Zero Feedback Impedance Amplifiers

Hi Eva,

Thank you for your comments:

Looking at the pictures I see a massive output transformer, I think it may whitstand 100W at 50Hz before saturation, but it's obviously overrated since I think this amplifier does not provide such a big output swing.

The transformer is designed to cope with low 25 Hz at reasonable listening levels plus full bandwidth audio on top of that, which is where I needed the headroom.

Also the transformer is generating the negative rail so one needs double the apparent core size.

http://www.susan-parker.co.uk/zeus-out-tx.htm

So this circuit may work somewhat better than lumanauw's one at the expense of higher dissipation and the requirement of a custom step-up drive transformer instead of a simple 2:1:1, 1:1:1 etc...

The step up transformer is not really any more problematic than any other small wide bandwidth audio transformer with fine wire windings that have to be matched for exact turns.

The input transformer was made for me by E A Sowter Ltd. It is a simple split bobbin with each half having one secondary (one ends up with two primaries, one on each half which are then wires in parrallel). The bobbin is wound as a (S/2):(P):(S/2) (i.e. split secondary wound either side of the primary) with a 1:10 turns ratio.

Sowter designated it as part # 8160 and they recently quoted me UKP 38.25 (plus VAT) each ex-works for some more (they list specials they design for people, but they don’t charge a big up front design fee).

The imput transformer could be made more exotic by increasing the number of interleaves e.g. (S/4):(P/2):(S/2):(P/2):(S/4) but I doubt that the extra complexity is necessary.

However as a cheep alternative for proving the circuit it would be possible to use one of those low profile encapsulated PCB transformers of around 10 VA with two 120 primaries and two 12 volt secondaries connected backwards to try out the power stage.

e.g. http://www.airlinktransformers.com/lowprofile.html

10VA 0-12V,0-12V LP1012

Within the frequency bandwidth this should enable one to experiment with bias levels and measure distorson etc.


========

Whilst apparently similar to some of the other circuits mentioned the amplifier is not the same as the amplification is in the input transformer (which does not add any noise), not in the mosfets (or transistors).

My novel step was to turn the whole topolgy upside down and use impedance amplification where I trade the impedance differential between input and output windings to get the voltage amplification to drive the mosfet gates (which are high impedance inputs).

I actually went back to first principles and worked forwards from there - I didn't copy any circuit, this is just how it come out.

My basic reasoning was that a speaker driver is a coil in a magnetic field, the same as a transformer, except the secondary is mechanical movement not a current.

To drive this I wanted to cause current to flow in the loudspeaker coil, which I can generate by making current flow in a transformer using followers because I was convinced of not using negative feedback to linierize the output devices.

It took a while to figure it all out.

Signal to noise is OVER 130 dB - I can't actually measure the noise with the equipment I have.

I have also redesigned the amp to work VHF between 30 to 75 MHz.

========

Hi Hugh,

Novel circuit - outstanding! Have you considered running a phase splitter tube for input voltage amplification?

Thanks. I have though about using other input stages but the whole point is simplicity and low i.e. below 50 volts power supplies (this puts the amplifier into a less stringant CE / FCC compliance bracket). (Just don't mention to the test house that the output transformer generates the 50 volt negative rail so one gets nearly 100 volts effective power supply.)

The amp is designed to sit next to the loudspeaker and be driven from the pre-amp using figure 8 bell wire (UNSCREENED). This saves one may tens of thousands of dollars in speaker connection cable which I believe is around $5000 per yead/meter these days.

Because of the input transformer the amp is immune to earth loops and RFI/EMC pickup. As metioned, it is important to use unscreened cable to avoid the capacitance of the braid. Properly driven with 600 ohm line drivers the amp could be hundreds of feet away from the pre-amp, and on a seperate mains supply circuit.

Of course for an integrated stereo amplifier one could dispense with the imput transformers and drive the MOSFETs with a tube or solid state phase splitter.

I personally prefer mono-blocks from an ideological standpoint as the correct place for a power amplifier is next to the load it is driving.

... And would it be possible to drive the speaker from across the mosfet sources, eschewing the secondary of the transformer, using it only as a centre tapped choke?

It would be possible to drive the speakers from across the mosfet sources, however there are several disadvantages.

1. In the event of a device failure substantial DC could be placed across the speaker - which lets the magic smoke out.

2. With two sectioned secondaries I can match for the loudspaker impedance by series or parrallel switching, whereas this is not possible using the center tapped choke method.

And with the options of the secondary windings I can even reconfigure to drive sub ohm impedance loudspeakers.

Q. Okay, why would I want to do that?

A. Line arrays.

My amplifier can drive line array speakers with ALL the drive units wired in parrallel. A line array with 16 off 2" full range 8 ohm drivers become a 0.5 ohm impedance loudspeaker.

Try driving that with any solid state amplifier!

================

I have been asked about sound quality.

Reproduction is better than anything I have ever heared, including the esoteric Japanese silver wound transformer single ended triod designs (although the price tag might have been influencing my hearing a bit there).

With a zero feedback amplifier loudspeaker matching is more important and some just won't sound right - but this is often the case anyway with amps and speakers.

Phase coherance is accuratly preserved, and it is possible to hear the movement of a violinist as she/he sways about when playing.

Piano played at high volume (where the original sound is at high volume - not just turning up the wick) allows all the complex harmonics and decays to be heared.

Of course I might be a little bit biased, but I genuinly wouldn't want to swap my amps for any others. I know that I can improve things a bit here and there - better matching of the mosfets for example but overall I am happy with the sound.

Others might not be so tolerant of valve level distorsion but for me phase accuracy is very important, more so than for many other people perhaps.

One of the design goals I set out to acheive was to design something that was easy and simple to build without sophisticaled set up techniques and with "Instant ON" performance.

Any mains power supply transformer manufacturer can make these transformers cheeply.

The mosfets are cheep and readily available - all the componets being umbiquitious.

The amp can be powered off a couple of car batteries and driven by a walkman headphone output.

No circuit boards are required in the construction - although a tag strip does make life easier.

The sound quality is "as good as" any mass produced solid state amp system, but has almost zero componets and is easy to maintain and repair by anyone with basic electrical skills i.e. it does not need an electronics engineer/specialist technician to fix.

Therefor the amplifier is also ideal for third world use where simplicity, robustness and repairability is more important than sophistication.

===========

Please forgve me if I have missed out any particular points.

Thank you all for your interest.

Best wishes,
Susan.
 
  • Like
Reactions: 1 users
If the sources are in voltage swing, should the gate driver looks like this?
 

Attachments

  • transformeramp1.jpg
    transformeramp1.jpg
    44.6 KB · Views: 7,011
:D

On Topic: Susan: Great innovative design! How severe is the problem with capacitive loading of the input signal, through the voltage amplifying transformer? Can it run with a passive preamp?

Your amplifier really gives meaning to the phrase: keep it simple but not stupid. Others never bothered hearing the line to end ;) They just made it simple :D
Your amplifier has a good power economy, reasonably low output impedance and no feedback. NICE!

The idea of driving the speakers on the primary side of the output transformer is very good i think. The problem with DC in case a MOSFET blows is not significant ... First because the transformer winding would short the DC to GND, and secondly because the PSU fuse would blow very quickly. And third and lastly, the MOSFET's are very unlikely to blow in the first place.

All the best from

Lars
 
You mange to get me totally lost

Hi Susan,

Congradulations for the reception you got with the new thread. I am also glad that it turned out to be sound advice.

BUT

you managed to get me totally lost with your tx design. And being a native speaker (yes, I live in Germany; no I am not German) with sufficient IQ , I guess I might not be the only one -- whether they admit it or not. So maybe you would be kins enough to do some detailed explanation. A few pictures or sketches would of course help a long way.


> Single chamber bobbin, EI 96 x 45.7mm
> Quad Filar wound,

meaning 2 x primary and 2 x secondary windings all side by side ?

> 0.80mm insulated copper wire

Is this empirical, or do you design on internal resistance or current density ? How do you scale this with output power

> No insulation between layers.

Why is this important (or unimportant) ?

> Terminations; to tags on bobbin:

I take it take you are not really centre tapping from one winding (or that the 2 primary windings are not ends on.

> 2 x Primary:
> 12 full layers ( approx. 144 turns each).
> Start and finish at same side of bobbin

> 2 x Secondary:
> 6 + 6 full layers ( approx. 72 + 72 turns each).
> Start and finish at same side of bobbin ( opposite side from
> primary).

What do you mean by 6+6 ?
In your circuit, you showed a secondary with a centre tap (in dotted lines). Are you using a step down ratio of 4 (i.e. no of turns in primary = 4 x no of turns in secondary) ?

> Primary: Start Finish
> Winding 1: 22 24
> Winding 2: 25 27

What do the numbers 22, 24, 25,27 refer to ?

> Secondary: Start Finish Start Finish
> Winding 3: 13 15 17 19
> Winding 4: 14 16 18 20

Same question as above.

> Starting from same bobbin end, wind all four for 6 layers,

All 4 meaning 2 x primary and 2 x secondary, each 0.8mm dia. ?


Thank you in advance for enlightening.

Cheers,
Patrick
 
Hi everyone,

Thank you for your comments.

"Gate follower" is my terminology, probably incorrect as I am not a power engineer. Also dyslexic.

bocka said:


Hi Susan,

also this is off-topic, can you give me more information about this? I'm looking to a similar solution for 40MHz and about 80 Watts for a welding amp. What do you using as in- and output transformer? RF transistors are very difficult to get, toxic inside and expensive! :bawling:

Hi, I am using iron core rings (light blue) but you would probably want to drive in the conventional manner with gain in the output devices as you are not so worried about THD. RF MOSFETS are a necessity at this frequency as the IRFP140s I have been trying out recently only drive up to 1 MHz or so (in my circuit - I havn't tried them with direct drive from a wideband op-amp).


lumanauw said:
If the sources are in voltage swing, should the gate driver looks like this?

Thanks for the schematic. However...

Nope! I only have a single supply rail so the two halves of the transformer cancel out any supply ripple (whichcan get to over 2 volts in the amp in operation) 10,000 uF caps are cheep, and do not put severe current spikes back onto the mains ac supply - otherwise I would have to do power factor correction.

One supply rail = 1 bridge rectifer, 1 = smoothing cap, and low i.e. below 50 volts DC power running around inside the box. Okay, the output transformer generates the negative power rail, but that is ac and so far less dangerious.

I would also avoid caps - particually electrolytic - in the audio signal path.


Lars Clausen said:
:D

On Topic: Susan: Great innovative design! How severe is the problem with capacitive loading of the input signal, through the voltage amplifying transformer? Can it run with a passive preamp?

Your amplifier really gives meaning to the phrase: keep it simple but not stupid. Others never bothered hearing the line to end ;) They just made it simple :D
Your amplifier has a good power economy, reasonably low output impedance and no feedback. NICE!

The idea of driving the speakers on the primary side of the output transformer is very good i think. The problem with DC in case a MOSFET blows is not significant ... First because the transformer winding would short the DC to GND, and secondly because the PSU fuse would blow very quickly. And third and lastly, the MOSFET's are very unlikely to blow in the first place.

All the best from

Lars

Hi Lars:

Thanks.

No, it needs to be driven by a proper 600 ohm ballenced line driver circuit.

There are ways to deal with this, but an active circuit is the simplest.

A point to note is that as the cable is ballenced and of proper impedance there are no signal mismatch phase reflections unlike using phono cables which are 75 ohm (i.e. video) impedance into an effective open circuit. Try running a monitor with video when the end isn't terminated properly and you will rapidly get the idea.

You could drive the input transformer straight from a DAC current outputs, although I am not sure what level of gate drive voltage you would get.

Driving the speakers on the primary side is fine if you know what you are going to use and have speakers of 8 ohm or higher impedance (e.g. LS3/5As which actually sound okay).

However the secondaries gives one a lot more flexability in the speakers (and also allows one to do things like run the amplifier output into an explosafe area as the output is isolated from the amp).

Using triple insulated output windings also helps with CE / FCC approvals as this is nominally a mains powered device.

And one of my interests is running large (up to sixteen) and very sub-ohm impedance line array speakers with all the drivers in parrallel , which necessitates the secondaries.

==========

Best wishes,
Susan.
 
I think it's brilliant. Only 9 components - that puts 47Labs claim to shame:D

I think people forget that some speakers have transformers, Quad electrostatics and Audax gold dome piezo tweeter to name but two.

The design is the polar opposite to some, like Hawksford et al. with their ever increasing complexity.

Yet our hearing is very sensitive to location, and not so sensitive to distortion, so your design is probably more attuned to our hearing abilities than most, something that you have hinted at in your description. The use of line array speakers probably emphasises its strengths in this respect.

Should I dump all my current projects for this minimal approach? what a dilemma:xeye:
 
Re: You mange to get me totally lost

EUVL said:
Hi Susan,

Congradulations for the reception you got with the new thread. I am also glad that it turned out to be sound advice.

Thank you and thank you, as I am new to this type of forum.

Also I have put in a lot of design effort into this amplifier and then been constantly pooh pooed by "those that know" so I must admit to have been a bit anxious about airing it publicaly on the forum.

Most people tell me to ditch the transformers - totally missing the point - and use conventional circuits. What is the point of that sort of comment? It then just becomes another "me too" design and I wouldn't have started on this path in the first place if what was available worked properly.

BUT

you managed to get me totally lost with your tx design. And being a native speaker (yes, I live in Germany; no I am not German) with sufficient IQ , I guess I might not be the only one -- whether they admit it or not. So maybe you would be kins enough to do some detailed explanation. A few pictures or sketches would of course help a long way.

Dont worry, it took me two years and lots of research to figure it out and understand what I was doing.

>> Single chamber bobbin, EI 96 x 45.7mm
>> Quad Filar wound,

> meaning 2 x primary and 2 x secondary windings all side by side ?

Yes. See, not so hard to understand.

This is the key design feature of the output transformer, and why it is NOT the same as a valve transformer.

>> 0.80mm insulated copper wire

> Is this empirical, or do you design on internal resistance or current density ? How do you scale this with output power

Mainly empirical although I did do some maths - long forgotton.

I have a transformer for a certain power over the bandwidth I want to attain.

For 25 Hz to 250 kHz small signal bandwidth I have determined a rough rule of thumb that I needed four times the VA mains rating for the peek wattage I could have (for a 1:1 output that's supply rail x 2 across the load impedance - nominally 8 ohms). So for 50 watts peak I need 200 VA core section.

To scale with a bigger core I would check the bobbin cross section and see how large a diameter of wire I could fit whilst maintaining the same number of turns - i.e. keeping the same(ish) inductance.

For a 75W tranformer I was going to use 1.0 mm wire and for a biggie to push 120 watts plus I was going to use 1.2 mm wire.

If I didn't want the bass response - say I was filtering from 200 Hz - then I could use a smaller core section.

>> No insulation between layers.

> Why is this important (or unimportant) ?

Because mains transformer manufacturers are used to puting the tape in and look puzzled when there isn't any between the primaries and secondaries. So I explicity stated the fact rather then assume it would be understood.

>> Terminations; to tags on bobbin:

> I take it take you are not really centre tapping from one winding (or that the 2 primary windings are not ends on.

Nope, you are correct in the first instance - all four windings done simutaniously. hence quad-filar.


>> 2 x Primary:
>> 12 full layers ( approx. 144 turns each).
>> Start and finish at same side of bobbin

>> 2 x Secondary:
>> 6 + 6 full layers ( approx. 72 + 72 turns each).
>> Start and finish at same side of bobbin ( opposite side from
>> primary).

> What do you mean by 6+6 ?
> In your circuit, you showed a secondary with a centre tap (in dotted lines). Are you using a step down ratio of 4 (i.e. no of turns in primary = 4 x no of turns in secondary) ?

The schematic is the simple version. For the amp in the pic (and this winding specification) I have made provision for configuring the secondary for different impedance loudspeakers. So I split the secondaries in half to get four seperate sections which can then be configured as required.

>> Primary: Start Finish
>> Winding 1: 22 24
>> Winding 2: 25 27

> What do the numbers 22, 24, 25,27 refer to ?

>> Secondary: Start Finish Start Finish
>> Winding 3: 13 15 17 19
>> Winding 4: 14 16 18 20

> Same question as above.

These are the tag numbers on the bobbins that I was using and therefor specified to the transformer manufacturer so I knew which winding went where.

>> Starting from same bobbin end, wind all four for 6 layers,

> All 4 meaning 2 x primary and 2 x secondary, each 0.8mm dia. ?

Not this time.

All four is 2 x primary and 2 x secondary.

Start with four wires from four spools stacked together on a broom handle wedged into a filing cabinate draw (this last bit is optional).

Wind six layers.

Pull out two of the four wires in a loop which will connect to the appropriate tags.

Return the loop so one still has four seperate wires.

Wind six layers.

Wrap in tape and then cut the ends of the wires to length and solder to the tags.

(I would varnish impregnate but I don't have the facilities to do that muself.)

Assemble laminate and attach frame.

Finished.

I hand wind the output transformers, holding the bobbin in one hand and feeding the wires with the other. It takes about 20 minuits to half an hour.

> Thank you in advance for enlightening.

> Cheers,
> Patrick

My pleasure. I hope it now makes some sense.

Best wishes,
Susan.
 
Very interesting - I have a few good-sized silicon steel toroids I may use to wind output transformers to try this out. The only down side is possibly dying of boredom while winding. I'm no stranger to layering and multifilar techniques for reducing leakage/extending bandwidth in transformers, as I design switching power supplies for a living and wind all the prototype transformers myself. This project goes in the pile with the 3-4 other projects that haven't been boxed/listened to just yet, so it may be quite a while before I report resuts. What thickness laminations were used for the output transformer?