0
\$\begingroup\$

Background

I am designing a low noise gain section for a power amplifier. This section consists of 2 parts:

  • the input section, taking care of gain
  • a single-ended to balanced buffer with unity gain (doubling with added inverter)

After extensive simulation and tweaking I managed to get to very low simulated noise output figures of -115,1dBu (20-20kHz) for the input/gain section and -117.6dBu for the differential converter/buffer.

Known facts

  • I am aware that noise in opamp configurations can sometimes have a larger than signal gain (but I can't see how this is related to this problem, nevertheless I like to mention it).
  • I am aware that noise can be added with the rule: Total noise = SQRT(N1^2 + N2^2)
  • I am aware that the above rule is only valid if noise sources are uncorrelated (as in originated from different parts for example).

Assumptions and question

  • As far as I see it, input gain section and balanced buffer stage are uncorrelated.
  • The balanced buffer stage doubles voltage. So if I tie this to the output of the buffer stage, i should be able to add noise by adding twice the input/gain stage noise plus the balanced buffer stage.

If I do this, here are the figures:

  • gain stage simulated noise output is 1.37µVRMS (-115dBu)
  • balanced buffer stage (differential) simulated noise output is 1.02µVRMS (-117.6 dBu)
  • adding these as described leads to SQRT((2*1.37µV)^2 + 1.02µV^2) = 2.92 µVRMS (-108.5 dBu)

If I simulate both sections cascaded, however, I get a different noise output (20-20kHz integrated) of 3.16µVRMS or -107.8dBu. That's a 0.7dB difference.

Am I making some wrong assumptions, should i check some overlooked details or is this an effect of other effects that I don't understand or know about? I basically want to understand why the cascaded simulated result is different from the calculated result.

Edit

first stage circuit: enter image description here

second stage circuit: enter image description here

\$\endgroup\$
5
  • \$\begingroup\$ What's the output impedance of the 1st stage? What's the input impedance of the 2nd stage? \$\endgroup\$
    – Andy aka
    Commented Dec 21, 2020 at 15:47
  • \$\begingroup\$ If you add 2x the output stage noise (once for each leg) it comes out pretty close. Might that be what's happening? \$\endgroup\$
    – user16324
    Commented Dec 21, 2020 at 15:47
  • \$\begingroup\$ output impedance of first stage is an opamp output, so near zero, input impedance of second stage is an opamp unity buffer input, so near infinite. \$\endgroup\$
    – gommer
    Commented Dec 21, 2020 at 15:53
  • \$\begingroup\$ "I am aware that noise in opamp configurations can sometimes have a larger than signal gain (but I can't see how this is related to this problem, nevertheless I like to mention it)." Could we see your opamp configurations? \$\endgroup\$
    – Math Keeps Me Busy
    Commented Dec 21, 2020 at 15:54
  • \$\begingroup\$ output noise of second stage is already measured across the differential output. Both when simulating independently and cascaded \$\endgroup\$
    – gommer
    Commented Dec 21, 2020 at 15:55

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.