Has anyone seen this front-end before?

[Michael Chua]

Originally Posted by Edmond Stuart

Hi myhrrhleine,

You are on the right track. Indeed, the Baxandall super pair closely resembles the Hawksford cascode (hence the reference to his article on my website), but a crucial difference is that my TIS has a current gain of just 1x (instead of beta or even beta^2).

Cheers,
E.

I see cascode Q10 base is actually driven by Q8. Is there an improvement on the cascode performance as compared to the traditional method of using diodes/zeners/resistors?

[Edmond Stuart]

Originally Posted by simon7000

E.

It has lots of similarity to a number of circuits. Cherry at first came to mind, but the exact design with your intent seems like a new twist. (The adage of enough chimps pecking at typewriters will eventually produce Shakespeare comes to mind, preventing an absolute claim at exclusivity.)

It's precisely the twist that counts.

So go for it!

ES

Sure, I'll go for it.

Cheers,
E.

[Edmond Stuart]

Originally Posted by Elvee

I have never seen or done something exactly identical.

I have used variants, mainly without the cascodes, and with the collectors of Q5/Q6 directly tied to R15/R16.
A sort of an afterthought, a way to recycle an active current that would otherwise be wasted.

But anyway, I am not very fond of double LTP's and their closed translinear loop, I didn't mind very much, and I didn't notice anything extraordinary.
I may have missed something.....

Hi Elvee,

I'm afraid you missed an important aspect. I'm sorry if I'm not clear enough. Perhaps it helps if you also read my website on this subject ('where TMC really shines').

Cheers,
E.

[Edmond Stuart]

Originally Posted by Michael Chua

I see cascode Q10 base is actually driven by Q8. Is there an improvement on the cascode performance as compared to the traditional method of using diodes/zeners/resistors?

Hi Michael,

Absolutely! Please see the Hawksford cascode: Malcom Hawksford, "Reduction of Transistor Slope Impedane Depedent Distortion in Large-Signal Amplifiers.", J. Audio Eng. Soc., Vol. 36, No. 4, 1988 April. His paper can be downloaded here: http://www.google.nl/url?sa=t&rct=j&...zktaKw&cad=rja
My implementation differs slightly as it is based on: Dimitri Danyuk's, "On the Optimization of Enhanced Cascode", AES paper, October 2008.
It differs in that the base current of the cascode transistor is almost completely 'recycled' (i.e. by a factor alpha), while in the Hawksford cascode this base current flows back partly into the preceding trannie and partly into the emitter degen. resistor. So some of it gets lost.

Cheers,
E.

[keantoken]

One of the important things in this circuit is the high impedance at Q8's collector and it's complement. This makes the Baxandall pair work exceedingly well. However it may work too well and oscillate.

Also, the 100k resistors connected to the current mirrors will eliminate PSRR issues, which is one problem with the Baxandall cascode, it needs a good source.

It does not correct the basic nonlinearity of the input pair Vbe's, but they could easily be modified with someone's solution to this. Degenerated cross-comp (Doug Self's book) comes to mind, although without degeneration it may just be more nonlinear.

The Baxandall cascodes may cause serious problems with stability, do you know what to do if this happens?

I like it a lot!

[homemodder]

Hi Edmond

Interesting design with very good performance.
Never seen a design or patent with this concept.
Do you have the article by John Ellis, Ive been working on a certain design where I need to curtail ULGF by a fair margin and there might be useful info I could use.

[Edmond Stuart]

Originally Posted by keantoken

One of the important things in this circuit is the high impedance at Q8's collector and it's complement. This makes the Baxandall pair work exceedingly well. However it may work too well and oscillate.

Hi keantoken,

You're right and I'm aware of this caveat.

Also, the 100k resistors connected to the current mirrors will eliminate PSRR issues, which is one problem with the Baxandall cascode, it needs a good source.

Nice (and simple) trick, isn't it?

It does not correct the basic nonlinearity of the input pair Vbe's, but they could easily be modified with someone's solution to this. Degenerated cross-comp (Doug Self's book) comes to mind, although without degeneration it may just be more nonlinear.

Well, yes and no. Nothing has been done to lower the distortion of the input stage itself, that is, when operated in isolation. BUT... the crux is that everything has been done to lower the (differential) signal level. At full power and at 20kHz in the order of only 1uA at the collectors or 10uV between inverting- and non-inverting input. With such low signal levels, 'tanh' can't do any harm.

The Baxandall cascodes may cause serious problems with stability, do you know what to do if this happens?

Yes. Please have a look at my (updated) website , Fig.2, C10 & C11.

I like it a lot!

Thank you. I would be very flattered if you would build this front-end.

Cheers,
E.
BTW, what's your real name?

[Edmond Stuart]

Originally Posted by homemodder

Hi Edmond

Interesting design with very good performance.

Hi Homemodder,

Thank you!

Never seen a design or patent with this concept.

So it might be new.

Do you have the article by John Ellis, Ive been working on a certain design where I need to curtail ULGF by a fair margin and there might be useful info I could use.

Regrettably, I've lost that article. Maybe somebody else would be so kind to send us a copy.

Cheers,
E.
BTW, what's your real name?

[bear]

Dizzy. The music goes round and round...

Speaking of which, while I have all of you here looking at that schematic, might I request that you all take a look here, and lend a hand? There may be some similarities, but I am not sure:
Les Sage Amps
Regardless of similarity, the question is how does this one work?? It is alleged to have both sliding bias and also SOA protection of some sort. (post 44 and beyond have real schematics)

_-_-bear

[Edmond Stuart]

Originally Posted by AndrewT

how do I1 & I2 get current from/to the supply rails?
The only route I can see is through the shorted CtoB of Q2 & 3

Hi Andrew,

Please have a look at my (updated) website , Fig.2, Q7 & Q8.

Cheers,
E.