1
\$\begingroup\$

I'm using this simple circuit to amplify a small signal:

schematic

simulate this circuit – Schematic created using CircuitLab

Input voltage is defined by: Vin = 1.5V + 5mV*sin(wt) with w = 5 kHz / (2*Pi).

The op-amp I use is the MCP6142 (application requires very low current consumption). I use the second op-amp only to probe the output signal from first op-amp (as scope impedance - 1 MOhm - is much higher than the resistances I use).

The problem I have is that some extra spectral components are added to the output signal:

enter image description here

Here is a capture of the input and output signals over time:

enter image description here

We can see that the input signal - blue - is "pure" as it is generated by the waveform generator of the scope. However the output signal - yellow -, though it is correctly amplified, has some extra spectral components, whose frequencies are multiples of the input frequency. Do you have any idea why this happens?

\$\endgroup\$
  • \$\begingroup\$ Notice the gain is also much less than you would predict. Look at the output waveform, I think you will see it's clipping, i.e. the output is hitting a supply rail (or both). How close to the supply rails is this opamp (no link to datasheet) spec'ed to go? \$\endgroup\$ – Brian Drummond Oct 11 '16 at 21:16
  • \$\begingroup\$ You may also be close to the slew rate limit of the op-amp. I calculate if the op-amp operates ideally, you'd have 19 V/ms slew rate at the output, while the datasheet spec is 24 V/ms, so you might be getting close enough to that to cause some distortion. \$\endgroup\$ – The Photon Oct 11 '16 at 21:21
  • \$\begingroup\$ @BrianDrummond Output is not clipped, as the GBW of the op-amp is rather small (see datasheet fig 2.14 page 9). At this frequency, gain is only 30dB. \$\endgroup\$ – Vincz777 Oct 11 '16 at 21:21
  • 1
    \$\begingroup\$ It's picking up hum, and that's causing some of your problems. The vin- terminal is a "virtual earth" sourced from very high impedance, but drawn on the schematic as a huge area. Make it as small as possible in physical area and, if necessary, put a copper foil screen over it. (If that's with the new, lower, impedances, then the hum's coming from somewhere else) \$\endgroup\$ – Brian Drummond Oct 11 '16 at 21:55
  • 1
    \$\begingroup\$ Did you put a guard ring around the input terminals (see page 18 of the datasheet)? You might be picking up noise from elsewhere in the circuit. \$\endgroup\$ – Andrew Spott Oct 11 '16 at 22:00
1
\$\begingroup\$

Although the Schematic is just a logical diagram , if the loops were as big as the high impedance paths suggest, it becomes a big loop antenna to E Fields created by power line cables. Considering the possible ground sources from Sig Gen and Scope, there is ample opportunity for large CM noise effects on intermodulation or basically distorted amplification. Considering the attenuation of 50Hz thru your high pass filter is -12dB @ 50Hz, it is still bigger than your signal. Make sure a supply cap is near IC.

So many times I have been telling folks to use short inter-connections on high impedances, caps on supplies and minimize single wire loops, otherwise use twisted pair. It's lack of willingness on users to research proper breadboard skills and measurement methods.

Start with clean ground signals, then clean supply , then check each side of each component for noise and make the breadboard look it should and show before and after photos of your construction and results.

\$\endgroup\$
  • \$\begingroup\$ Okay if you're sure problem comes from the noise, I'm going to dig into it (haven't clearly study the question so far). However I don't understand the -12bB you calculated. For me the op-amp amplify strongly high frequencies and practically doesn't amplify low frequencies. Am I wrong? \$\endgroup\$ – Vincz777 Oct 11 '16 at 23:02
  • \$\begingroup\$ you mean you haven't yet figured out the HPF breakpoint? or cant \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Oct 11 '16 at 23:06

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.