Zero Feedback Impedance Amplifiers

[mastertech]

erictoucan-
But, 0.1% crossover distortion could be very audible while 0.1% distortion in magnitude is definately inaudible.
------------------------------------------------------------
lets keep things in perspective above a certain level each kind of
distortion becomes audible, if you gona say otherwise show the
evidence and the scientific data to prove it

"0.1% crossover distortion could be very audible "

please show us what you mean

Reagards

[Susan-Parker]

Hi,

Originally posted by mastertech
Have these amps been reviewed, how do they sound?they look
more sophisticated then np's zen, are they better?has anyone
listened to them to tell us what he liked about them

See post #570 of this thread

#570 Review

Hope this helps.

Best wishes,
Susan.

[Susan-Parker]

Hi Lineup,

Originally posted by lineup
To use transformers and/or coils in input and/or output of amplifiers
may to me to be asking for distortion.
What I know, magnetic fields are not behaving very linear.
It is certainly not like a resistor.

There is a lot of non-linearities in the whole audio chain - especially in the ear where the mechanisms of hearing are complex and active.

With the relative recent resurgence of tube amplifiers (on a large(ish) scale) vanishingly small THD figures are now more accepted as not being the be all and end all of hifi performance.

The follower configuration I use in my amplifiers drives the transformer from a lower than normal impedance source which is the key to reducing transformer distortion/non-linearities.

But we still use to pass signal through coils and magnetic fields
in our loudspeakers.

I started with a clean sheet with my amplifier development and asked myself the initial question "what is a loudspeaker, how does it work"? Really thinking about the basics and trying not to let any assumptions that I might have (which may be correct or erroneous) cloud my thoughts.

Logically it wouldnt be more bad to use coils in amplifiers.
And this is no problem for all them tube builders.

Most tube builders have to use transformers of necessity! I have based my amplifier design around the premise that the transformer is the primary component from which everything else evolves.

For me, it just takes some time to adjust my thinking
and try to accept this way to achieve a good amplifier topology.

No pressure. It took me several years to develop my amplifier.

And behavior, mathematics and wiring of coils/transformators
is not quite as simple for us normal people.

Um, one of the things that I was looking for in making the transformer based amplifiers was a design that could be made by "normal people" such as myself.

Maths is not my strong point, and I still have not got a satisfactory transformer model for Spice simulation...

... and winding multi filar transformer bobbins is more a matter of patience than any innate skill (honest).

Best wishes,
Susan.

[Susan-Parker]

Hi Rozak,

Originally posted by rozak
My transformer coupled amps sounds incomparable better than my usual SS amps.

As per the schematics and pictures on your thread:

My new (old) way

I assume?

Of note is that we both have four transformers and a PP mosfet follower 2:1 output stage referenced to ground.

I would imagine that the sound is similar although I perhaps have a higher second harmonic in relation to the third with the SE input stage (so only one TX-102 is needed per channel, rather than two) particularly in the tube version?

You have had some time to "live" with the sound of the system and from your comment I take it that you are pleased with what you have.

Best wishes,
Susan.

[rozak]

Yes, Susan,

sound of my system is very good, distortionless for ears at least,
big and powerful. 25W sounds like 250W.
Because of my poor English I can't describe it better.
Only one coment from frends and visitors-WOW!

Transformer topology does something to sound!

It take very long burn-in period (400-500 hours) with ups and down to reach best quality.
Any burn-in story of your system?

[angel]

Hello, Susan.

I note that you have had time to try the Exicon VFETs, but with rather poor results. I'm assuming you used the ECF10N20 or ECF10P20.

Could you post some details about this?

I'm assuming you've checked for oscillations; these devices are nearly as twitchy in that regard as DMOS devices, if I recall correctly. In fact, you'll probably want a source resistor between the FET and the output transformer to improve stability, as well as a zobel (either on the FET side or the speaker side).

According to the public datasheet, these devices have 500pF/700pF Ciss, 300pF/300pF Coss and 10pF/25pF, with a Vds(SAT) of 12V and 0.7S<Yfs<2S.

However, blatantly missing from the datasheet are a few important stats, such as Yos, Igss and Rds(on).

I expect Yos is probably on the order of 350-700mS<Yoss<200-400mS, while Igss is probably on the order of 1-100mA and Rds(on) is presumably at least .5<RdsON<1.4 at low output currents (more at higher currents).

Hence, the output transformer probably needs to step them down a bit to get the same damping factor as for the other FETs you've tried (just do the regular quadfilar winding, but section it and try various versions of series and paralell to see what sounds best). The drain supply would also benefit from 6-12V more headroom than the other FETs. I suspect you need to use a different value gate stopper too.

I also expect you'll need a gate buffer or low-impedance source; the gate resistance of a VFET is far lower than for a MOSFET (on the order of ohms, according to the application note).

I know you resist the idea of a second semiconductor, but a complementary JFET buffer adds 1-100ppm THD+N, with nothing over 3rd order. A single-ended CCS-loaded JFET buffer adds marginally more.

For single-ended, Linear Systems LS843 (Dual N-JFET with 1mV and 5uV/C max Vgs difference; 1.5mA-15mA), Vishay 2N5911 (Dual N-JFET with 5mV and 5uV/C max Vgs difference; 7-40mA) and Fairchild J500 (Single N-JFET with minimum 500mA Idss) spring to mind.

For complementary, 2SK146V/2SJ73V (Obsolete dual N-JFET with 20mV max Vgs difference; 15-30mA - no complete datasheet) or 2SK364V/2SJ104V (Current production Toshiba single N-JFET with 10-20mA Idss; try a Vdg of 10V at 12mA - for <10pA Igss and very linear operation with zero tempco)...

I'll try to show an idea for a circuit to do this... simpler works fine, too... Just don't drive it with a GaAsFET or heterojunction

Sample driver, one half shown (done before bedtime, so don't shoot me ;)...
                       ===
                        |
                        |
                        |
                     |--+
                     |    Q3a
              +--+-->|--+------O Vss
              |  |      |
  ===         C4 V      +--|
   |          |  |  Q1a    |
Q5 +--|       +--+------+--|<--+--R--+                / / /
      |                 |      |     |                 --+--
+--+--|<--+             +--C5--+     |                   |
|  |      |             |            |                   |
L1 R1     R             R4a   +------|--C3-----+         |
|  |      |             |     |      |         |         |
+--+------+--R2--+--R2--+-----+--R6--+--+--R3--+--R3--+  |
|  |      |      |      |            |  |             |  |
C1 +------|--C2--+      R5a          |  C3      +--+--+  | +--O HOT
|  |      |             |            |  |       |  |  |  | |
|  (      |             +--C5--+     |  |       |  |  )*||*(
|  (      |             |      |     |  |       C  |  ) || (
|  ( T2   C2  +--+------+--|>--+--R--+  |       |  R  ) || ( T1
|  (      |   |  |  Q2a    |            |       R  |  ) || (
|  (      |   C4 V      +--|            |       |  |  ) || (
|  |      |   |  |      |               |       |  |  |    |
|  |      |   +--+--<|--+------O Vdd    |       +--+--+    +--O COLD/GND
|  |      |          |    Q4a           |             |
|  |      |          |--+               |          [VBIAS]   =
|  |      |             |               |                   ===
|  |      |             |               |                  =====
|  |      |             |               |                    | GND
+--+------+-------------+---------------+--------------------+

R1/L1/C1: 1 ohm / 1 uH / 47nF - example zobel.
R2/C2/R3/C3: 4th order Sallen-Key lowpass filter (e.g. 250KHz? Should complement trafo)
R4/R5/R: Self bias resistors (1-1000R typ) and gate stoppers (10R, 221R, 1000R?).
R6: Z_IN bootstrap/feedback resistor.
C4/V: Sub-nanoampere voltage source (e.g. DPAD diodes), where If(V) = Igsx(Q3|Q4).
C5: Ciss at lowest Vdg swing.
T1/T2: Input/output trafos, single leg.
Vbias: Measure Igsx(Q1)-Igsx(Q2) through the secondary, and get a suitable diode with Vref=Vgs(Q5,Iq)
Vdd/Vss: Used to cross decouple cascodes between balanced halves of the circuit.
Q1: N-channel JFET follower, low Vgs(off), high Gm, low r_ds, high current, low noise.
Q2: P-channel JFET follower, low Vgs(off), high Gm, low r_ds, high current, low noise.
Q3: N-channel JFET cascode, high Vgs(off), high Gos, high current (2Idss(Q1)<Idss(Q3)<20Idss(Q1)), high (BR)Vgss (>30V).
Q4: P-channel JFET cascode, high Vgs(off), high Gos, high current (2Idss(Q2)<Idss(Q4)<20Idss(Q2)), high (BR)Vgss (>30V).
Q5: Exicon EC-10N16 N-channel VFET follower.

You could also try to test a compound follower between about 100pcs BF246A (or similar cheap N-JFET) in paralell and a single P-channel Exicon VFET..

I wish there was a vBcode for line spacing.. I've just spent hours (literally) getting this post to render before discovering the list hack...

[el`Ol]

Hello all!

I know this is the totally wrong thread, but I hoped to find some transformer experts here.
There is a Japanese Fullrange speaker with two voice coils and a transformer between, at least for low frequencies.
Interesting new fullrange drivers out of Japan
I ask myself whether a simple 1:1 transformer between the voice coils of a dual voice coil woofer would improve the performance in combination with a low dampening factor amplifier.

Regards, Oliver

[erictoucan]

Hi Susan,

I find a thread www.bonavolta.ch/hobby/en/audio/2sk135se.htm

They call it SE Super Nemesis Compensated.

It has equivalent idea to my previously proposed magnetic balanced topology. But, I do not see the point of connecting the gate of balancing Mosfet to the collector output by 12K. What is the exact purpose in doing that? Do you see it?

Eric

27 Dec 2006

[ilimzn]

Originally posted by erictoucan
It has equivalent idea to my previously proposed magnetic balanced topology. But, I do not see the point of connecting the gate of balancing Mosfet to the collector output by 12K. What is the exact purpose in doing that? Do you see it?

It pays to know some history this approach has been used for a loooong time (back when tubes were the only amplification capable component in electronics).
Regarding the 12k, it forms a low pass filter with the 4700uF cap which acts as a 'DC servo' of sorts keeping the balance correct. There is actually a simpler way of doing this, I am surprised they did not think of it.

[erictoucan]

Thank you for your remind, I think you are correct, take the same bais as the amplifing Mosfet to have exact balance. But, do we need that exact? Would 20% deviation significant? I think not and we can just use resistor or lamp of suitable resistance(no need for heatsink) to balance it without the mosfet.