Let’s re-visit the voltage meter for a moment - the actual circuit used in the next test differs in a few resistor values, but it’s still essentially the same as the triple op-amp setup described earlier. Here are the two main op-amps involved (omitting the 3rd one, used for the reference voltage):
Let’s inject a 10 KHz square wave into this circuit, and see how it behaves:
The input signal is shown at the top, in purple. It swings from +5V to -5V, and since R1 is only 1 MΩ in the actual setup, the voltage divider is 2:1, not 10:1.
The blue trace at the bottom is the output of U1A. In the test setup, R5 was shorted out and R6 was absent, making U1A a follower, so the output is ±2.5V.
Lastly, the yellow trace is what is fed into the ADC, i.e. the output of U2B. It varies between around +1.3V and +2.3V, nicely within the ADC allowable 0..3.3V range.
As you can see, the LT1413 output is far from a square wave - did we go wrong somewhere?
Yes and no: if the input signals we’re going to measure only interests us in the “static” DC range and in the low audio range anyway, then this setup is actually fine. We can easily measure any fixed DC voltage with this, as well as mains A/C at 50 and 60 Hz.
But there clearly is something very unexpected going on, in terms of using an op-amp as simple amplifier or voltage follower (i.e. amplification set to 1). Where did the “distortion” come from?
Here are the main specs, as presented on the first page of the LT1413 datasheet:
All very respectable figures. Nothing unusual here. Let’s look a little further:
Aha! Check out the “Slew Rate” specifications: what the LT1413 datasheet is saying, is that this particular op-amp’s output level can typically change by no more than 0.3V per microsecond.
In this circuit, the input signal swings between +5V and -5V, i.e. a 10V
change - which in the case of a square wave happens very quickly of course.
At the junction of R1 (1 MΩ) and R2, i.e. V+ of U1A, and at the output, the
signal swings between +2.5V and -2.5V, i.e. 5V in either direction.
Surprise! The LT1413 is behaving exactly as specified: 5V / 0.3V/µs = 16.7 µs rise & fall times.
That doesn’t make the LT1413 a “bad” op-amp in any way. In fact, it has pretty impressive specs for several of its parameters. But in this circuit, if we want to handle higher frequencies, we’re going to need an op-amp with better slew rate specs.
(The slew rate is in fact often intentionally limited by the manufacturer, to avoid oscillation in voltage-follower mode, but that’s a whole different story, involving phase shifts and “poles”…)
This is typical when working with op-amps: they all make lots of choices and trade-offs, so you really have to select and match the op-amp for every single use case. In the next article we’ll go through a brief summary of the main parameters of op-amps, what they mean, and then try to pick a better one for our voltmeter setup.
For the true path to insight and wisdom, check out The Art of Electronics book: over 1200 pages filled with facts and experience to help the Electronics Engineer, including several tables with op-amps, their key criteria, and how they compare.