Arny Krueger wrote:

"Jim Lesurf" wrote in message


Well, I don't know the design of SET amp you refer to, so can't
comment on it specifically. However when I had a student build a
transistor SET amp some years ago the expensive and difficult part
was the o/p transformer. This is a bit easier in some ways than for
valve as it didn't have to isolate very large voltages, and didn't
have a large turns ratio. But you still end up requiring a
significant (expensive) transformer if you want it to work OK even
down to LF.

I don't know if I still have a copy of his project report. When I get
a chance I'll have a look for it. My (perhaps unreliable)
recollection is that the performance was limited by the transformer
not by the choice of gain devices. Hence such theoretical
'simplicity' can become quite complex and costly to attain. Looks
simple on a circuit schematic, though. :-)

Yes, that is a big concern of mine as well.

One of the nice things about push-pull amps is the flux cancellation in the
output transformer. Of course no output transformer is usually even better.

The OPT for a single device class A mosfet or BJT
amp can be very easy to wind because instead of say 2,000 primary
turns of fine wire used for an SET amp, only about 200 turns of much thicker
and easier to
handle wire are needed.
The OPT will have to be the same weight as that for an SET, and the secondary
will have to be the same as for the tube amp since the turns per volt
will be the same.
A good diyer should be able to make a suitable OPT if he knows his theory,
but most have little idea.

But for 8 watts, a single pair of output devices in an complementary pair work
fine,
and if 2 mosfets are used in source follower then the thd at max po will be
about 1%
and acceptable at a watt, and all that's needed is a linear driver stage
capable of producing
about 9vrms at say 0.3% thd so it don't add much to the two output devices.

With a 6 ohm load, such a pair share the load between them and each device sees

12 ohms, so their open loop gain is about 10, and the amount of NFB applied via
the source follower
connection is about 20 dB with a 6 ohm load.

The driver device could be a high gain signal bjt with emitter current FB
and perhaps cap coupled to the fet follower.

But because you don't have triodes, don't expect the sound to be as acceptable.

The input device would perhaps better be a high Gm j-fet like 2SK369.
Load could be in the drain circuit, and global NFB used like it has to be in
pentode amps
to straighten out what is the open loop mess.
The beauty of the common source connection of mosfets is that
only about a volt rms of drive is needed to the gates, very doable
with a drive fet I have nominated, and the amount of
global NFB can be set up to the minimum amount to get
Ro = 1 ohm.
All devices are then included in the NFB loop.
The mosfet gate capacitance is high in common source mode, but low voltage
drive is needed.
In common drain, C is low, so the same drive can still swing the reqired
voltage either way
because the current in the driver device will be sufficient.

One can also connect the output mosfets as a totem pole connected pair of the
same
NPN or PNP polarity, and set them up as a mu-follower, or SRPP
so liked by tube orientated people for preamp stages.
This way there is a mix of SE and PP action, low drive, slightly
more 2H, and very easy to set up.

But BJTs?, nah, leave em for the boyos who
like to have 39 devices in an amp, and think all that junk is "elegant".

Patrick Turner.