You realize that the Thevenin source impedance is the combination of R2 and R8, and therefore any large value for R8 has very little effect on the effective source impedance?
Last edited:
Do you have anything better to say than wise *** comments; I see them toward
many different members here and really, they are not that clever or entertaining.
You have no idea what is already in my technical library.
Why don't you just send me the book?
Last edited:
As a matter of fact, you've done this before here:
http://www.diyaudio.com/forums/soli...lls-power-amplifier-book-806.html#post4952477
and:
http://www.diyaudio.com/forums/soli...lls-power-amplifier-book-806.html#post4953090
I bought everything that Linear Audio has to offer and after reading this in Kolinummi's book, showed
that he was wrong in his analysis further back in this thread - I didn't bother to read the rest of his book:
If you take the "output" for the FFT at the base of the non-inverting input then you
will measure the distortion due to the non-linear input current or input impedance alone.
My understanding is that you want to match the DC path resistances at the two inputs
so that the finite base current produces the same voltage drop across those path resistances.
This maintains the idle match in Ic (balance) for the diff pair assuming they are also matched
for Vbe and beta as they would be in simulation.
I've never read about any other reason to keep those path resistances matched - anyone?
Hi Pete,
Please have a look at Super TIS fig. 5 and the text just above fig. 6. If I alter the value of R1, thus introducing an unbalance, the distortion does increase (at least according my simulation). But this does not happen if I zeroed all the capacitances of the input transistors. So I concluded that the distortion increase is due to the (nonlinear) Miller effect.
Not sure if this applies to other input configurations, of course.
Cheers, E.
Hi Pete,
If one follows your suggestion, then this is too pessimistic in my understanding. Measuring the same at the inverting input will also reveal distortions, which partially cancel at the summing point feeding the "VAS" input.
But as I said: perhaps we are speaking about different things.
I will withdraw from this -- interesting and difficult -- discussion until next year, as I'm moving to an Internet free zone for some days.
Kind regards,
Matthias
my impression is that we are running into trouble due to "weakly defined terms". I always understood, and maybe Douglas Selfs's definition is another one, that this input distortion is something additionally to the other contributions.
If one follows your suggestion, then this is too pessimistic in my understanding. Measuring the same at the inverting input will also reveal distortions, which partially cancel at the summing point feeding the "VAS" input.
But as I said: perhaps we are speaking about different things.
I will withdraw from this -- interesting and difficult -- discussion until next year, as I'm moving to an Internet free zone for some days.
Kind regards,
Matthias
Thank you, Edmond, for sharing this idea. I will try this with my BJT input circuit, next year.
Kind rergards,
Matthias
Thanks Edmond, I'll have to study that in some depth.
I've always blamed the nonlinear Miller capacitance for increased distortion at HF with
higher source impedances but never noticed a balance effect - interesting.
Keeping the differential source impedances the same can improve PSRR if the LTP is not cascoded. But other than that, and DC offset effects, the linearity gained from optimizing differential source impedances is usually only fractional and not necessarily worth designing the circuit around.
After all, if it matters that much you can just add buffers to your LTP. I don't know why this seems to have gone out of favor, but it does help enormously with the robustness of a BJT input amp. All the chips do it, why not us.
After all, if it matters that much you can just add buffers to your LTP. I don't know why this seems to have gone out of favor, but it does help enormously with the robustness of a BJT input amp. All the chips do it, why not us.
After all, if it matters that much you can just add buffers to your LTP. I don't know why this seems to have gone out of favor, but it does help enormously with the robustness of a BJT input amp. All the chips do it, why not us.
I just sim LTP input with buffer several days ago and yesterday (different topology) and then read this.
But my purpose was to get high input impedance and high slew rate. I can set LTP current high enough.