I am working on a redesign of a test box we have at work. There is a transmitter that is sending data to a recording system (in my case the dewetron trendcorder with Trion BNC modules). We previously outsourced the design of this test box and that company used unity gain buffers from the transmitter side into the recorder.

The output impedance of our transmitter is less then 10K ohms and the input impedance of the recorder is 10M ohms. With that ratio I am not worried about a significant voltage drop but I am worried about channel ghosting as we will be switching channels. The trion module doesn't specify the input capacitance for its internal ADC.

We are trying to make this redesign as simple (and cheap) as possible and not having to design a PCB (even though its fun!) would save a lot. I am trying to decide if the vendor was overly cautious with these unity gain buffers or if they do serve a purpose. I am using this tutorial from NI for reference. Ghosting tutorial

I am leaning towards the side of just designing the PCB to be safe. Is there something else I need to be aware of here?

  • \$\begingroup\$ I must admit I have absolutely no idea what a dewetron trendcorder is (might as well be a Rockwell Retro Encabulator), but things like switching rate and signal bandwidth would absolutely be necessary to help you. \$\endgroup\$ Mar 31, 2017 at 16:17
  • \$\begingroup\$ also, are you positively sure these are unit gain amplifiers and not say, low pass filters? \$\endgroup\$ Mar 31, 2017 at 16:25

1 Answer 1


If I understand your question correctly, your cautious predecessors have put unity gain buffers to reduce the source impedance seen by the mux. You have concerns that these buffers aren't needed. Make an experiment. Temporarily remove the buffers and replace them with shorts. Compare measurements with and without buffers (all other things being equal).

I wonder why you're worried about these buffers in the first place?
Why are you worried about this test hardware being as cheap as possible? This is a one-off test boards.
The buffers are already there. Not that you need to add them. It ain't broke.

  • \$\begingroup\$ I don't have the test box and I can't get it. I pretty much have to do a complete redesign but I can use the old design for reference. Not having these unity gain buffers would save time on not needing a PCB. There are very tight time constraints. The sampling rate of the recorder is 204.8 kS/s per channel. \$\endgroup\$
    – Mtk59
    Mar 31, 2017 at 16:53
  • \$\begingroup\$ Switch-noise/charge-injection can bite you. If you're switching that mux to support multiple channels at a 200kHz sample rate, then I'd recommend that you include the buffers as a precaution (especially if you can't fully characterize the system). The alternative is to reduce the switching speed of the mux for more settling time. Adding the buffers also gives you the opportunity to add support for over/under-voltage protection, scaling, and filtering, if needed in the future. \$\endgroup\$ Mar 31, 2017 at 17:19
  • \$\begingroup\$ what @ChrisKnudsen said, but the question really is: what is your signal bandwidth? How often do you switch? \$\endgroup\$ Mar 31, 2017 at 17:20
  • \$\begingroup\$ The data sheet says .38 fs. I think its best if I do include the unity buffers. Thanks! \$\endgroup\$
    – Mtk59
    Mar 31, 2017 at 21:42

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.