Don Pearce wrote:

On Thu, 13 Jan 2005 14:35:24 +0200, "Iain M Churches"
wrote:


"Don Pearce" wrote in message
...

I've been the minimalist route for amp design, and the problem is that
it doesn't stay that way. You always end up thinking "if I just do
this, it will work SO much better" and by the time you are done, you
have put in all the buffered current mirrors etc that really make it
work. And in the end, why not? The bits cost bugger all.

The cost is not really a factor. The idea was to design and build the
SS equivalent of Andre's simple SET amplifier, so that those
interested could build both for listening comparison.

Iain

I understand that, and it could be an interesting experiment -
although the very different transfer characteristics of a BJT and a
triode would make it of limited value. And of course in SS it is so
much more trivial a task to do all the good stuff.

If only you could get valves that worked on positrons!

I wouldn't buy tubes that worked by emitting positrons....

Its hard to conjure up such a device, and just because that would enable
some of the combinations routinely used with NPN and PNP
SS devices may not bring more music to anyone's ears.

I think it will be quite some time before anyone comes up with an
8 watt two transistor single ended amp which has no global NFB
and which sounds as well as a 300B amp when correctly set up.

Stewart Pinkerton and I discussed the prospects of a kiSS amp
at RAT some weeks back, and so far, no SS prototype has arrived on the
doorstep of our
consciousness from Mr Pinky.

There are several problems with BJTs ensuring that simplicity is almost
impossible when one uses them.
Its impossible to use them without NFB, since the collector resistance is so
much higher
than a speaker's, like an SE pentode amp.
OK, so lets say we allow the BJT amp to use the same amount of NFB as the
internal NFB
which exists in a 300B and then proceed.
This rules out the use of setting up the output trannie as an emitter
follower, since that is a case
of a vast amount of series voltage NFB.
And input voltage must be below 1vrms.
The low base input resistance of the bjt can spoil the simple is best
idea.....
Even if you set up a single MJE21194 or 93 as an EF , the Iin will be
emitter output current / current gain, so for 1 amp of speaker current, and
8v into 8 ohms,
and if current gain is 30, then you need to apply about 8.1 v at the base
at 33 mA, so Rin = 242 ohms, and needing that other driver transistor
If the driver transistor was also in EF, and had a current gain of 100,
then Rin is still a lousy low 24k ohms.
Methinks any other form of applied loop FB to get Vin 1v, Ro 1 ohm,
and thd 1% will still result with Rin as too low.

Simple SE SS amps were routinely used in auto radios of the 60s and 70s,
and often had a few driver transistors or an opamp to power a single BJT
in class A with a transformer coupled speaker, all running off the
12 volt car supply which allows a swing of 7 vrms, giving 12 watts into 4
ohms
if the idle current is high enough.
All used lots of NFB, and were not as simple as an SET 300B amp
when considered in terms of electronic elements used.
I once repaired such a radio from an Aston Martin.
The output tranny was a large germanium type in SE class A.


But why muck around using BJT's?

They are OK in complex amps working in nearly class B circuits, and their
cheapness is why they are used.

Mosfets provide a far better solution because they have a high gate input
impedance,
and The Nelson Pass Zen amp using just a single mosfet as the output and gain
element
with a CCS instead of an OPT with some 12 dB of shunt voltage NFB is
comparable with
a 300B amp.

People may complain that mosfets are horribly non linear.
But when used in the middle of their transfer range and with some NFB
they are quite adequately linear for a few watts, and most of the thd is
a combination of mainly 2H and 3H, rather like a pentode.
Efficiency is about 45% with SE, so for 8 watts output
you only need about 19 watts of idle power, much less than a triode amp
with 25% efficiency..
So its comparable with the triode amp to begin with and Rin is quite high.
Cin is also high, but not high enough to stop the use of a single j-fet
driver device such as the
2SK369 from being used very effectively, and in a circuit like the cap
coupled and
transformer output coupled tube amp.

BJTs have appalling voltage linearity when set up in common
emitter with the load in the collector circuit and the emitter grounded.
They have a lot of voltage gain, but linearity is worse than a pentode with
no NFB.


One thing seems certain to me, it is easier to make a good sounding tube amp
than a good sounding solid state amp.

The knockers will maintain that the tube amp measurements are always going
to be **** compared to the 0.0001% available from an SS amp,
but so what?

A class A PP williamson amp designed and built to the original
spec of Mr DTN Williamson in 1947 with a pair of KT66
will make 16 watts at a max of 0.1 % thd with 20 dB of series voltage NFB.
One can remove that NFB and still only get 1% thd.
There are 4 input triodes and two outputs.
School boys found such amps easy to assemble and get working.

At a few watts, thd is 0.02% with NFB and perhaps 0.2% without.
I'd suggest that 0.02% of thd at 2 watts of signal level is inaudible, so
the reduction of thd down 200 times to 0.0001% as Halcro claim to achieve
cannot make much difference.


Then if we were to consider the use of KT88 in lieu of KT66, but with exactly

the same circuit, we'd get nearly twice the power.
The Quad 40 is an example of a simple PP tube circuit using 2 x KT88,
and two signal pentode driver tubes, and although pentodes have a reputation
for
more thd than triodes, the Quad 40 and the preceding Quad II amps
have only 0.025% thd at a couple of watts when all the tubes are matched
fairly close, and arranged more maximal 2H cancellation.

I recently built a high current BJT 300 watt/channel stereo amp using 5 x
MJE21193, and 5 x MJE21194
in the outputs, and with a shirtload of NFB so thd is negligible at all
levels and
its just does not sound as smooth and non fatiguing as many tube amps.
There are not too many school boys who could manage to build and get such a
thing working.
Its not as good as the same power mosfet amps I have built.
It ought to be great, and I have followed all the rules Douglas Self and
others recommend,
even with a choke filtered collector rails, but methinks its only good for
sub-woofer
power.
The crossover distortion with a hefty idle current is utterly negligible
at audio F and the dominant thd is the non linearity of the main voltage
swings.
Still, it measures very well since a total of over 100 dB od NFB is used.
I didn't like the way the cross conduction during clipping at just above the
audio
band caused enormous currents between rails.
I tamed all that as best I could, but mosfets don't have all these ****ty
problems,
so the drive circuit and the mosfet PP complementary pair outputs is a lot
simpler than that
needed for the BJT amp.

Give me tubes or fets, and to hell with the rest.

Patrick Turner.



Pearce Consulting
http://www.pearce.uk.com