CFP For The VAS

[Ultima Thule]

You've got mail. BTW, welcome back! It seems you've been gone for quite some time.

My questions were mainly rhetorical in nature. I was just trying to get you to think about it in a different way. With that in mind, I've attached a picture to try to clarify what I'm getting at.

A properly designed VAS should act in many ways like the op-amp integrator shown in the picture. The current source is meant to represent the diff amp output current for either the simple Rc case or the current mirror. Likewise, Cout and Rout represent the output impedance of the diff amp. For the simple Rc case, Rout=Rc and Cout~0. For the current mirror case, Rout is the output impedance of the current mirror and Cout is its output capacitance.

The open-loop gain of the integrator is assumed to be very large, so the voltage from the inverting input of the op-amp to ground is very small. So then, what effect does Cout have on the circuit?

Hi Andy, thanks for the email (will read the paper a bit later) and thanks for your welcome!
Yes, due to some changes in my life I haven't spent much time here, and gotten a bit rusty when it comes to audio stuff.

To your question, about Cout effect on the circuit, it's not my strongest point to express in an academic way but by frequency the voltage swing for the differential gain stage increases as the gain goes down for the integrator, so the non-linear intrinsic transistor capacitor, here called Cout (right?), the non-linear portion grows. But you already stated Cout~0 so it wouldn't have any effect then... and if the gain is very high (..in comparison to the diff stage we might perhaps explicitly tell..) the non-linear Cout has a very small effect as we know from the mathematics world by magnifying a portion of any part of the integral curve it becomes virtually a straight line.
Well, sorry I don't know what you are fishing after.

 

[Ultima Thule]

Well it simply doesn't make any sense to evaluate the worth of a current mirror on the LTP without analysing its effect on the circuit as a whole.
As for buffering the VAS with an EF (NOT a Darlington), yes, the improvement in (VAS) linearity is still mostly due to the increased Miller loop gain. And what role do you think the EF has in combatting the effects of the VAS transistors Cob???

And we still have not ventured into the full impact of that CM on the LTP.
That increased Miller loop gain not only improves the VAS linearity, it also reduces the VAS output impedance at HF and more effectively pole-splits the effects of the driver input C. Ever wondered why the "Blameless" works so well with just a double EF power output stage?

Hi Glen,

I have a bit difficult to follow your posts. The EF, now you say buffering the VAS with an EF, you mean putting an EF after the VAS?
As long as I talked about EFs in any posts in this thread so far it was meant to be interpreted as put it in between the LTP and VAS stage.

One thing I might have to correct myself on is when I wrote the Miller capacitance I meant the intrinsic non-linear capacitance inside the BJT, sorry for that if it was interpreted as the external Miller feedback component we add between Collector/Base, Drain/Gate or Plate/Grid.
And for the rest of the post I do in fact agree and I think I never said anything else for the matter, I think just we are maybe crossing each others lanes due to some language barrier or how we express many times same things in different ways, no hard feelings mate!

Cheers Michael

 

[GK]

I’m still talking about the EF between the LTP and the VAS.
I’m still not sure why Cob should be a mitigating factor in the case of an ‘unbuffered’ VAS, without an EF. The Value of Cob in parallel with Cdom remains unchanged, and these two capacitances will still feed back the same proportion of signal with respect to each other. The only difference is, when the impedance at the VAS base is raised by eliminating the LTP load resistor with the addition of a CM, the Miller loop gain is increased proportionally, and that is pretty much a guarantee for less VAS distortion.

I guess it is true that no feedback system can be more linear than the feedback element, but I think in this case one would need to be using a terribly unsuitable transistor for the VAS to not get a worthwhile improvement in VAS linearity when adding a current mirror to the LTP.

 

[Lumba Ogir]

Glen,
evidently, you need more spambots.
Essential requirements for voltage amplification are very small conduction losses and high speed. A transimpedance amplifier delivers an output voltage that is a function of the input current, ideally, having zero Ohm input impedance, driven by a high impedance source. The grounded emitter topology does not offer the most favourable characteristics for this task, neither buffered (doing far more harm than good) nor unbuffered. The grounded base topology is much faster since, Cob being put out of the running, Cbe is much smaller, and does not suffer from the Miller effect. Its current gain = 1. The high isolation between the input and output results in little destructive feedback and high stability. An active load improves linearity, and is highly beneficial regarding both output voltage and output current.