My update lin topology

[GEirin]

Hi all.

R16: Load resistor, for to limit the Vas gain.

R5,R7,C4 network: for to stable the diff input.

C8: Phase lead capacitor, for frequency compensation.

R9,C6: Filter Vas input, for no to use miller compensation
capacitor.

D1, R15 network: for reduce cross conduction and protects the drivers transistors.

I will study the comment for improve my project.

Thanks

Guillermo

[GEirin]

Hi all.

My version II the update lin topology amplifier.
Any comment for improvement are welcome.

Thanks

GEirin

[AKSA]

Hi Geirin,

Nice circuit; simple, effective, musical.

Can I ask why you use FETs for drivers, rather than bipolars?

Isolation from a reactive load? If so, maybe it's a good idea......

But you could more cheaply achieve this isolation with a bipolar driver such as a 47943/1837 and keep the Vbe of the predriver more constant for superior linearity.

You should be able to run your VAS at a lower current, too, around 6mA. Keeps dissipation lower in the VAS and reduces the lag comp constraints.

I don't believe dissimilar voltage supply rails are needed; use the output stage to clip, not the negative swing of the VAS.

Cheers,

Hugh

[GEirin]

Hello Hugh.
Thanks you very much for you reply.

Why mosfet drivers? Because in the other thread "RMI-FC100" (Roender), in post 66 you said "Your output triple is good, but I´d
try replacing Q1/Q3 with low current thermally robust mosfet, say IRF 710 series.... And I think, good idea.

I´m a diy, but not an expert in amplifier designer, I´m Industrial Designer and diy audio is my hobby.

You think: a bipolar driver such as 2sa1837-2SC4793 is the best. OK. I will use it.

Sorry, any questions:
1- What is your opinion of the Sziklay VBE multiplier?
2- Stopper resistor in predriver: yes or not?

Regards
Guillermo

[lumanauw]

Hi, GEirin,

If you aim for isolation, will it be better if the mosfets are placed for Q6-Q7, instead of Q8-Q9?

[AKSA]

Hi Guillermo,

My pleasure. Remember, I only have an opinion, and nothing is more dangerous...... I may well be wrong.

I don't enjoy it when people throw back what I said weeks or months ago!! Posts don't convey this well, but consider that I might have been saying one thing but meaning just the opposite; this is common with Australians.

Steve McCormack, a famous US designer, uses mosfets as drivers too. His response is that it gives excellent isolation, but his configuration is, AFAIK, source follower then emitter follower.

However, the advantage of the CFP as used also by Mihai is that the input impedance to the predriver, the 3423/1360, is VERY high, on the order of megohms, so loading on the VAS is much reduced. This is very important.

But it's a little higher with a little less local feedback using a mosfet than it is with a bipolar, and the distortion introduced by Vbe variations will be less with bipolar.

Lastly, consider the cost factor. I think the 4793/1837 combination I suggested could be cheaper and more robust during manufacture than the mosfet......

I could not say about the Sziklai Vbe multiplier. This is usually a critical part of bias stability for CFP output stages, though this stage will be more forgiving because the base/emitter of the output devices is in the thermal loop, unlike a two device CFP. A standard Vbe multiplier might be too enthusiastic; you'd need to put a resistor into the emitter circuit as you have done. Finding the best value for this resistor would depend on thermal coupling with the output stage and you would need to arrive at the values empirically. I think a diode is not a good idea as it shares a similar tempco to the Vbe junction and this would double the slope of the Vce variation with temperature.

A stopper in the predriver of 100R is a very good idea.

Hope this helps,

Hugh

[lumanauw]

Hi, AKSA,

However, the advantage of the CFP as used also by Mihai is that the input impedance to the predriver, the 3423/1360, is VERY high, on the order of megohms, so loading on the VAS is much reduced. This is very important.

I've been wondering about this too. If we swap Q8-Q9, then it becomes a darlington driver. How about the input impedance of this (darlington driver) compared to CFP driver, is it about the same or quite different?

[roender]

Originally posted by lumanauw
Hi, AKSA,


I've been wondering about this too. If we swap Q8-Q9, then it becomes a darlington driver. How about the input impedance of this (darlington driver) compared to CFP driver, is it about the same or quite different?

David,

Did you receive my e-mail regarding this question?

Mihai

[destroyer X]

A good idea to construct to evaluate real world hermano!

muito bom.... felicitaciones.

regards,

Carlos

[GEirin]

Hi.

Thanks for yours replies.

Hugh:
Thanks you very very much for your detailed explanations. I will improve my amplifier.

Hermano X
Muchas gracias. Haré los cambios en el circuito.
Saludos.

Regards,

Guillermo