No announcement yet.

1st topic here! Distortion and global feedback

  • Filter
  • Time
  • Show
Clear All
new posts

  • 1st topic here! Distortion and global feedback

    I'm in the final build phase of a quite unique amplifier design. My own discrete jfet/mosfet design. It is to some extent inspirered from Nelson Pass, but also from some other giants in the industry. But, I have made it in my way, so no copies.

    One thing that I find quite hard ot understand is the effects of global negative feedback in terms of harmonic distortion and especially intermodulation distortion that may be created by the feedback itself. Of course negative feedback reduce THD, output stage impedance etc. To make a good feedback loop I think one should start with a really good open loop performance and just use negative feedback to improve linearity and reduce distortion a bit. And not too much. My experience is that some global negative feedback can be good if the forward path is fast and perform well to begin with. In my current design I have only 6dB feedback and 0.3% THD at 100W/8ohm (spice). I have worked quite hard to increase the speed of the three stages and now I have a total lag of <200ns. I think this is an important figure for being able to have a good design using negative feedback. To lower the output impedance of the amplifier I use a little (adjustable) positive current feedback. I have an adjustable H2-harmonic function for getting a what I think a good distribution of harmonics across the spectrum.

    But, what about intermodulation distortion assuming some global negative feedback? It seems like with music as sound source, negative feedback may create complex intermodulation components a some level are spread over the spectrum. Would be very interesting to discuss thoughts and ideas within this area.

  • #2
    Usually what goes on with feedback is that the feedback can suppress the innate distortion of the amp, but in doing so adds some of its own through a process of bifurcation. Additionally IM components can be introduced at the feedback node.

    If you get enough feedback, then the amp can compensate for distortion added by the feedback. That seems to take north of 35dB or so. Obviously getting enough Gain Bandwidth Product to pull that off is a bit of a trick.

    Higher ordered harmonics are perceived by the ear as brightness and harshness. As a result in most applications, feedback results in the amp being brighter and harsher than real life. It appears that enough presence of the lower orders (2nd, 3rd, 4th) can mask the presence of the higher orders, resulting in a smoother sound. This is part of why tube amps tend to have a smoother quality to them.

    Good luck with your design! It sounds quite interesting!


    • #3