I want to build a programmable gain instrumentation amplifier with power of two possible gains (1x, 2x, 4x, 8x...). Is there a way to use a R-2R resistor ladder and an analog multiplexer in order to tune the $R_{gain}$ ? Classic instrumentation amplifier

On the web I found this way to make programmable gain amplifiers using a R-2R ladder and an analog multiplexer but I don't want to attenuate my input signal as it is already small, I'd lose it in the noise in this way. R-2R PGA

  • \$\begingroup\$ You can consider using the R-2R ladder in the NFB path (RF and RG) in the circuit above. An MDAC (AD7528 etc) makes a convenient R-2R ladder... \$\endgroup\$ – Brian Drummond Jan 4 '16 at 12:07

You could instead buy an AD8231: -

enter image description here

Digitally/pin-programmable gain of 1, 2, 4, 8, 16, 32, 64, or 128 plus you get a spare op-amp.

Or you could use a PGA281: -

enter image description here

Or maybe consider using a digipot with one of these: -

enter image description here

Trying to control gain by adjusting the resistors where you have shown in your question is going to be problematic especially if you input signal is very small to begin with. Factors that rule this out include: -

  • Injection of extra noise
  • Injection of bias currents from analogue gates
  • Inaccuracies / non-linearities of analogue gates
  • \$\begingroup\$ @KBowser I was wondering if this solved your problem? \$\endgroup\$ – Andy aka Feb 1 '17 at 8:31

Don't use an R-2R ladder. The classic way to switch the gain of an instrumentation amplifier is to change the value of Rgain (which is why it's called Rgain by the way!).

There are some subtleties due to the distortion of an analogue multiplexer however, its variation in series resistance with voltage. Simply using a mux to select various values in the place of Rgain puts the non-linear switch resistance in series with Rgain. For 5v/10v 4060 type switches, this is a delta of a few ohms over its voltage range, which may be neglgible for your application compared to a few kohms.

If it's not negligible, then abetter way to do it would be to use a dual mux. Consider the string of resistors R1-Rgain-R1 as a long string of small resistors, with each mux selecting an appropriate tap point to feed back to the first amplifiers. This way, the switch resistance delta is compared with the amplifier input resistance, which is huge, rather than the resistors making up Rgain, which are much smaller.

A way to 'cheat' and only use one mux would be to note that the third amplifier has a differential input as well, so does not need the first stage to be fully balanced, though having it fully balanced maximises its common mode range and rejection. Use a single mux to feed the top amplifier from tap points along R1-Rgain string, and feed the lower amplifier from the Rgain-R1 connection. You would need to do your own calcuations for first stage gain with this arrangement, as most publications do not expect you to use this unbalanced arrangement. But it works, and works well, combined with the second stage has a fully differential behaviour, and only sacrifices some input common mode range.


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.