[Prune]

It would be interesting to see a design for a Stax amp that uses Pass' supersymmetry idea (as discussed in this patent and at diyaudio; note that the folded cascode mentioned in the abstract is a red herring; the main thing is the "feeding its distortion and noise contribution positively, via input gain transistor conduction, to the other stage").
Or perhaps an existing design could be changed.
How about a KGSS/BlueHawai modified with supersymmetry?

 

[kevin gilmore]

By definition if you are using my amp in fully balanced mode
it is already super symetry as defined by nelson pass.

However, please pay attention to the necessary voltage
levels required. +/- 400 volts. 800 volt swing. 900 volts
or more to be safe.

900 volt N channel fets have large gate capacitance
and sound like crap. Great for switching power supplies
and little else.
P channel versions at 800 volts DO NOT EXIST.

There is only one pnp transistor available from anywhere
that is 900 volts and it is the one i'm using.

There are only 2 npn transistors rated at 900 volts and
i'm already using the best sounding one.

Stack lower voltage fets?? That is what the koss amp does.
Absolutely one of the most horrid sounding things i have
ever heard.

High voltage n channel fet into a resistor current source. That is
what the sennheiser hev60 does. Another horrid sounding
item.

 

[Prune]

Quote:

Originally posted by kevin gilmore By definition if you are using my amp in fully balanced mode it is already super symetry

Thanks, but can you elaborate a bit? I'm just trying to understand here, excuse a newbie...

In SUSY, while feedback carries the signal in inverse phase, distortion/noise is fed in positive phase through "input gain transistor conduction" to the other side (in some circuits feedback is cross-connected depending on the inverting/non-inverting stages to get phase correctly, but that's besides the point), and with the same error signal present on both outputs, it cancels out. I don't see how this works here. In the SUSY circuits at diyaudio feedback went to the gates of the input differential pair, and the input transistor sources are coupled by either a very small value resistor or just a wire so the error signal has a low impedance path to the other side.

 

[Petter]

Kevin,

Prune's question is precise in reference to Nelson Pass' patent on "SuperSymmetry" which rely on feedback around a single folded cascode stage.

I see a lot of clever stuff in your designs, but I certainly cannot detect any "SuperSymmetry" in this one. Perhaps you are not familiar with Nelson Pass' work in this area and understood the question to be "balanced circuitry".

While I am at it, let me thank you for getting me to sell my tubed Stax unit and start building my own. Your original articles were certainly extremely helpful and motivational

My $.02

Petter

 

[kevin gilmore]

quote
I see a lot of clever stuff in your designs, but I certainly cannot detect any "SuperSymmetry" in this one. Perhaps you are not familiar with Nelson Pass' work in this area and understood the question to be "balanced circuitry".

I'm familiar with the super symmetry design. In fact my
uberamp uses this identical front end feedback arrangement.
Whether it is a fet or a bipolar transistor also does not
make a difference.

http://gilmore.chem.northwestern.edu/uberpage1.jpg

I cannot see the difference between returning an inverted
version of the output signal to the gate of the fet as opposed
to a non inverted version returning to the source of the
fet. They are absolutely identical. The fet only sees the
difference between the two signals.

Now pay attention to this one more time.
You cannot get a voltage amplifier with a gain of 1000 with
only 2 fets, even if you could find ones that run at the correct
voltage.

Power amps like the super symmetry design have a closed
loop gain of less than 50.

Amplifiers that drive 8 ohm speakers are very different
from amplifiers that drive electrostatic devices with impedances
in the 10's of megohms.

If you think you can design a better amplifier, then by all
means go ahead and try. I've been doing this for the better
part of 8 years, and the blue hawaii is the best of the many
designs i have done. I'm actually tired of being the only
one around here designing electrostatic amps.

 

[Prune]

Quote:

Originally posted by kevin gilmore I cannot see the difference between returning an inverted version of the output signal to the gate of the fet as opposed to a non inverted version returning to the source of the fet. They are absolutely identical. The fet only sees the difference between the two signals.

While feedback is necessary for SUSY, it's not the point; it just provides a path for what Pass calls the errorsignal (as separate from the distortion that feedback corrects). The point is that the sources of the input FETs are connected together, and in most designs there is not even a resistor between them but just a straight wire (for example the Aleph X); a low value resistor is in some circuits with multiple stages, according to Pass, to prevent a hall-of-mirrors effect. The errorsignal travels through the feedback connection to the input, but then is isolated from the feedback and presents itself at the sources connection which allow it to go to the other side, and thus ends up duplicated in phase on both + and - outputs; and of course commond mode signal in balanced output will not end up in the sound.
Quote:

If you think you can design a better amplifier, then by all means go ahead and try. I've been doing this for the better part of 8 years, and the blue hawaii is the best of the many designs i have done.

Mr. Gilmore, I doubt anyone here thinks they can design a better amplifier. I hope you didn't misinterpret my questions as a challenge, because it certainly is not, and I share the great respect for your designs that the other head-fi members have.

 

[Petter]

Quote:

Originally posted by kevin gilmore quote Whether it is a fet or a bipolar transistor also does not make a difference.

Indeed!

Quote:

Originally posted by kevin gilmore quote Amplifiers that drive 8 ohm speakers are very different from amplifiers that drive electrostatic devices with impedances in the 10's of megohms.

Also very true. My own calculations indicate that a typical Stax headphone requires about 15mA of drive at the highest frequencies in the worst of cases.

Quote:

Originally posted by kevin gilmore quote If you think you can design a better amplifier, then by all means go ahead and try. I've been doing this for the better part of 8 years, and the blue hawaii is the best of the many designs i have done. I'm actually tired of being the only one around here designing electrostatic amps.

Well, if it helps, I am also into electrostatic designs. I do not pretend that I have the experience that you do and apologize if my earlier post indicated lack of respect - it would be hard NOT to give you respect for your work - especially since it was what got me started in the first place.

I will of course check out the uber-amp.

Petter

 

[kevin gilmore]

quote

While feedback is necessary for SUSY, it's not the point; it just provides a path for what Pass calls the errorsignal (as separate from the distortion that feedback corrects). The point is that the sources of the input FETs are connected together, and in most designs there is not even a resistor between them but just a straight wire (for example the Aleph X);

The schematic you list has no such straight wire. Furthermore
this circuit only really works well with a real balanced input.

How nelson decides to describe his circuits is up to him, but
feedback is feedback is feedback, and an error signal is
an error signal, and includes all non-linearities, phasing
and other distortions. I'm certainly not going to pick on nelson,
at one time or another in the past 25 years he has designed
and produced at least one version of every amplifier design
ever built.

Fact is that this is what i call a cross-coupled feedback circuit.
Exactly like my uberamp except i'm using all bipolars, and
nelson likes to use fets. There is no difference.

You could take either of my current electrostatic amps, change
the feeback from the sources of the fets to the gates of
the fets and at the same time flip the feedback from the
two outputs. Then insert a 1k resistor in the input line, making
absolutely sure that the input side of the resistor is also
terminated by 10k at all times. Otherwise on turn on the
high voltages blow up the input fet. What you have done in the
process is changed the input impedance of the amp to
about 5 to 6 k which pretty much eliminates the possibility
of an input attenuator. Other than that absolutely nothing
has changed.

 

[Prune]

Quote:

Originally posted by kevin gilmore The schematic you list has no such straight wire.

Sure there is! Q5 and Q7 are directly connected to each other (at their connection with Q6).
Quote:

this is what i call a cross-coupled feedback circuit

Well, Pass says there is something more than the cross-coupled feedback. From this diyaudio thread:
Quote:

Originally posted by alaskanaudio The cross-coupled feed back from one side to the other is typical balanced audio line drivers. Nothing new there.

Quote:

Originally posted by Nelson Pass In point of fact there is something new here, and you seem to have missed the essence of it.

This simple example doesn't even cross the feedback (already inverted).

 

[kevin gilmore]

quote
Sure there is! Q5 and Q7 are directly connected to each other (at their connection with Q6).

This is a standard differential amplifer without emitter degeneration
(or in this case source) resistors. Q6 is the current source
for the differential amplifier.

Just like my front end except up side down, and i use a
transistor as the current source. NO DIFFERENCE...

emitter degeneration resistors are used to lower the gain.
Same as a single resistor could or would be.

Replace the fets with tubes (and turn everything upside down)
and you have a circuit that goes back to early 1950.

Nelson can put any spin on his design that he wants to, and
just because he has a patent on it does not mean it works
the way he said it does.

 

[Prune]

OK you win!