2
\$\begingroup\$

I realize that there are a lot of questions relating to op amp offset here on electronics.se; however, none of them seemed to answer my exact question. If some of them are too similar, feel free to close.

I have a very simple non-inverting single-supply 3.3v-0v LTC6240 circut. This op amp doesn't have an offset null pin. I am running this op amp at a very high gain (>1000) with a 40khz input signal. The circuit works perfectly fine most of the time; however, I have an unpredictable ~500uV offset on the negative input, which (with some op-amps and some populated boards) works out to an output offset that the rest of my circuit can't compensate for.

I thought the best solution was this. I put a 10uf cap in series to ground as the answer describes and somehow, my gain at 40khz went away. I've replicated this in ltspice, even though I don't understand why it happens; the impedance of a 10uf cap at 40khz is only ~0.5 ohm. Here's a schematic of how I connected it:

enter image description here

As for putting a potentiometer in to correct the offset as described here, unfortunately I don't have a negative voltage available on the board, so all the pot can do is raise the offset.

\$\endgroup\$
7
  • \$\begingroup\$ You should probably add a circuit to show how you're connecting things. I have a feeling that the reason why the cap at 40khz crushed your gain is because you just high-pass filtered your output at 40k. It's also possible you loaded the op-amp output down too much by creating a virtual ground at 40khz. \$\endgroup\$
    – horta
    Commented Apr 9, 2016 at 17:06
  • \$\begingroup\$ @horta One schematic coming right up. \$\endgroup\$
    – 0xDBFB7
    Commented Apr 9, 2016 at 17:11
  • \$\begingroup\$ @horta there we go. \$\endgroup\$
    – 0xDBFB7
    Commented Apr 9, 2016 at 17:42
  • \$\begingroup\$ Is it really 7mOhm? Because if you look at the basic non-inverting amp equation, that results in a gain of (0.007/500+1)*V+ ~= V+ or a gain of 1. \$\endgroup\$
    – horta
    Commented Apr 9, 2016 at 17:50
  • \$\begingroup\$ m = milli = \$10^{-3}\$. M = mega = \$10^6\$. Capitisation matters. Also "F" for farad and "Hz" for hertz. General rule for SI units is capital letter if abbreviation is used but lower case if spelt out. e.g., V for volt, A for ampere, K for kelvin. \$\endgroup\$
    – Transistor
    Commented Apr 9, 2016 at 17:54

1 Answer 1

4
\$\begingroup\$

You need to use meg instead of m for the resistor value. You have specified a 0.007\$\Omega\$ resistor, which will tend to reduce the gain to 1.

Your amplifier has a GBW product of 18MHz meaning it will only have an open-loop gain of about 500 at 40kHz, not the 14,000 you appear to be expecting. However it will have extremely high gain at DC so any offset will be amplified - so only a few uV will saturate the amplifier.

You also need to bias the amplifier so the inputs and outputs are somewhere in the operating range- right now it will sit at the negative rail- you want it to be about Vdd/2.

It's better to have a few AC-coupled stages with gain of ~100 or less rather than one stage which needs astronomical gain.

\$\endgroup\$
1
  • \$\begingroup\$ Aha! Biasing the input! That's perfect. \$\endgroup\$
    – 0xDBFB7
    Commented Apr 9, 2016 at 18:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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