Given that you appear to have a gain of 20 this could result in an output offset of anything in the range +/-80 mV.

Maybe this is the problem? It sounds to me, on the face of it, that this device isn't suitable as a pressure transducer amplifier. I mean, it has a bandwidth of 145 MHz and, with a big bandwidth things are sacrificed in other areas (such as DC accuracy).

Try using an instrumentation amplifier - there are plenty to choose from and I doubt very much (even under really strange measurement situations) if you need a bandwidth more than 50 kHz. There are plenty of IAs that are sub 100 uV input offset across a full temperature range. I understand you need a differential output of course and this can be accommodated by an inverting op-amp on the output of the IA.

Answer 2

Either use a ADC with differential input or at least use two channels ADC and sample them synchronously, then subtract the two results, like poor man differential ADC.

Please note that this opamp is not suitable for bridge transducer, IMO. You would need an instrumentation amplifier for this task, that has very big input impedance. In your case the sensor itself has the influence on the gain and its variation. Having 20k and 1k resistors, it would make a gain of G=20, but in your case it is not gonna happen.

Have a look on AD8237, MAX4208/MAX4209 and similar new gen. of instrumentation amplifiers with indirect current feedback and bias subtraction.

Answer 3

Op-Amp giving negative differential voltage

You have your ADP+ signal connected to the op-amp's VOUT-, and ADP- connected to VOUT+. So you should expect an overall negative differential gain between the input voltage and output voltage.

An increase in the voltage supplied by the sensor will cause a decrease in the differential output voltage \$V({\rm ADP_+}) - V({\rm ADP_-})\$.

I hae got some inputs like apply some voltage to Vocm to get rid of the offset.

If the op-amp is working correctly, applying a voltage to VOCM will only change the output common-mode voltage, not the differential output.

Does anyone here have a suggestion what can I try next?

As others have said, you can use a differential-input ADC to be able to read both positive and negative differential inputs.

Another option is use an in-amp rather than your differential op-amp to generate a single-ended output that you can read with a single-ended ADC.

Answer 4

1K input R's will cause a gain error.

Your OA has high input offset and differential output. I think you only need single ended with an INA with Vio<50uV and Zin>=10 M as most DPS sensors are rated for load >=1 M with Zout being a few hundred ohms, thus 1K will cause a gain error.

Make sure that you use shielded twisted pair cable to improve noise immunity or a balun CM choke.

It is also possible in App Notes to null the offset at 0 bar.

Any of these INA's should work.

Pick the best one for you.

These are all low offset and 1% gain tolerance, work on single supply 3.3V but this can also be tuned with a single gain control R.

Then you may increase the gain to 250 or so.