Patrick Turner wrote:
Below Pinkerton bull****s to everyone about how to make a simple SS amp.

He did it last year too.

He's repeating the same mistake.

He didn't make or test the amp he tried to foist onto us. Nobody here afaik tried to
build what he suggested;
we wouldn't, since we prefer tubes.

I even said he could use 12dB of NFB because that is how much is in a 300B by my
estimate.
He didn't even check that out of course because he has not the slightest idea how to
establish how much NFB there is in any triode.

But his SS design used a huge total amount of NFB. But he can't cheat while I'm
around.

How come Stewart Oinkerton, the class A pork seller can't build a simple amp?

More class G for gross, I think

Nelson Pass beat Oinkerton in this race of course.

Sue Parker also has beaten him.

Patrick Turner

I couldn't care less about "beating" Pinkerton. I thought we were here
to exchange information to everyone's mutual benefit. I don't even see
what Pinkerton posts until someone replies and most of the time I just
skip those posts.

Less time spent on useless clowns leaves more time for building amps.

Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"wonderfully well written and reasoned information
for the tube audio constructor"
John Broskie TubeCAD & GlassWare
"an unbelievably comprehensive web site
containing vital gems of wisdom"
Stuart Perry Hi-Fi News & Record Review

Stewart Pinkerton wrote:

On Wed, 29 Mar 2006 15:32:14 +0200, "Ruud Broens"
wrote:


"Stewart Pinkerton" wrote in message
.. .
: On Tue, 28 Mar 2006 23:07:46 +0200, "Ruud Broens"
: wrote:
:
:
: : Bjts are so damned non linear with regard to voltage gain that large amount
of
: : NFB, local or otherwises must be applied
: : to linearize the outcome, as well as raise input impedance.
: :
: : Some BJTs are more linear than say a 300B. Why do you keep ignoring
: : this *fact*. Heresy?
: :
: : It isn't ludicous to drag people screaming back to the basics.
: : The basic fact is that bjts WITHOUT ANY NFB, LOCAL OR OTHERWISE applied
: anywhere
: : have hopelessly poor
: : voltage gain linearity.
: :
: : Utter bunk, check the spec sheet for the 2SA1216. Gain linearity is
: : flat as a pancake at elevated junction temperatures, as would be
: : experienced in a Class A amplifier.
:
:
: yep, checking, i found it he
: http://users.otenet.gr/~athsam/database.htm
:
: the third graph shows Vbe - Ic, the _run hot_ curve
: does 0.8 V Vbe @ 6 A, 400 mVtt giving 8 Att with 1.6 % distortion
: certainly impressive. but, now realize the curves are taken at a
: constant Vce=4V - not very realistic, is it ?
:
: now, Hfe about 100, at this 24W class A idle point,
: 60 mA of base current is required. oops, at 10 A, Hfe has
: gone south to only 60, creating an additional few dozens
: of THD %.
:
: 2nd take:
: 0.6Vbe, 2.2 A idle now Hfe stays flat with 200 mVtt in, 4Att out,
: distortion is 7 % or so that's indeed quite an achievement
:
: but considering it's taken at constant Vce
: reality will be worse
:
: : Statements have been made on the group that what i am saying is untrue.
: :
: : That's because it is - see above.
:
: hm, about 12 W / 8 Ohm available, 7 % THD, 8.8 W idle with the 2SA1216
: you're saying there are no tubes that can beat that ?
:
: I read less than that into those curves, but even if true, that's
: better than you'll get with the 300B.
:
: You are of course conveniently forgetting something else, the greatly
: superior gain of the 2SA216. In the KISASS design, Tr2, the equivalent
: of the 300B in KISS, has its emitter and collector loads set for a
: stage gain of 6.25, to make it as similar as possible to the KISS amp.
: This applies more than 20dB of local regeneration, equivalent to the
: internal feedback of the 300B, cutting the distortion to about 0.5%.
: Apples to apples, KISASS totally outperforms KISS.
:
: --
:
: Stewart Pinkerton | Music is Art - Audio is Engineering

not forgetting anything here, just sticking with the topic - again -
being inherent device linearity, not what you can do with fb - remember ?

Yes, but PT already conceded that the 300B has internal feedback, so
he had no problem with local degeneration in KISASS. After all,
putting in a larger emitter resistor doesn't create any of the
supposed horrors of loop feedback. Note also that you already conceded
that it's as good or better than a 300B, without *any* feedback.

it seems you didn't get the impossibility to create 12W out with 8.8 W idle
class A, the '-4V Vce from the graph' * 2.2 A idle.

Sure I got it, but obviously KISASS doesn't use a 4 volt rail.....

to make it work for an 8 Ohm, 12 W output, you'll actually need about
40V supply voltage and a current sink pulling 2.2A, together using 88W
to get that 12W out - 15 % efficiency and a large 1K/W heatsink for
that 2SA1216 and say an LM350.

See KISASS, if you can't remember the basic parameters of KISS.
KISASS runs a 30 volt rail at the voltage amplifier, and is designed
to run 100 mA idle current in the voltage amplifier stage.

To reiterate the design process:

KISASS came about as a suggestion by Iain Churches that I design a SS
equivalent of Andre Jute's (so far mythical) KISS design, supposedly a
single-ended tube amp with 300B output tube, a double-triode
input/driver stage, zero loop feedback, and an output of less than ten
watts. Ignoring a bunch of tube-centric 'rules' spouted by Patrick
Turner, I had a think about how this should be done.

In terms of parts count, we cannot beat the Nelson Pass ZEN design,
which uses a single MOSFET, so we will use BJTs, the 'bete noir' of
the tube fan. In terms of excellent performance from a minimal parts
count, we have the classic 1969 Linsley Hood design, which ticks most
of the boxes but does use global NFB.

Where to go from here? OK, let's start from the simplest possible gain
stage, a common-emitter BJT with emitter and collector loads (Tr2 in
the schematic, the 2SC2922 device is recommended). This will set the
transfer curve, and hence the sonic signature, of the amplifier. Such
a stage does of course have a quite low input impedance, so we place a
simple emitter follower buffer (Tr1, the MJE15028 will do, or any
other decent medium power NPN BJT) ahead of it to allow a wider choice
of source devices. The output impedance (as with the 300B) is much too
high to drive a loudspeaker, so we need an impedance transformer. As
one of the great advances of SS was that it allowed the removal of the
OPT, we will not use iron. A simple pair of emitter followers (Tr3 and
Tr4, for which the 2SC2922 and 2SA1216 will work well, but other
modern complementary pairs may substitute) provide output impedance
reduction without affecting voltage gain, and there we have the design
in a nutshell.

Further philosophical decisions were the avoidance of bootstrapping
from the output to the gain stage (that could be construed as loop
feedback), and the use of a heavily filtered power supply to keep the
noise floor clean. The latter was in the cause of maintaining the
'first watt is the most important' philosophy, without which this
design is utterly pointless. R15 and R16 reduce the rail voltage to
give a maximum output power of less than ten watts into loads between
8 and 4 ohms, meeting one of the principal design parameters, while
R12, R13 and R14 maintain the rail voltage of the gain stage at 5
volts higher than the output rail, allowing better driving of the
output while maintaining the natural transfer function of the gain
stage.

The bias current of the output stage is set to about 1 amp by R6,
while the quiescent output voltage point is set primarily by R1/R2.
This amplifier should be d.c. and thermally stable, but will reward
careful layout, particularly in the utilisation of the 'star earth'
philosophy, i.e. one central ground point to which all 'zero volt'
connections are made. Given this, I'd expect less than 0.1% THD plus
noise right across the audio band at an output of 1 watt into 4-8
ohms, rising to several per cent just below clipping, in the usual
'SET' style. The output impedance will also be in the half-ohm region
over most of the audio band.

In terms of construction, it should be noted that R15 and R16
dissipate 5 watts each, so should be at least ten-watt rated, while R5
dissipates about 3 watts, and should be at least a 5-watt component.
R12, R13 and R14 should be 2-watt rated, the others should be
half-watt metal films. Note that for thermal stability, D1, D2, Tr3
and Tr4 must share the same heatsink.

you are also conveniently forgetting that - as is -, you have an input
impedance of about 5 Ohms ...

See KISASS for the solution. Tubes use lotsa volts, BJTs use lotsa
milliamps, same thing in the end for the 'black box' solution.

more of a win some - loose some scenario, it seems,
Stewart ;-)

Not really, as KISS suffers similar problems. Naturally, one wouldn't
expect a tubie to acknowledge this.

T reiterate
--

Stewart Pinkerton | Music is Art - Audio is Engineering