2
\$\begingroup\$

I want to measure micro volt signal from 10 micro volt to 1000 micro volt. for this i have used ADA4528 IC. And used three Op amp in series, first as voltage follower, second as amplifier of gain 10, and third as amplifier having gain 100. Is it okay combination for micro volt measurement? Or instrumentation amplifier will be good for micro volt measurement?

\$\endgroup\$
  • 1
    \$\begingroup\$ What's the source impedance of the sensor? How much bandwidth do you need? Besides what Andy said about noise, for maximum BW (and the same opamps) you'll want to split the gain equally. (BTW you've got a gain of 1000 which will raise 10 uV to 10 mV.) \$\endgroup\$ – George Herold Dec 16 '16 at 14:21
4
\$\begingroup\$

The first stage should also apply as much gain as is possible to reduce the impact of self-generated noises in the op-amps. For instance, the ADA4528 has a noise figure of 5.6 nV/\$\sqrt{Hz}\$ and if your 2nd stage also was a 4528 then the noises would add to give about 8 nV/\$\sqrt{Hz}\$. This is when there is no gain in the first stage.

However, if your first stage had a gain of (say) ten then the equivalent noise at the input to the 2nd stage is 10x 5.63 nV/\$\sqrt{Hz}\$.

Noises add as RMS terms do hence they add using \$\sqrt{A^2+B^2}\$

So, get as much amplification into the front-end as you can in order to reduce noise effects.

The ADA4528 is a good choice but beware, at 200 kHz the noise (due to internal switching circuits) is significantly higher. If your bandwidth is much smaller than 200 kHz then you'll be OK.

\$\endgroup\$
  • \$\begingroup\$ Andy sir, is it will be better to use instrumentation amplifier ? \$\endgroup\$ – Satish Patil Dec 16 '16 at 9:39
  • \$\begingroup\$ @SatishPatil Sometimes but not always. An IA uses op-amps but you won't find an IA containing op-amps as good as the ADA4528 but this, as always, is application dependent. \$\endgroup\$ – Andy aka Dec 16 '16 at 10:05
3
\$\begingroup\$

You should have more gain in the front end (first) amplifier. I would suggest 100, since that is the gain that the noise is specified (but not guaranteed) at. There may be output noise in addition to the input noise that will reduce performance at G=1. As Andy rightfully points out, the uncorrelated portion of noise will add in quadrature.

Zero-drift op-amps are kind of a can of worms- without more information on your application I won't go into this further but you should read more about how they work and how the inputs behave with regard to EMI and their internally generated switching noise. Most likely it will be just fine, but it's better safe than sorry.

You only need an instrumentation amp if the signal is differential and both inputs have to be high impedance. Otherwise it will be inferior performance (bought ones anyway) and more expensive/complex.

You didn't mention what you are shooting for in terms of accuracy but to get down to less than a few uV of drift will require some care in terms of reducing the effect of thermal EMFs, such as keeping air currents off the PCB. In extreme cases, low thermal EMF solder is sometimes used, but it is expensive and probably not RoHS compatible.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.