Post Go back to editing

Noisest Opamp?

Now that I've got your, no, this is real.

My company wants to sell into China without having to wait months for an export license, we can accomplish this by making the resolution of our systems worse, especially in the bandwidth below 100 Hz.  I know that FET amps tend to have the most noise down in that area, but what is the noisest one?  We have amps in the SOIC-8 package (duals).  I'm also going to sprinkle some carbon comp resistors in there to boost the noise.

I know that I can do a redesign to add a noise generator, but I'm trying to get around this without a redesign and re-layout.


  • There two AD8512 and one AD8672 in the circuit.  This image is the first amp after the demodulator.

    I need to increase the noise below 100 Hz.


  • Hi Steve,

    Have you thought of just dropping in a LM741  (or LM324). The specs on these old opamps are pretty awful.

    (below is from the OP27 datasheet) ,input offsets may be a problem then.

    Putting say 0.1uF in series with R25 will cripple the CMRR below 100Hz. That may be sufficient to achieve the desired effect.

    Many video and super high speed opamps are noisy at low frequencies , if the noise isn't specified on the data sheet, then look for offset voltage drift, anything around 100uV/degC will be real noisy below 1kHz .

    Your circuit has close to unity gain, so won't add much noise unless followed by a high gain stage.

  • Thanks much.  While I like the idea of the cap in series with R25, I'm going to have to do something without going through a board turn. For gain, I had planned to perhaps boost the gain at this stage and allow the later range adjustment stage work lower.  There is a log amp in-between, so there would be some linearity change.  My thought was to find an old FET opamp since those usually have bad low frequency noise.

  • Hi Steve,

    Ok looked through some old databooks, pretty much everything (including older FET designs) is designed to LM741 noise figures. Your source resistance tends to favour FETs anyway. Better off looking at amps designed for lower Rs.

    Two other choices

    (a) use a video amp , these are designed for Zin=100ohm , thay also have 1/f knees in 1kHz to 10kHz range , they can have really low voltage noise above 10kHz, BUT current noise is much higher.

    So take for example the OPA2674 (this only gets to +/-12V but there are more out there) it has an impressive voltage noise of 2nV/rtHz, but current noise of 100nA/rtHz at 100Hz ,

    so at 100R source resistance, at 100Hz its 12.8nv/rtHz

             ~3K source resistance will yield 300nV/rtHz at 100Hz

             ~30k source resistance yields 3000nV/rtHz

    So you can tailor the noise by changing the input/feedback resistors (and the caps).

    Also a possibility the video amp may oscillate (might fix this by lowering C26, C27 and swap R27 R28 for ferrite beads (Zo~ 2k)

    (b) Use a push-pull comparator (or differential line receivers) (with same pinout) these aren't intended as opamps, so would expect higher noise, you will need to test them first as they may oscillate, and they will have some crossover distortion, and possibly slower slewrate.

  • And regarding "For gain, I had planned to perhaps boost the gain at this stage and allow the later range adjustment stage work lower.

    Putting all your gain up front generally produces lower overall noise; So if you are trying to make the overall system noisier, then use lower gain up front. 

    If there is a log-amp in between then adding gain after the log amp is fairly pointless,  consider a random log-amp example:

    Vin 1mV-> 600mV out and  Vin 10mV-> 800mV out  and 100mV -> 1000mV out and 1000mV in -> 1200mV out

    So the changing the first stage gain by a factor of 10 changes the output by ~ 20%.

    Note that the log amp will also produce intermodulation,  so for example if you have a 20kHz carrier with some signal modulation, then the extra noise you add will appear as noise on your signal (and on the carrier), but maybe that's the effect you were after in the first place.

  • bobsbots,

    Sorry for the late reply.  Apparently someone in management actually grew part of a brain and realized that no matter how we make the units noisier, the customer can always modify it to improve the performance. That was about a month ago and nothing has happened since.  Then the sales guy who was pushing for this had given his notice.  Management will probably do what it always does, do nothing hoping it will go away.

    Thanks for all the ideas, I've marked one of your posts the 'Correct Answer'.