Zero Feedback Impedance Amplifiers

[janneman]

Originally posted by Susan-Parker
[snip]However the Nemesis is a valve design with the power fet instead of a tube, and the gain stage is the FET transfer characteristics, which as Eva has pointed out are a bit shakey.

These single ended designs by their nature have an ASYMETRIC output impedance, which will affect the reproduced sound quality. They can only drive in one direction, the other direction is dictated by the back EMF of the transformer or by the current sink.

Again this is not necessarily good or bad, it's just how it is.

[snip]Best wishes,
Susan.

Susan,

I think this is incorrect. The Nemesis is a class A design, and it does not depend on the transformer to function properly. In fact, if you increase the load by the transformer ratio it runs happily without the iron. I think this is different from your designs that rely on the transformer to generate the neg swing, so to say.

Jan Didden

[erictoucan]

I find your idea of Push Pull Mosfet Amp interesting. I have the following questions

1. For your SE version of Mosfet power amp. Is it possible to add an additional winding on the output transformer to exactly balance the DC baising current. In this way we do not have to have so big output. And, if the power of the additional winding is taken from the same power rail of output transformer. It can offer a good common mode rejection. So, the save in the output transformer cost can be taken to invest a choke before the filtering capacitor. It enhance speed and purity.
If I use the ordinary main transformer as output for experiment. I can use the mains primary to feed opposite DC to magnetic balance the magnetic DC according to the following:

Iprimary=Isecondary / Eprimary / Esecondary

2. For the single mosfet pre-amp, is it possible to use lower power rating mosfet. 2SK214 or the like. I find 500mA baising is a bit high. Is it really necessary? And same as power amp, Is it possible to add a additional winding to balance the DC bais and the size of output transformer can be reduced.

Eric

[Susan-Parker]

Hi Eric,

Thank you for your post and interest.

Originally posted by erictoucan
I find your idea of Push Pull Mosfet Amp interesting. I have the following questions

1. For your SE version of Mosfet power amp. Is it possible to add an additional winding on the output transformer to exactly balance the DC biasing current. In this way we do not have to have so big output. And, if the power of the additional winding is taken from the same power rail of output transformer. It can offer a good common mode rejection. So, the save in the output transformer cost can be taken to invest a choke before the filtering capacitor. It enhance speed and purity.
If I use the ordinary main transformer as output for experiment. I can use the mains primary to feed opposite DC to magnetic balance the magnetic DC according to the following:

Iprimary=Isecondary / Eprimary / Esecondary

The SE version of the Zeus makes use of the Back EMF of the transformer's magnetic field "collapse" to generate the negative rail (this replaces the capacitively coupled load and resistor or constant current sink found in other designs e.g. Hiraga Nemesis, Pass Zen ).

The bias must therefore be sufficient to produce enough BEMF current to drive the reflected load from the speaker.

With an 8 ohm load and 2:1 step down about 1.5 amps bias is required for full swing to the supply rail, with 1:4 step down 750 mA is sufficient.

However the power mosfets work best when biased quite hard as can be seen from the graphs on my web page:

http://www.susan-parker.co.uk/zeus-se-amp.htm

2. For the single mosfet pre-amp, is it possible to use lower power rating mosfet. 2SK214 or the like. I find 500mA biasing is a bit high. Is it really necessary? And same as power amp, Is it possible to add a additional winding to balance the DC bias and the size of output transformer can be reduced.

As above, although I am normally using smaller mosfets for the pre-amp they are still higher power than the one you suggest.

===

So for Zeus larger gapped transformers are needed for SE operation. And as you say they are a lot larger than ungapped ones:

Comparison between SE and PP output transformers (800x487 93KB JPEG image).

Which is one reason why I am focusing on using a SE pre-amp/line driver stage (with mosfet or triode) and a mosfet PP power stage.

Hope this answers your question.

The simplicity of the Zeus design allows experimentation and quick tests of different components and conditions. So please do have a go and try out these ideas.

Best wishes,
Susan.

[thorstenlarsen]

Hi Susan !

I think you missed his main point.
He wants to eliminate the DC-magnetization of the transformer core by drawing an opposite current through another winding.

Hi erictoucan !
Sure you can do this. Just make sure that you use a current-source (high AC impedance ) to draw the opposing current, or else the extra winding will work as a (shunted) secondary.
I believe that your calculations are correct, even though they are a bit confusing. ( primary of the amp or primary of the mains ? ? )
If you want to use the secondary of a mains transformer as the primary of your amp and the primary of the mains transformer for the opposing DC current, then you can use the voltage-ratio ( being the same as the turns-ratio ) to calculate the current necessary to balance the DC.
Same thing has been done to a version of the "Nemesis"amp.
I think it`s a great idea.
Go ahead ! But report back !

Thorsten Larsen

[Susan-Parker]

Hi Thorsten Larsen,

Thank you for your post - nice to see some movement on this thread.

Originally posted by thorstenlarsen
[B]Hi Susan !

I think you missed his main point.
He wants to eliminate the DC-magnetization of the transformer core by drawing an opposite current through another winding.

Yes, I understand that.

But a SE Zeus uses the stored energy in the transformer's magnetic field as a "battery" to generate an effective negative rail from the Back EMF.

By superimposing an inverse bias the magnetic field is nulled and the Back EMF is lost, therefor the circuit ceases to function.

If the bias is reduced or is insufficient the negative cycle "flatlines" at a given level and is quite distinct.

Note that the natural reverse polarity amplitude of the Back EMF is greater than the applied voltage, hence the follower will always be constraining it's amplitude until the magnetic field is exhausted.

Therefor in this application the described technique is not applicable.

To reduce the effects of supply ripple simply use an active voltage regulator.

In the PP circuit the back EMF still has an effect on the overall operation, making the circuit more efficient by dumping some of the stored energy back into the load reducing the current drawn from the power supply.

The real question is "Can one use standard mains toroids in SE mode without the core saturating?".

The answer is yes, if it is big enough. But there will be higher intermodulation distortion, as can be seen on my pages, than with the PP mode.

Using a mosfet the intermodulation distortion is still relatively low because of the low impedance drive. Low enough? Well that would need listening to.

Best wishes,
Susan.

[AudioGeek]

i think they meant just a dc current for cancellation...

there still would be the ac signal flowing in the output transformer.

[erictoucan]

Hi Susan,

Thank you for thordtenlarsen reply. I think he understood my point.

What I was saying is to use a winding to magnetic balance the core due to the idle baising from power mosfet. I really cannot see what is the difference of this operation in terms of magnetic induction between the SE and PP. If it does not work, why the Push-Pull work??? Why I think this could sound superior is the whole sine wave is generated slimless by one Mosfer (instead of 2 in PP). In this way, for 1.5A bais, the mosfet actually swing from app. zero to 3A. This kind of SE operation does not take another active device to sink the current to provide negative swing. I also think basically non-negative feedback amplifier drive high efficiecy speaker (more than 95dB/W) and the power output from the amplifier is relatively perfect and noble to the BEMF of speaker.

The reason why I want small power Mosfet for pre-amp is the input capacitance of High Power Mosfer is too high and affect the high frequency performance (usually more than 1000pF). With smaller Mosfet, the input capacitance will be 500pF or less. So, the high frequency cut-off will be higher. Well, for this, I am not completely sure if it is good enough to balance the high bais advantage. But, more than 500mA bais in pre-amp seems a bit too high to let all the components to work to their best. That is my point

[Susan-Parker]

Hi AudioGeek,

Thank you for your post.

Originally posted by AudioGeek
i think they meant just a dc current for cancellation...

there still would be the ac signal flowing in the output transformer.

The DC current makes the "battery" to produce the negative supply.

If one removes the DC component then there is nothing of any substance to pull the signal negative of ground.

Particularly driving low impedance speaker loads, even with the reflected step up through the turns ratio, one just runs out of steam on the negative cycles.

This is not theory or the results of some Spice simulation, this is what happens on the bench and can be seen on my oscilloscope (and heard on the speakers).

Best wishes,
Susan.

... Music is asymmetric...

[Susan-Parker]

Hi,

Originally posted by erictoucan
[B]Hi Susan,

Thank you for thordtenlarsen reply. I think he understood my point.

Yes, I am sure he does.

I also understand what you are asking and why you are asking it.

What I was saying is to use a winding to magnetic balance the core due to the idle baising from power mosfet. I really cannot see what is the difference of this operation in terms of magnetic induction between the SE and PP. If it does not work, why the Push-Pull work??? Why I think this could sound superior is the whole sine wave is generated slimless by one Mosfer (instead of 2 in PP). In this way, for 1.5A bais, the mosfet actually swing from app. zero to 3A. This kind of SE operation does not take another active device to sink the current to provide negative swing. I also think basically non-negative feedback amplifier drive high efficiecy speaker (more than 95dB/W) and the power output from the amplifier is relatively perfect and noble to the BEMF of speaker.

Several points to the above:

1. Why complicate something that is simple to start with? Even if it did work the output transformer itself may be smaller but you are adding extra complexity in making a counter bias power supply.

2. The pair of mosfets/tubes in PP are coupled through and driven by the identical windings of the input transformer secondaries, and can thus be considered a single stage.

3. Using a SE pre-amp/line driver gives you a SE signature whilst preserving the inherently benefits of the PP output stage (lower overall distortion across the audio frequency band etc.).

4. If you are using speakers with more than 95dB/W sensitivity then you only need a few watts and can comfortably use the 4:1 step down configuration. You can also use a lower power supply of perhaps 24 volts (even two deep cycle lead acid batteries if you so wish).

5. Just to make sure we are all referring to the same thing the BEMF I am referring to is NOT the speaker's Counter EMF but the magnetic field of the amplifier's output transformer flux changes.

The reason why I want small power Mosfet for pre-amp is the input capacitance of High Power Mosfer is too high and affect the high frequency performance (usually more than 1000pF). With smaller Mosfet, the input capacitance will be 500pF or less. So, the high frequency cut-off will be higher. Well, for this, I am not completely sure if it is good enough to balance the high bais advantage. But, more than 500mA bais in pre-amp seems a bit too high to let all the components to work to their best. That is my point

The input capacitance of the mosfet when used as a follower is the Reverse Transfer Capacitance i.e. Crss, which is better than an order of magnitude lower than the input capacitance Ciss normally seen when using the mosfet as an amplifier.

E.g.
Crss for a STW34NB20 is 90pF
Ciss for a STW34NB20 is 2400pF

You therefor don't need to worry about the HF cutoff of the mosfet, which is about 2 MHz for the STW34NB20.

For the pre-amp/line driver I do generally use smaller TO220 packaged mosfets with a lower supply and bias than the output stage. But lower is relative. Note that Crss may not be smaller for lower power mosfets. (And one doesn't want the value to be too low otherwise the mosfets become very prone to parasitic oscillation.)

Overall I select for performance, measured and heard, to optimize my design. I am after the best possible sound with as little compromise as I am able.

Therefor to me the difference between 200mA and 500mA bias is very small addition in overall power requirements and does not add significantly to global warming.

Best wishes,
Susan.

[erictoucan]

The thing I was worry about the input capacitance of Mosfet is driven by the input transformer. For 2000pF input capacitance at 10kHz, the impedence is 7958 ohm. It is fairly low to small input transformer in the pre-amp. I also think the input capacitance is the problem of the early versions of mosfet amp getting dark high band. People thought Mosfet's input impedence is high. They used simple and high impedence driver stage. But, it is not the story in high frequency. So, I make balance between the power, current, linearity and input capacitance when choosing Mosfet Power Transistors. I am not here to criticise anything. In fact, I love the Push-Pull version too. But, I just want to come to discuss and the best ever design. Thank you very much for your reply again