1. Home
  2. Electronics Design
  3. Gain Non-linearity Measurement Woes

Gain Non-linearity Measurement Woes

Hello! I've recently been interested in measuring the gain non-linearity of various op-amps and composite op-amp configurations for use in integrating ADCs. Following the circuits and measurement techniques posted in these app notes from ADI and TI, the results I got were unusable:

formatting link
problem I found with these circuits is that they are basically noise amplifiers with a high-gain network created by the top feedback resistor and the divider on the inverting input. Instead of a clear line, I get a flood of noise that comes from the resistors and op-amp (the noise waveforms are very characteristic of band-limited noise). I don't understand how the clean lines in the app notes were obtained. I tried both a small capacitor from input to the summing junction as recommended in the TI app note, as well as a film capacitor (10s of nF) across the top resistor on the divider on the inverting input. The latter fix helped remove the noise, but the XY plot was flat even for the incredibly poor LM358. I had a couple other people repeat these measurements for me, and they had similar results. Is there a sane way to measure gain non-linearity that actually works? Regards, NNNI

Reply to
NNNI
Loading thread data ...

Just signal average a bunch.

Reply to
John Larkin

...and low-pass filter the scope inputs. If a sweep frequency of around 1Hz is used, then maybe filter at about 1kHz. Both app notes are silent on this.

John

Reply to
John Walliker

Most scopes do bw limiting and signal averaging these days. One can always do some creative graphing too.

Reply to
John Larkin

I've tried both averaging and bandwidth limiting, but that didn't help much. The only scope I have on hand right now is a HP54600A. In two weeks I should get a Rigol DS1054Z, I'll try the measurements again. Thanks for the tip about low pass filtering the scope input, I'll try that as well. Regards, NNNI

Reply to
NNNI

Are you picking up ambient noise? An Altoids or a Danish Butter Cookie can would be a good shield, with some lead bypasses.

I hope you are not using one of those dreadful plastic breadboard things.

formatting link

Reply to
John Larkin

I wouldn't dream of using a solderless breadboard for anything!

formatting link
've been using Manhattan-style as of late. Regarding the noise, I'm not 100% sure if shielding will help, because at such high gains, you might as well be amplifying the noise of the resistors and op-amp.

Reply to
NNNI

Wave your hands around it and see if the noise changes. Look for 50/60 Hz too maybe. FFT it!

Nice style, similar to mine. I use gold-plated FR4, which looks good and doesn't tarnish over time.

formatting link
Here's one I just did, to blow up some Panasonic solid-state relays.

formatting link
SSR data sheets don't have SOAR graphs.

Looks like I can add a roughly 10 ohm wirewound resistor and have the SSR survive shorting 35 volts for long enough for software to shut things down. Now I need to blow up some surface-mount wirewounds.

Reply to
John Larkin

Air convection is a classic source of super-low-frequency noise, 1/f sort of thing. Here's a little cover over a thermocouple front-end, which really helps.

formatting link

Reply to
John Larkin

IIRC there's an old PMI app note about the OP27, which they tout as having vastly superior open-loop gain linearity compared with the OP07.

I remember it being full of plots and test circuits and stuff. I'll see if I can find it.

Cheers

Phil Hobbs

Reply to
Phil Hobbs

To amplify a current shunt, I used an opamp feedback network of about

10 ohms Thevenin, to keep the Johnson and current-driven noise down. That makes the opamp work hard making much swing, which creates thermal hooks. So, compound amp.

I used LT1028.

Reply to
John Larkin

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.