OPA2192: Request for additional information regarding THD+N for OPA2192 and OPA2141

Hi Kai,

Thanks for the response and please excuse me for being a little bit vague with the details - I am somewhat limited as to what I can post in a public forum.

Essentially I have two inputs, across which appears a small desired differential signal (tens of mV RMS) at a particular frequency f0. A much larger amplitude common mode signal (a few volts RMS) at a higher frequency f1 is also present. The two inputs are buffered by the OPA2192 (or OPA2141) and drive the inputs of an instrumentation amplifier (so a very high impedance load). The output of the instrumentation amplifier is digitized and a synchronous detector is used to reject f1. The interfering signal f1 is only present under a specific test condition, so I am able to see how much of a shift it causes in the demodulated output (i.e. relative to the demodulated output that is present when f1 is not). The output of the demodulator that I am interested in is the RMS sum of the I and Q components.

With the OPA2141, the shift is small and tolerable. With the OPA2192 it is larger and not tolerable. I know that the demodulator adequately rejects f1, therefore I assume a third frequency must be present. I have verified that there is no saturation taking place anywhere in the signal chain, I am not operating the OPA2192 in its crossover region, and both parts are operated within the specified typical value for slew rate (which is the same for both; also note that there is no guaranteed minimum for either part).

The only difference I see on the datasheets that might be relevant is the higher level of THD+N of the OPA2192. I am hypothesizing that this is resulting in a higher amplitude (relative to when the OPA2141 is used) of some third frequency (maybe an intermodulation product) that is not as well rejected by the demodulator. I can't prove this yet, but to help my investigation, I am looking for any additional information TI might have regarding THD+N for these parts (see questions in original post).

By the way, in your response, did you mean to say "... which makes the distortion increase by a factor of ten when increasing the frequency by a factor of ten"?

Thanks

Shane

No kai was right on gain as well - and you are right on frequency - both are related to LG degrading which is what controls HD (partially) -that breaks down as you start approaching slew limiting, in fact I have a rule of thumb about that, but in the meantime a very thorough discussion of HD in the context of a really good measurement methodology we developed to get down into the -140dBc area - 

Xavier ultra low distortion article.pdf

also Shane, have you considered filtering the higher frequency F1 in the buffer amps so they do not have to try and follow it? 

Thanks for the replies. f1 has already been filtered to the extent possible given other application specific constraints.

Hi Shane,

have you thought about using the OPA837 or OPA625? These OPAmps have very low distortion specs at 100kHz.

By the way, intermodulation can be identified by the appearance of sum and difference frequencies.

Kai

Hi Kai,

Thanks for the suggestions. Unfortunately both of those parts would saturate due to their limited supply voltage range relative to the OPA2141 and OPA2192.

It looks like the additional data I was hoping to see probably doesn't exist so I am going to close the thread. I will mark your first reply as the resolution.

Shane

Hi Shane,

I would suggest that the low pass filtering in order to suppress f1, recommended by Michael, is entirely carried out by the help of passive filters, or by other words, before the signal runs through any OPAmp. This should reduce the intermodulation distortion to a minimum.

Kai