Charge injection in chopper-stabilized opamps.

Rather than hi-jacking the thread "Precision op amps - is 1 microvolt offset believable?", I decided to open a new topic.

After reading the thread above, I remembered that I had an OPA735 (bought brand new from Digikey) and decided to play with it a little. You know, I had nothing better to do on a dreary Sunday afternoon.

I built a classic diff amp, with 1Meg and 100k (0.1%) feedback and input resistors, supplied it with 5 volts, and biased it with a 2.5 volt virtual ground from another g.p. opamp. Proper decoupling caps used throughout. My intention was to amplify a 20 mV DC signal 10 times. Perhaps to use in the future as a current shunt preamp.

My initial verification was to check the offset voltage with inputs shorted. To my surprise, the output was reading 15mV (with a DMM), which meant 1.5 mV on a chopper opamp? Impossible!
I know enough to know when seeing such offsets in a DMM, is good practice to actually see the waveform on a scope, to see if noise, hum or an oscillation was causing the spurious reading.

The scope waveforms showed a quite significant spike, meaning that most likely the problem was being caused by input charge injection.

Reading different chopper opamp datasheets, in addition to the OPA735, there were recurring descriptions to lower the overall loop resistance to minimize charge injection.

For that matter, I replaced the feedback and input resistors to 10k and 1k (0.1%) respectively.
The output offset voltage now dropped to 0.1 mV, which was what I would expect on this opamp.

While I was expecting some chopper induced noise, I was surprised that high impedance resistors would generate some much DC offset......

Has anyone had similar experiences?

I had TI in house talking about this stuff a few months back and they think you should still be careful but that modern parts like OPA189 have almost solved these problems. The 189 spec says:

Charge injection from the integrated switches on the inputs can introduce short transients in the
input bias current of the amplifier. The extremely short duration of these pulses prevents the pulses from
amplifying, however the pulses may be coupled to the output of the amplifier through the feedback network. The
most effective method to prevent transients in the input bias current from producing additional noise at the
amplifier output is to use a low-pass filter such as an RC network.

I think I recall them saying that their app people tried 100k input impedance and couldn't detect problems (implying that with older parts they wouldn't have gone that high).

The 189 is also impressive for its bandwidth of 14Mhz for a chopper. On the other hand other types of opamps keep improving too such as the bipolar OPA2210 (5uV typical, 18Mhz) with 4nA max at 85C (competitive with Jfet and CMOS at that temp) which I quite like.

Indeed, another solution advocated in many data sheets is to use a capacitor in parallel with the feedback resistors to create a low pass filter.

Haven't attempted it yet, though.