LTP to VAS interconnection.

[Dave Zan]

There are a several ways to interconnect an LTP IPS to the VAS.
For instance in a "Blameless" amp, the load of the input transistor is simply connected to the VAS input transistor base.
There are several variants on the Borbely/Lender scheme where the load of the other transistor of the LTP is also used. This can be connected to the emitter of the input transistor of the VAS, as shown in D. Self p. 129, or to the second transistor in the VAS, as used by Bongiorno and also forum contributor "Bonsai".
J.L.Hood published a circuit that connected both sides of a current mirror loaded LTP.
Bob Cordell's book touches this only a little and Self has only the brief reference already mentioned. Didn't find much on DIYAudio but perhaps I missed it in all the noise.
Anyone have any references or an analysis?
If there is little published information then perhaps we can discuss it here and I will post a few pics to clarify.

David

 

[wahab]

Generaly , using the VAS emitter as non inverting input driven
by the non inverting output of the LTP is just totaly useless
given that there s a complete mismatch between the IPS
high output Z and the VAS (emitter) non inverting input that
has low Z.

In other words , the IPS non inverting output has too low
output current to valuably drive the VAS by the emitter ,
so this arrangement is more aesthetical than anything else.

John curl also use this topology in one of his early symmetrical
differentials input amp , the JC2 if IIRC.

 

[Dave Zan]

The apparent mismatch in impedance always looked problematic to me too.
But Andrew "Bonsai" said VAS emitter drive reduced distortion a little in his amp simulations so I decided to re-examine the idea.
I checked the JC2 and your memory is correct, thanks for the reference.

Best wishes
David

 

[jcx]

impedance mismatch is good for signal transfer - not for power matching

"matched impedance" makes both side maximally interact with the other's impedance

for signal transfer we like mismatch = releative independence from either source or load Z variation


if you are collecting oddities then how about Wurcer's AD797 folded cascode/bootstrapped mirror driving unity gain output directly

diff-pair to folded cascodes to mirrors that then can swing full V at the hi Z node are sometimes called a "2-stage" topology - only needs a unity buffer after

 

[Dave Zan]

What I initially found problematic was that the impedance of the loads in the two sides of the LTP looked to be mismatched. D.Self emphasises the importance of current balance and so I assumed mismatched loads would be non-optimal. But he points out that apparently "balanced" resistor loads can be deceptive so perhaps my assumption was premature.

I understand the difference between an impedance match for maximal power transfer and the typical audio amplifier desire for a mismatch.
I do not fully understand how any LTP to VAS impedance mismatch works out when the (so called) VAS is actually closer to a TIS and the impedance is neither practically zero or infinite. And I need to think about the impact, if any, of the use of Miller Input Compensation rather than ordinary Miller Compensation.
So thank you for the comments, I always look at Scott Wurcer's work with careful attention and I will consider more before I reply on that.

Best wishes
David

PS. Didn't Scott actually call this a "one stage" amp? Op-amp people don't count the output buffer but power amp people should

 

[wahab]

But Andrew "Bonsai" said VAS emitter drive reduced distortion a little in his amp simulations so I decided to re-examine the idea.

His amp has very high LTP current , about 10mA , wich allow
for a better drive but even with this feature THD reduction
is marginal at best.

Didn't Scott actually call this a "one stage" amp? Op-amp people don't count the output buffer but power amp people should

Actualy it s Luxman that did call this topology a one stage op amp
(single stage) when they implemented it in their C02 preamp.

http://www.thevintageknob.org/luxman-C-02.html

 

[jcx]

< ~1% imbalance from mirror base current error is usually a lot better than collector R with VAS Q Vbe fixing one collector current nearly independent of the diff pair tail current

with discretes the added mirror current imbalance is probably below even most sorted diff pair Q Vbe match

 

[Dave Zan]

... better than collector R with VAS Q Vbe fixing one collector current nearly independent of the diff pair tail current

That is another expression of the mismatched loads in the LTP halves, so we concur I think.

with discretes the added mirror current imbalance is probably below even most sorted diff pair Q Vbe match

And Samuel has pointed out that mirror current imbalance can be eliminated entirely by cancellation with the VAS bias current. That makes mirrors even more attractive and I plan to use them.
But even with mirrors assumed there are still alternative ways to drive the VAS. I wrote that D.Self covered it briefly. I should have noted that he references a book by Herpy. I probably need to reread it.
I would also like to read the text of the JLH article with the comp. sym. current mirrored LTP with VAS emitter drive. Or can anyone summarize it?

Best wishes
David