Mosfet amp design i just dreamt up

[martseger63]

The circuit itself isn't too complicated for me at all, also i've made quite a lot of self designed and etched pcb's, including even some smd stuff. But i'll look around in this nice forum i just discovered. For now i have to finish new cabinets for my speakers and tweak the xovers a bit. After that i'll be concetrating on the amp, it's a long and dark indoorsy winter period, so i have time to work on this stuff, in summer i fly rc aircraft..
Btw, how long does the moderator preview my posts before they're actually posted here? Is it linked to the number of posts?

 

[Tekko]

So heres a preliminary board layout for this amp thing:

An externally hosted image should be here but it was not working when we last tested it.

 

[Tekko]

Stupid photobucket and their bandwidth limiters...

 

[Tekko]

Third revision, bias circuit now has a trimmer for easy bias adjust:

 

[Tekko]

I upped the voltage on the godzilla amp to 2x50V, had to remove the BD139/140 to get it to work without oscillating though.


Heres the result:
YouTube - 214 watt amplifier

 

[unclejed613]

so, since photobucket has made it impossible to look at the rest of this thread, could you re-post the schematics?

 

[Tekko]

Heres the schematic for the high power amp:

Schematic for the low power amp:

And again, the pics of the stereo amp:

 

[Tekko]

Heres one channel made on a piece of chinese copperclad using chinese rtesistors from one of those cheap 2500pcs resistor kits from ebay, using small signal transistors bought on ebay.

A video:
YouTube - Own design mosfet amp

 

[artu]

This design has this problem (current mirror in the input leg) if you stimulate with a square wave at input with high dv/dt ie: 1usec or less the amp will enter in oscillation (ringing). Remedy: put some resistance at the emitters of Q8 Q9 and add more neutralization (capacitance) for Q2. The drawbacks are less bandwith and more distortion, less slew rate
Cheers
Arturo

 

[Tekko]

In the real amp the cap on Q2 is 100pF. This is still a prototype.

 

[artu]

Maybe 100pf could be enough, but I recommend you to do a ac simulation from 1hz to 100MHz to verify a clean and smooth BW plot (no peaks), remenber that the signal transistors have gain well above 100MHz, and the feedback loop has several frecuencies where the phase is out 180°, that mean positive feedback = oscillation
Cheers
Arturo

 

[unclejed613]

nice... it's basically a big op amp.... don't know that the VAS has enough oomph to drive 12 mosfets though. you need to drive a lot of gate capacitance... 1700pf per device.... that's about .02uf (20nf)

 

[Tekko]

The 12 mosfet model was strictly intended as a subwoofer amp.

This is the simulated frequency response:

For this circuit:

 

[benb]

I see C7 in the schematics in post #27, it's between the bases of the current source (or technically, current sink) transistors and ground. This should instead go directly across the top and bottom of the three-diode string, otherwise it will amplify any voltage variation between ground and the negative rail, and it could also contribute to instability/oscillation problems.

I'm not sure exactly what it's for to begin with, presumably to stop any noise generated by the three forward biased diodes. It looks like a rather low impedance node, so I'm guessing it should have a larger value to go down to 20Hz.

 

[Tekko]

I'm not actually using the C7, its not actually needed, but those who are allergic to any kids of noise may want to use it anyways.

 

[Bonsai]

Any reason why you don't degenerate the LTP pair?

 

[Tekko]

Update:

 

[martseger63]

What is on the separated piece (the one with Q8) and where does it connect?

 

[Bonsai]

With the now added LTP degeneration, the value of C1 can be reduced to improve the slew rate.

 

[Tekko]

It's the bias transistor, and it connects to the pads on each side of the feedback resistor.