opamp differential amplifier high frequency/loading effect

Too low will load the input signal and ultimately load the opamp output, and waste power.

Too high will add Johnson noise, input bias current noise, input offset from bias current, and can add poles to the closed-loop frequency response and make the mess oscillate.

In a range of situations, the range of resistors might span many orders of magnitude.

If you need precision or high common-mode rejection, you are further constrained by the resistor values you can get.

This is, incidentally, a rather interesting lopsided circuit. The inverting input doesn't affect the ni input, but the ni input affects the inverting. So the concept of input impedance is sort of tricky.

John

Depends on how fancy the opamp is and the source impedance and the accuracy requirements. If the source impadance is high (< a few 100 ohms) it is generally better to use a three opamp instrumentation amp with the front two as high impedance buffers to the differential amp.

If you only have one opamp to work with use highish resistor values in the

100K range (say 499K and 49.9K max). The amp impedance looking onto the inputs will be at least 50K. If thats too high you need the three opamp design or a true blue instrumtation amp.

If the resistor values go too high you could get oscillation. Even with the

499K you could be on the edge and you should put a 47pF or so cap across them to reduce the gain at high frequencies (above where your're interested)

Hi, I am trying to use just a single opamp. Is there some reference where i can read about the effect of the feedback and load on the gain/phase of the closed loop amplifier? Is it that as long as the opamp can fan out the load, the gain and phase is not affected? Regards, kristo

You could just download LTspice, free from

, and run some simulations with stepped values for feedback and load resistors, and look at the gain/phase plots.

- Tom Gootee

Some simple rules of thumb to get you started:

You say R2/R1=R4/R3=10. It's usual to have R2 = R4 and R1 = R3, too. Makes the analysis easier.

Make the lower valued resistors (R1 and R3) *at least* 10 times the resistance of the source they're measuring across, ideally 100 times, and at least 10 kilohms in any case. This reduces their loading effect on the source (Kirchoff's Law - you need to make the current used for measurement insignificant, so the thing under test isn't affected).

If the feedback resistors R2 and R4 are surface mount, and are higher than, let us say 2.2M, you are likely to begin seeing problems with leakage currents on the PCB (from dirt - these effects are more usual with small surface mount components, simply because the terminals are closer together) unless you have a little experience with this kind of circuit.

--
Nemo