Opinion about this hi-fi amplifier

[viktor1986]

Hi

What's your opinion about this amplifier?

[Gold_xyz]

Hi viktor1986

at the first look it seems good !

[MikeB]

I think, you should not use the "current mirror" formed by Q7/8. This way the outputvoltage to the 2nd stage is not clearly defined, giving unreliable bias in the 2nd stage.
It would work better with r21 replaced by a ccs, but the common mode voltage is still not defined, which can be anywhere between 0v and -Vcc. (Set by the ratio of currents delivered by Q7/8 and input-diffamp)

It's likely that the amp built that way will have no bias at all, or burn up r21.

For a better DC-balance you might move r1 to the other side of r2.

Okay, enough critics for one post...

Mike

EDIT: For a better understanding: This topology needs a voltage output from 1st stage rather than a current output.

[jmateus]

It really looks good! Is there any layout yet?

[XEAGLEKEEPER]

Looks real similar to ESP P101 lateral mosfet amp.

[P.Lacombe]

Hi MikeB,

Q7/Q8 don't form a current mirror, but a dual constant current sink. This can improve the linearity of the 1st stage, and increase dramatically the open loop gain...
... which can be the very best or the very worst thing, depending of the phase response of the whole amplifier. In other words, building of this amplifier is reserved to experienced persons...

[Wavebourn]

I like power filters with diodes.

[MikeB]

Originally posted by P.Lacombe
Hi MikeB,

Q7/Q8 don't form a current mirror, but a dual constant current sink. This can improve the linearity of the 1st stage, and increase dramatically the open loop gain...
... which can be the very best or the very worst thing, depending of the phase response of the whole amplifier. In other words, building of this amplifier is reserved to experienced persons...

Yes, that's true, that's why i wrote "Current mirror".
It does increase openloop gain, but the issue with voltage source vs current source keeps. The bias to 2nd stage is undefined. Try to calculate bias in 2nd stage... (without sims)
This would work perfect if 2nd stage would connect to 1st using folded cascode.

Mike

[darkfenriz]

I completely agree with Mike, this simply won't work.
R. Cordell used active loading of differential stage, but with bias feedback, so that voltage is defined. I also saw something like this in a class D amplifier (by classD4sure ?). Check them out.

http://www.diyaudio.com/forums/showt...746#post952746
http://www.cordellaudio.com/papers/m...rrection.shtml

[lineup]

Originally posted by viktor1986
Hi

What's your opinion about this amplifier?

What is good with Lateral MOS, like 2SK1058 / 2SJ162,
is they are easy to parallel, compared to IRF HEXFET and Bipolar power transistors.

What I have read, you don't even have to use any emitter resistors
when parallel several www.renesas.com Laterals ( former Hitachi line ).
I don't think this is true for all lateral MOS.
There are several variants of laterals with different thermal behavior.

In your case you use 0.1 Ohm.
It wont hurt to use emittor resistors anyway
and also makes it much easier to set and measure the Idle Bias Current.

I think 2SK1058 has got considerably HIGHER RDS-on than most power HEXFET.

This also means that they will start clipping maybe already when
Volt Drain-Source is 4-6 volt.


So, when using Lateral MOS instead of vertical IRF HEXFET
you need maybe 3-6 volt higher voltage supply both in V+ V- half.
This makes a total of 6-12 VDC
and so transformer needs to be 4-8 VAC higher.

For example one amp IRF240 / IRF9240 uses 2x30 VAC. ( 60 VAC )
If using 2SK1058 / 2SJ162, maybe need 2X32-2x34 VAC. ( 64-68 VAC )
For same power output.


lineup