I'm still trying to understand real-life op-amps. I've built this circuit:


simulate this circuit – Schematic created using CircuitLab

Op-amp is an MCP6141 (datasheet) I've chosen for its low-power consumption.

\$V_{\text{in}}\$ is sourced by a waveform generator, and has the following equation:

$$V_{\text{in}} = 1.5\text{ V} + 5\text{ mV} \times \sin(2\pi \times 3.8\text{ kHz} \times t)$$

When \$R_1\$ is between 0 and \$5.9\text{ k}\Omega\$, I have a normal behaviour, signal is amplified, and gain increases. By the way, I obtain a gain of 40 dB for \$R_2 = 5.9\text{ k}\Omega\$, which is strange twice, because

  1. \$20\log(5900/100) = 35\text{ dB}\$ which is lower than 40 dB
  2. datasheet says op-amp gain at 3.8kHz is around 30 dB.

enter image description here

As we can see output signal (yellow) is slightly distorted but everything happens as expected (except the gain as explained above).

FTT shows the main frequency is amplified by 40 dB for \$R_1 = 5.9\text{ k}\Omega\$:

enter image description here

But suddenly for \$R_1>5.9\text{ k}\Omega\$, things change and output signal becomes strange:

enter image description here

enter image description here

Does somebody has any idea why there is this strong discontinuity in the regime of the op-amp when \$R_2\$ crosses \$5.9\text{ k}\Omega\$?

  • 2
    \$\begingroup\$ Try changing C1, and see if the "break-point" changes. Try several different values, and see if there is a relationship. \$\endgroup\$ – WhatRoughBeast Oct 19 '16 at 19:11
  • 1
    \$\begingroup\$ Yeah C1 looks small. Also the GBW of that opamp is 100kHz! you've got a gain of ~60 and a freq of 3.8kHz... ~220 kHz. That might be causing an issue.... what's it look like at lower frequency? \$\endgroup\$ – George Herold Oct 19 '16 at 19:30

That's a very low impedance feedback path for an ultra-low power op-amp. The output resistance is probably of a similar order to the R1 which is causing you problems (you can 'measure' it from the SPICE model, they typically don't discuss such dirty little secrets on the datasheet).

I would suggest increasing it by at least an order of magnitude, taking R2 to 1K and C1 to 10nF, preferably by two orders of magnitude (R1-> 10K and C1->1nF). Gain is specified with a 50K load to ground.

Check your calculation of the C1 value too- it should have negligible reactance at 3.8KHz compared to R2, so I get more like several uF than 100nF for R2 = 100\$\Omega\$. So maybe 10K/39nF for R2/C1 and R1 from 0 to 2M

  • \$\begingroup\$ I took R2=1kOhm and C1=10nF and it works great, I couldn't reproduce the "breakdown" effect. Could you please detail how the feedback impedance and the output resistance are related? When is it an important parameter when choosing an Op-Amp? Thanks if you can explain, because I've been struggling with op-amp recently. For C1, my goal is to have a gain of 1 for DC input, so I guess it's okay to have a small reactance at the frequency of interest. \$\endgroup\$ – Vincz777 Oct 19 '16 at 20:17
  • \$\begingroup\$ In addition, when I increase R2, gain increases and reaches a maximum value (41 dB), then decreases again. I don't understand why! \$\endgroup\$ – Vincz777 Oct 19 '16 at 20:20
  • \$\begingroup\$ If C1 is too small you'll have reduced gain at the frequency of interest (and significant phase shift). Try going up to my suggested values (you can use 100nF rather than 39nF) and see what happens. You need to worry about loading when the output resistance is of similar order to your feedback network, then C1 starts to load your output in a frequency-dependent manner. Output resistance is maybe 100 ohms for a normal op-amp that draws a mA but yours works on 600nA so much higher- try measuring it with SPICE (you can do it open-loop in SPICE and sweep a voltage source with (say) 10K series. \$\endgroup\$ – Spehro Pefhany Oct 19 '16 at 20:23
  • \$\begingroup\$ I just put 100 nF and now the gain is topped at 27 dB, whatever R2 is. This is mind boggling! \$\endgroup\$ – Vincz777 Oct 19 '16 at 20:43
  • \$\begingroup\$ Now increase R1/R2 by another 10:1. \$\endgroup\$ – Spehro Pefhany Oct 19 '16 at 21:17

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.