# Characterizing raw op amp gain

I thought as a first exercise in op amps, even before setting up a feedback circuit, I would determine the gain of the raw op amp.

I tried with an LM324 and an LM358 which should be essentially the same. I assume the input currents are really small, I believe when I glanced at the data sheet it was nA. So I used a 4 AA power supply with 6v (not fresh batteries). To get an input signal, I created a voltage divider from a 1M and 1K resistor (V0). I can measure .0056mV with the meter. Then I put a second divider of 1M and 1k across the 1k (labelled V1), which should give roughly 5.5 uV though I can't measure that.

For the LM324, pin 2 is V-, pin 3 is V+, and pin 1 is Vout. If I set pin 2 and 3 to 0V, essentially 0 comes out (some small noise as expected). If I set pin3 to Vcc, 4.8v comes out (roughly 1.2v drop through the amp).

But when I set pin 3 to the first stage voltage divider, 5.6mV, the output voltage does max out, but I notice the input voltage does change a bit. That one I didn't record, see below for more. Troubling, considering there should be essentially no effect from the op amp. Worse, once I put in the 5.5uA signal, I expected to see something like a gain of 850,000 or 900,000. I swear I did this before, and had planned a lab on it. But now I can't get this repeatably. When I tried again, this time replacing R3 and R4 by 100k each, V1 changed from 2.9mV to 3.2mV I don't see how that could possibly happen.

I tried again with positive feedback mode, I can get a gain of 101 using a 100k and 1k resistor. I used 1M and 1k, and then across the 1k, a second voltage divider with two 100k. Again, voltage V1 is 2.9mV but when plugged into the + side of the op amp it changes, only to 3.2mV, but I wouldn't have thought there would be any change at all.

I'd like to know whether I am simply using the op amp out of the range in which it behaves well, whether it should work in which case I will try to double check my power supply, my wiring. The voltage is still 6v, and the LM324N should be good down to 3V, so I can't believe it's the power. This whole experience is making me doubt my sanity.

• Please post a schematic of what you did. It will make it much easier for us to understand. Commented Oct 12, 2012 at 22:07
• Note LM324 has a typical offset voltage of 2 mV. Imagine this as a tiny voltage source in series with one or the other of the input pins. Imagine its value varies slowly over time, mostly due to temperature changes in the chip. Commented Oct 12, 2012 at 23:02

You are trying to use the "opamp out of the range it behaves well". Used open loop like this it becomes a makeshift comparator.

Opamps are meant to always be used closed loop. The open loop gain is pretty variable so cannot be relied upon, the only real requirement is that it's as large as possible (ideally infinite) When you close the loop though the huge gain makes the (closed loop) gain very accurate and the open loop variability (usually) makes almost no difference.

In theory, with an ideal opamp what you are trying would work okay, but in real opamps you have slight mismatches, causing things like input offset voltage which just by itself will cause this experiment to fail.

If we look at page 3 of the datasheet, we see the typical offset voltage for the LM324 is 2mV, up to a maximum of 7mV (3mV for the LM324A). This swamps your tiny input signal, and will cause the output to do just about anything, probably saturate.
There is also an offset current, external noise, and other non-ideal parameters that are variable with temperature that will cause issues here.
For an example of how variable the open loop gain is, here is a graph of the open loop gain over supply voltage:

We can see at 5V, the open loop gain is around 100dB (100,000) compared to 120dB (1,000,000) at 15V. This is a factor of 10 difference. Also we can see it varies considerably with load.

You really need to read the datasheet. Every opamp is different and has its own wrinkles. You can't use a opamp without knowing what that particular part can and can't do.

The LM324 is specified for 3 mV offset voltage. When you gave it 5.6 mV offset, it pegged to one end just as expected. When you gave it just a few µV from the second divider, the difference between the inputs was well under the maximum offset voltage, so the result is unpredictable and can vary from part to part, accross temperature, supply voltage, phase of the moon, etc. Again you got what was expected since any output voltage would have been valid.

Op-amps and in fact most circuits are (to state the obvious) meant to be operated in the mode that they were to be designed for. Amongst the various possible violations:

• input range - inverting input is tied to ground -turns out this device can handle, but it can't handle being tied to the other rail.

• It's meant to be operated with feedback, ie. the bias and operation presumes a current flow through the differential input. Tying one input to ground and slewing the other input means that the internal circuitry is running way out of spec. or signal flow is corrupted.

to test this properly you must bias one input mid rail and then bias the other input to the same point and then change the bias slightly to see the open loop gain. It would be best to inject a signal so you can sweep with frequency. By doing this you keep the op-amp "close " to operating the way it was intended (i.e. a virtual ground).

There will be certain op-amp designs that can handle this and others that can't.

What you've really gained is the insight that not all op-amps can be operated as comparators and therefore you can conclude that op-amps and comparators designs are very different internally.

I think you have to trim the input offset 1st and beware of the load resistor and V+ effects on gain.

Reference Analog Devices

Earlier comments are correct that op-amps are intended to be used withing specific well defined ranges of feedback. This does not mean that they will not work if used outside those well known parameters...just that you must experiment and define your own set of operating parameters for your specific device if using in ways not anticipated by the original designer. For those who can live with resultant low bandwidth in exchange for higher gain, it may be permissible to run op-amps with no negative feedback. This is a situation where traditionalists will caution against working outside traditional specifications, but where doing so may yield interesting and useful results.