F5 Semi-Turbo with Lateral Mosfets...

[fab]

Hi,

This is my version of an F5 "semi" Turbo with Lateral mosfets. I call it "semi" because of the limitation in current of the Lateral compared to verticals and I want to use only +/-42Vdc.

As a reminder one issue with Lateral in the common source configuration such as the F5 is the low transconductance which limits the output stage gain.

Second issue is the low threshold voltage compared to Laterals which reduces the jfet load resistor thus reduces the gain of the first stage.

In order to try to solve both issues, we can use 4 Lateral mosfet to have a similar output stage gain as a single vertical. Also, we can use an intermediate mosfet stage which serves 2 purposes:
1) increase the Load resistor of the Jfet input stage thus increase gain (vertical mosfet has a high threshold voltage)
2) Allow to use a higher drive current for the mosfet outputs and avoid using jfet in parallel

Since the GSD pin out of Lateral compared to GDS of vertical mosfets, there is limited PCB available in common source configuration output stage.

My goal is to use a standard F5 PCB available from multiple vendors and then:
1) use a custom daughter board to mount the cascode input stage (b/c 42V)
2) build a simple point-to-point output stage using Double-die Lateral to limit the number of connections.

Since I have already the power supply +/-42vdc and the heatsinks, it will be a class A-B with about 800ma bias.

See attached concept. Before I build it I would appreciate comments on the idea and improvements.

Thanks

Fab

[AndrewT]

What happens if the two cascode halves charge up at different rates?
Will that output a big offset?

Can the two 22uF be replaced with one 10uF and no middle connection?

The extra driver stage will introduce extra phase error. This may require compensation to maintain good stability. The standard F5 already has significant overshoot on fast signals.

[fab]

Originally Posted by AndrewT

What happens if the two cascode halves charge up at different rates?
Will that output a big offset?

Can the two 22uF be replaced with one 10uF and no middle connection?

Hi AndrewT

That is interesting questions but this cascode is exactly the same as the recent F5 Turbo V3 of Pass. See this link for Nelson comment: F5 Cascoded - an alternative for my interest in the Balanced F5
Also: F5 Cascoded - an alternative for my interest in the Balanced F5

I taught that the charge up would happen mainly at amp power up...

Originally Posted by AndrewT

The extra driver stage will introduce extra phase error. This may require compensation to maintain good stability. The standard F5 already has significant overshoot on fast signals.

I have also a standard F5 but I do not remember it had significant overshoot. However I will try to test it to confirm maybe tonight.
For this design with extra driver stage it does not seem to have a significant overshoot according to the simulator with 10n to 1uf in parallel with 8 ohms at Square of 10KHz... However, I can install a small miller cap if really required.

My goal here is to get a similar feedback factor as the standard F5 and improved linearity because of the Lateral mosfets. Hoping that the sound result will worth the effort.: rolleyes:

Thanks

Fab

[fab]

Hi

I have attached a scope picture of my standard F5 (not the semi-turbo one) response to 10KHz with 8 ohms- 1 Uf in parallel load. Finally the overshoot seems damped enough....[B]

[AndrewT]

Take out the capacitor loading and see if any overshoot remains.

[fab]

Originally Posted by AndrewT

Take out the capacitor loading and see if any overshoot remains.

See attached, no overshoot at all. The picture is blurry but it is a straight line. I must admit that I have a 68pf cap at input of amp but my generator is relatively fast (3MHz capacity for Square signal).

Thanks

Fab

[fab]

I have assembled one set on my bench. As Andrew pointed earlier, the stability is less easy to obtain than with a standard F5 and we can see the overshoot on only resistive load. I had to add 220pf miller caps and increase the input cap to about 470pf and then I got the attached results.

Notes:
I have not installed the cascode input stage so far. I have tested with about 30Vdc only.
I have only one set of Double die mosfet instead of 2 yet.
The output impedance seems to be high at about 1 ohm...

1st scope: 10kHz, 8 ohms only
2nd scope: 8 ohms and 470nf cap in parallel
3rd trace: 8 ohms and 470nf cap in parallel but with added inductor at amp output
Any comments?

Thanks

Fab

[Paulv]

Hello FAB
I allso want to build an F5 with the same lateral mosfets and I tried to imagine different sorts of current draw to solve the issue of the low threshold and I came to the following idea (sorry I am not used to post images):
between R1 (4 kOhms?)and jfet insert a zener diode (15 ...20volts)
allso connected between jfet and supply rail: a resistor (+- 2k Ohms depending on bias)
with LTSPICE (beginner and problems with right models...?) it seems to work quite well
I would apreciate any comments
Thanks
Paul

[Paulv]

I do not know if a zener diode would degrade sound quality
what make to use
or if I should put a cap in parallel
any advice?

[fab]

Hi Paulv

Are you removing the intermediate mosfet stage?
Where is the zener diode?
I am not sure to clearly understand your suggestion.

Thanks

Fab