K10amkii

[GK]

Something I conjured up last week.

10W class A into 4 ohms / ~20W class AB into 8 ohms.
THAT340 quad transistor array input stage.
2SK1058 / 2SJ162 complementary MOSFET output stage.

Fast slewrate / wide bandwidth full power bandwidth (~1MHz but then limited by input RFI filter).
VAS clamp designed so that no transistor enters saturation when input overdriven, giving exceptionally clean and quick clipping recovery.

The input stage and VAS has very low distortion (~1ppm 20Hz-20kHz in SIM) meaning that overall THD and harmonic spectra for all sakes and purposes is entirely governed by the lateral MOSFET output devices (ie low order).

Compared to BJT (especially RET’s) MOSFETs are not particularly linear. Ultimate THD performance will depend on how far the unity loop gain frequency can be pushed out whilst retaining a satisfactory phase margin.

Simulations posted in following posts performed with a conservative unity loop gain frequency and 80 degree phase margin, giving:

THD-20kHz ~0.01%
THD-1kHz <.001%

Who’d be up for a pair of double sided, ground plane PCBs?

Cheers,
Glen

[lineup]

I am glad to see at least someone
that is not Stupid Enough
to feed input and vas stages from the heavily current modulated output stage supply

such craziness
but even some supposedly intellligent old gurus at this forum
is still into such bad thinking

Probably learnt from see commercial massproduced designs
where they save a few bucks
from not having a small separate transistor for input/vas.
Mokee See ->>> Monkee Do

Selling 1000 amplifiers x 10 euros = 10000 euros higher profit.
One diy amplifier supplying everything via output supply
will make 10 euros money gain.

A Trade off for what?
For the loss in PSRR and performance will be the price you pay
for having only 1 trafo
(Not seen at all in Spice = ideal voltage source + ideal rails, but in reality)

As you say, GKlein
the improved dynamic headroom of voltage variations (margin above saturation) for vas transistors
is another benefit, besides avoiding all interference from Output currents.

I really do not know what our gurus think with
when deciding for only one supply for power amplifiers.
The Shining exception is John Curl, who understands these things

Lineup

[GK]

20kHz sinewave clipping. And before anyone asks, the Vbe multiplier is not mounted on the heatsink.

[GK]

500kHz squarewave.

[nikwal]

This is good, a little too many components for my taste but thery're all doing something. double buffers for the mosfets hehe yeah why not.. 20watt seems low, this would'nt be too hard to crank up a bit I guess..
I can see that c5 works nice as it's connected, I did that too on some amp, is it really that bad to have it parallel with r15, I kind of wanted to do it like in your schematic on my amp but it ended up in parallell anyway for some reason,when I was trying to get it to work with a capacitor load without output coil, hmm maybe I'll have to go back and try again..
I sim I strive for 60degrees margin,read that should be good somewhere, not sure why though..
(EDit : offtopic: Haay , nice landrover, I'm working on a supra myself, it's never finished, but it's getting closer)

[GK]

Hi

I can’t comment on exactly why your amp wasn’t stable with the feedback cap from the VAS collector. The cap has a significantly different effect on the loop gain and phase margin when connected in parallel with the feedback resistor. I try to avoid adding a cap in parallel with the feedback resistor as it forms a direct short to the inverting input for any RFI picked up by the speaker leads.
With and 80 degree phase margin the amp is overcompensated, but I’m just being conservative with the simulations so not to overestimate the amplifiers performance.

10W/20W is not a lot, but can be plenty with efficient speakers and the class A bias dissipation is manageable – also good for bi/tri amping. Here you can have your class A linearity and still make plently of noise in the average listening room.

I do the majority of my listening on amps of this power level, so I like to play with designs of this power level that perform better than average.
Very few take low power amplifier design seriously. I think this is silly.
Even less seem to be interested in the prospect of such as design being made readilly available

Oh well.


On the supra – cool . I’ve got a few other car projects on the go which will be finished one day also. All old school stuff though. Been thinking for a while about putting funds aside for a Series II Lotus Elise, since being taken for a ride in one. Far too underpowered for my liking, but there is a plethora of Japanese import power plants with huge tuning potential to choose from that would be a relatively straight forward engineering fit. One of these little plastic go karts with 250-300kW at the treads would annihilate just about anything, and could be build up on a pretty reasonable budget.

Cheers,
Glen

Oh, and the 20kHz FFT:

[MJL21193]

Originally posted by lineup
I am glad to see at least someone
that is not Stupid Enough
to feed input and vas stages from the heavily current modulated output stage supply

such craziness

Hi Lineup,
Do you really see this as being a significant problem? When you consider the average listening level will be very low power, there will not be much modulation of a well made power supply. If your PS is capable of 300 watts and only 4 are being used??? No need for that second transformer/PS.
You could (and some do) take this attitude to the electrical substation - with all of those different houses and buildings, street lights, etc modulating the power supply before it gets to your house. What about all of the draws within the house itself? Fridge cuts in and out, electric heaters, water heaters, clothes dryer, washing machine, etc, etc, modulating the power supply.
FRACK! We need our own electricity generating plant to cleanly power our audio equipment! Anyone have a fundamental understanding of cold fusion yet?



Glen! Nice design - too many Q's.
I suggest bootstrap.

[syn08]

Originally posted by G.Kleinschmidt
Something I conjured up last week.

10W class A into 4 ohms / ~20W class AB into 8 ohms.
THAT340 quad transistor array input stage.
2SK1058 / 2SJ162 complementary MOSFET output stage.

Diamond buffer drivers? You are certainly looking for trouble...

Nice design, BTW.

[atiq19]

Glen,

Do you have any update on this nice design?

Atiq

[GK]

Originally posted by atiq19
Glen,

Do you have any update on this nice design?

Atiq

Hi Atiq.

Unfortunately it never got beyond LTspice due to lack of interest combined with there being more than enough on my plate already.

Cheers,
Glen