6
\$\begingroup\$

So, let's say I have a precision voltage reference with the output fed to a voltage divider which is tapped at two places for voltages A and B. I've made sure to use precision resistors (or calibrate) and the divider doesn't overload the voltage reference. So at this point, I have maintained line regulation up to the taps but significantly degraded load regulation there by using a voltage divider for A and B but I thought that the best thing to do now is to use an op-amp voltage follower to buffer each of A and B. The characteristics that I've identified so far are that the input impedance should be very high, the input offset voltage should be as low as possible, low input offset voltage temperature drift and the input and output ranges should match. It seems like CMRR wouldn't matter in this application. On top of that, I'm not sure what else to look at. Noise should be low but how low and relative to what exactly?

To give a couple of examples, if I use the LM4040 4.096 Grade A then I should have, at worst, 14mV of error due to temperature drift and load, right? So how do I make sure that my buffer doesn't make that worse? What about something like the REF3240 where the error might only be in the tens of μV? Isn't it the case that something like the OPA340 would work fine in the first case but might not be quite good enough (just because of the input offset voltage and its temperature drift) for the second?

\$\endgroup\$
  • \$\begingroup\$ Another buffer characteristic: ability to drive a capacitive load. Very often, a reference will include a bypass capacitor to ground, because a buffer's high-frequency performance is not-so-great. As a buffer (gain near 1), oscillation may be a problem. And CMRR, PSRR are not good at high frequency. \$\endgroup\$ – glen_geek Dec 27 '17 at 18:44
  • \$\begingroup\$ What kind of frequencies are we talking about, do you think? If it's running from a SMPS then it'd be hundreds of kHz to MHz, right? But for LDO then it can't be that much, right? \$\endgroup\$ – Anthony Dec 27 '17 at 18:55
  • 1
    \$\begingroup\$ People, this is the kind of question you ask when looking for parts, not "gimme part no for XXX" \$\endgroup\$ – PlasmaHH Dec 27 '17 at 21:49
  • \$\begingroup\$ Maybe Power supply rejection ratio. PSRR. \$\endgroup\$ – mkeith Dec 28 '17 at 6:36
6
\$\begingroup\$

To summarize, you want to know what opamp characteristics you need to buffer a reference voltage.

The two most obvious are low offset voltage and unit-gain stability. Any offset voltage is a error directly added to the output. Without unity-gain stability, the thing will oscillate in voltage follower (unity gain) mode.

Other parameters depend on more specific circumstances. For example, will this thing be subjected to wide temperature swings. If so, low offset across a wide temperature range is important, not just offset at 25 °C or whatever.

Other issues come up depending on what supply voltages you have available relative to the reference voltage you want to buffer, and how stable those supplies are.

Do you expect the load to have a significant capacitive component? If so, you have to look at stability more carefully.

In any case, it would be a good idea to filter both supply inputs with a ferrite chip inductor or two followed by caps to ground. That way you don't have to count as much on the active power supply rejection capability at high frequencies.

\$\endgroup\$
  • \$\begingroup\$ Thanks. Where can I find out more about filtering the supply inputs? I'm only familiar with using bypass caps. \$\endgroup\$ – Anthony Dec 29 '17 at 14:53
  • \$\begingroup\$ @Ant: I don't know if there's a specific resource. It's just a basic application of a low pass filter, although one that has to pass the supply current without causing too much of a voltage drop. \$\endgroup\$ – Olin Lathrop Dec 29 '17 at 14:55
0
\$\begingroup\$

If the VREF is used in a charge-cancelling Analog Digital Converter, the ADC will be grabbing big hunks of charge during every conversion, and your opAmp must quickly recover from the output-transient-induced voltage sagging and perhaps even ringing.

\$\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.