Post Go back to editing

Source impedance for driving difference amplifiers


I am designing a conditioning circuit for a MEMS analog microphone and I would like to use an IC for single-ended-to-differential conversion in order to have a balanced line output capable of driving twisted-pair cables. I found that the difference amplifier AD8273 is recommended for such a task in AN-1165 (Fig. 9, pp. 6), and it is directly connected from the output of the MEMS to the inputs of the AD8273 (via the AC-coupling cap. C1).

Since my application demands constrained PCB dimensions (as small as possible) and low power, I would rather use the AD8476, particularly the circuit found in its datasheet Fig. 54, pp. 19. However in such a circuit I would expect that the source impedance, depending on its value, demands the need of a voltage follower right between the source and the input terminals of the AD8476; mentioned in the section "Driving the AD8476", pp. 18, and it should be kept below 0.1 ohm. In other words, an additional but undesirable op-amp should be included in the design.

So my main question would be to what extent the MEMS output impedance (say 200-450 ohms) will unbalance the resistor ratios of the AD8273/AD8476 and affect its performance? Are there any other alternatives for smaller/more compact designs capable of dealing with the MEMS impedance and providing a differential output? I was already thinking of the AD8295.

Many thanks in advance for your kind attention and support, I look forward to hearing from you.

Best regards,

Elías Zea

  • I also would use the AD8476.

    While you are correct that the microphone output impedance will reduce the otherwise perfect performance of the differential amplifier, it will not matter in the big picture. This is because the low output impedance and the twisted

    pair cable will ensure that any possible 60Hz pickup will be unmeasurable.

    You may want to drive your twisted pair via some small resistors and you may want to use a common mode filter on the receiving end of the twisted pair, because high common mode rejection on the receiving end is only for low frequencies. Higher frequency electromagnetic pickup is not necessarily removed.

  • Thank you for your message KurtBecker. Due to size constraints I had to change the design so now I intend to use 3.5mm audio cables instead of twisted pair. In this case I assume there is no need for resistors to drive the cables. Would it still be necessary to use the AD8476? Do you have any ideas of what would happen with common mode rejection then?

    Many thanks in advance.


  • I would not really overthink this issue. The trick is to prototype cheap and fast. Virtually any low noise amplifier will make a good audio amplifier.

    Unless you do some serious mistake somewhere, you will not be able to measure the difference between amplifier A and amplifier B.

    Power consumption matters, and you want to use a single supply only. How does the amplifier behaves when overdriven, what happen when your cable is very long and you have some serious noise pickup. No amount of reading the data sheet will answer these questions.

    Then, for long cables differential is better.

    In the old days, it was a power transformer (60Hz) which caused all the noise, even when the single ended cable was screened. If you have a large 60Hz magnetic field nearby and the cable is long, you must use differential. But the common mode rejection is not limited by the receiving amplifier, it is the precision of the twist in the cable.

    Again, just build it. Make yourself a bunch of litlle PCBs which take the tiny footprints from the opamp and solder up a "rats-nest" on a copper-clad board.

  • This question has been assumed as answered either offline via email or with a multi-part answer. This question has now been closed out. If you have an inquiry related to this topic please post a new question in the applicable product forum.

    Thank you,
    EZ Admin