Please rate designs for n-ch mosfet amp

[bjornagain]

I principal, there is three different designs for n-channel mos-fet power amplifier that I have come across so far. I like to keep it simple and believe a good performing amplifier don't have to be complicated. Inspired by Pass MF-12, Holton N-channel and others, I think it would be fun to design and build one myself, but which one? Target output fets are IRFP520.

I would appreciate comments or experience from any of these designs. Sound, general performance, stability is of course of interest. And of course, how do they compare?

[bjornagain]

Design A uses a double differential and is perhaps the most symmetrical of all designs. The second stage (Q1, Q2) don't have to be differential, a conventional CE works fine.

This design have been the hardest of the three to perform well in simulation.

Pros: Symmetrical, uses a real differential amplification.
Cons: ?

[bjornagain]

Design B uses a single transisitor for phase shift. All amplification is in the input stage (Q4), Q3 usally has a gain of one but can be more. It is simple, simulates good and is a proven design.

Holton N-channel is a working example of this design!

Pros: Simple, proven.
Cons: ?

[bjornagain]

This design I think is the most classical one, a two stage amplifier (Q3, Q4), with phase shift (Q1), and voltage feedback (Q2). It has been proven over many years and works well.

Pass MF-12 is an example of this design.

Pros: Simple, classical.
Cons: ?

[Tube_Dude]

How you adjust the bias of the output mos-fets?

[pjacobi]

If "simple" is in your shopping list, nothing beats true single ended operation against a current source. Of course this halfs the efficency, but something can be regained when using bootstrap to achieve rail to rail output operation.

Look at this example circuit (think MOSFETs instead of IBGTs):
http://www.linearaudio.de/scratch/IGBT-3.pdf

Regards,
Peter Jacobi

[bjornagain]

Originally posted by Tube_Dude
How you adjust the bias of the output mos-fets?

Essentially by adjusting current through the first gain stage, or in design C with a discrete and adjustable zener. Anyway, there must be some temprature compensation.

Originally posted by pjacobi
[B]If "simple" is in your shopping list, nothing beats true single ended operation against a current source. Of course this halfs the efficency, but something can be regained when using bootstrap to achieve rail to rail output operation.
...

I don't agree the complexity is an issue of class A or AB. The complexity and component count of your proposal is very much the same as in design C, and design B is simpler. Nice with CFB though! What do you think is the audible difference between VFB and CFB from your experience?

By the way, fet's are IRFP260N: Vdss=200V; Rdson=0.04ohm; Id=50A; Pd=300W. Nasty!

[pjacobi]

Originally posted by bjornagain
I don't agree the complexity is an issue of class A or AB. The complexity and component count of your proposal is very much the same as in design C, and design B is simpler. Nice with CFB though! What do you think is the audible difference between VFB and CFB from your experience?

That was overlap at posting. When I did my posting I haven't seen your simpler designs B and C yet.

CFB issues (Hey, where is the CFB police? Strictly, speaking this isvoltage sensed feedback into a low impedance node):

I can't do my day job, post on this forum and actually build these amps. Anyway, I'm waiting for more cheap grabs at Ebay, before my first prototype will be done. So I can't tell you of subjective differences.

It's more philisophical, Occams Razor as applied to amplifier design. Why the heck buffer the feedback signal with a follower when it has a source impedance of milliohms?

Regards,
Peter Jacobi

[azira]

Comments/Questions...

Design A: The collectors of Q1/Q2 go to different rail voltages. This should make it very hard/impossible to find resistor values that balance the currents and gains... does it work?

Design B: I don't think Q3 can be set for more than unity gain because it's also a phase splitter. If you add gain, it will only be added to the inverted phase.

Design C: Mosfet-12, I want to build one of these, I got some IRPF450s cheap...

It seems like to me that all these designs are considered 3-stage amps instead of 2-stage.

Design A vs Design C, I would suggest Design C. Same number of transistors. It's not as symmetrical but you will win on HF response.

[x-pro]

You can add my design to your list as well:

N-channel only output devices in a power follower.

Regards


x-pro