Zero Feedback Impedance Amplifiers

[Graham Maynard]

Hi Mike,

I wonder if I might enter my thoughts here.

I don't regard Susan's amplifier, or the one I am trying to make, as being 'passive'.

They are active 'followers'. But unlike NFB amplifiers they are not 'reactive'.

Thus no time delayed products due to loudspeaker back EMF generation are *reactively* countered by a second NFB amplifier input coincidentally operating wrt the input waveform. Yes, the loudspeaker current will still be self back EMF modified, and the follower amplifier's output does become passively loudspeaker current modified, but it does not reactively generate new fractional error voltages that would otherwise arise due to multistage phase change and low biasing.

The original four transistor JLH class-A already behaves very competently in both these regards, but these get so hot when greater power output is required.

An output transformer can have weaknesses at audio frequency extremes and when not operating in pure class-A (which excludes toroidals), but it does have advantages.
Identical per-half devices can be used such that the amplitude differences between halves are not going to generate an underlying alternating voltage error on tonebursts.

Susan's and Rozak's prior constructions utilise an output transformer, and it is only when I simulate realistic values for power rail characteristics, with input, output and bias all established wrt a common ground, that I see the psu does not need to be well regulated, merely power/voltage capable.

Non NFB controlled NPN/PNP complementary push-pull halves amplifiy differently, and thus introduce an underlying error.

An all NPN and output transformerless output stage has the upper half working wrt output and the lower half wrt the negative rail. Only on a simulator can this arrangement be made to work, because low frequency bias shifts arise with current draw when the virtual psu is given real world characteristics.

My conclusion - the output transformer is an essential part of a push-pull 'follower' design, and Susan's simple circuit has for long had much to commend it.

My efforts continue.


Cheers ........... Graham.

[EdwinR3]

Hello,

I’m new to this forum and hope I’m not posting in the wrong place.
I've been impressed by the simplicity of the Susan Parker design of the amp and my interest was increased by the mentioning of the suitability to drive ESL speakers.

I am at the moment building a firs prototype of a full-range ESL speaker and winding my own step-up transformers for them, on a second hand winding machine. So here's my question (i apologise if I missed some important issue, but i have no electronics background, being an architect).

1. Can the output transformer and the step-up transformer of the ESL be combined to a single transformer with a lesser step-up ratio so it might be easier to become a wide bandwidth ? If so I suppose the transformer must be completely redesigned ?

2. I understand that the Zeus is a constant current amplifier. Since in theory an ESL that is driven by a constant current has a flat frequency response at large distance, it can might avoid the insertion of (large) resistors between step-up and ESL-stators in order to convert the voltage source into current-source at the cost of efficiency. Is this correct ?

3. Is the layout of the Zeus affected in any way by the reactive load of the ESL ? Is there more/less risk of instability and ringing or can this be controlled in any way without sacrificing the simplicity of the design ?

4. The Zeus is described with several possibilities to connect speakers according to their impedance load. Since the impedance of an ESL is variable (normally starts at 10 Ohm, rising to maybe 150 Ohm and declining to 0.5 Ohm is not unusual), what configuration would be best suited ?

Looking forward for your educating comments.

Edwin

[mashaffer]

Thanks for the reply Graham. First I just want to clarify that I was only indicating that the voltage amplification was passive in that it was accomplished in the transformers. The active devices being (as I understand it) only acting as unity gain buffers.

I was wondering about the application of these ideas to a single ended topology. I assume that a SE EF stage capacitively coupled to the load would be undesireable but I am curious as to the use of a transformer load to a final SE EF stage. Would such a topology be materialy inferior to the PP orientation (other than power output capability).

Thank you for your patience.

mike

[Graham Maynard]

Hi Mike,

For a single ended design to work it would have to run in class-A.

Susan's design is class-AB, but without crossover distortion as it utilises the long gate base of a Mosfet.


Cheers ........... Graham.

[Graham Maynard]

Hi Mike,

Maybe I should re-qualify my class-AB wording above.

Susan's and Rozak's output stages first operate fully in class-A for normal listening levels, and the gradual drift into class-AB at higher output powers is observable via the gentle lift in distortion shown in Susan's measurements earlier in this column.

This is not like the sudden bias set shift from 'A to 'AB in a bipolar output stage which causes a sudden increase in distortion.

I enquired further about the Bartolucci output transformer and Audiomarketing state that a 5% bias offset can be tolerated, though they did not state what nominally designed bias current is intended. Also that it is a Premium Double C core, has seven split and interleaved winding layers, plus a layered secondary combination.
A 20kg transformer must be good for more than 100W; or does a premium core saturate more sharply ?
20kg is often the checked in baggage limit for budget air travellers these days, so it is little wonder that carriage for two pieces to the UK is 150 Euros. Ouch!


Cheers ........... Graham.

[DJNUBZ]

Has anyone found a suitable source for transformers in the us to build this amp with? Anyone willing to build them for DIYaudio members in the states?

[rozak]

Suitable transformers from http://www.plitron.com/
1. PAT4152-00
2. PAT 4144-00

From http://www.sowter.co.uk/ for Single Ended
SE14

[Roscoe Primrose]

Originally posted by rozak
Suitable transformers from http://www.plitron.com/
1. PAT4152-00
2. PAT 4144-00

While the PAT4144-00 would be usable, the turns ratio on the PAT4152-00 (7.8:1 on the 8-ohm tap) is really too high unless you have a: very high efficiency speakers, b: a very high stepup ratio on the input transformer (or another way to achieve a very high driving voltage) and c: a relatively high voltage supply in your amp.

Peace

[Susan-Parker]

Hi Roscoe,

Apologies for the delay in replying.

Originally posted by Roscoe Primrose
[B]The main purpose of the Zeus line driver is to provide a way to drive the 600ohm input of a Zeus power amp, or any other real 600ohm load. However, if you use a high-impedance input transformer, you get little to no voltage gain out of the line driver as it currently exists...

With my normal line driver configuration I have a 1:4 step up input transformer and a 1:1 line-output transformer, so an overall gain of four. With the 6C33C version the tube higher plate impedance lowers the drive to the line-output transformer and gives a gain of about three times.

More than enough for Hi level sources but I agree may be a bit under for "old" style equipment.

It occurs to me that for the power amp, we're loading the output stage with a 32ohm load source-to-source (2:1 stepdown, 8ohm speakers). What about using a stepup output transformer? Assuming the same 32ohm s2s load, that's a 4.3:1 stepup. Even is we went with a 2:1 stepup on the output (which should lower the distortion vs the 4.3:1 stepup) that'd be a help, and easier to wind, the (semi)normal quadfilar output transformer with half-way taps on the secondaries could be wired for a 2:1 stepup of you ran it backwards...

Susan, any thoughts?

In my testing I see a substantial improvement in the performance (reduction in distortion, lowering of output impedance) going from 2:1 to 4:1 step down for those who don't need lots o' watts. My personal feeling is that the best - impedance match to the speakers / distortion / bias level - trade off is between these ratios.

Moving to 1:1 (choke or transformer) or 1:2 (choke) step up is quite possible, and I specifically designed the Zeus75 output transformer to have to flexibility to be able to be operated in this manner. However the distortion and output impedance will be higher.

If one is running from a lower supply rail, e.g. a 12V automotive system then I would consider using a 1:2 step up.

Having a proper transformer output, rather than a swinging choke, also benefits in that the output stage can survive an inadvertent short circuit (whereas with the swinging inductor the magic smoke escaped).

Time and my own personal requirements (not needing much power for a small listening room) obviously affect the areas that I concentrate on, so input from others is always welcome.

Many thanks.

Best wishes,
Susan.

[Susan-Parker]

Hi Graham,

Thank you for your post and suggestion.

Originally posted by Graham Maynard
...
Nice work with the 6C33Cs.

Thank you. It makes a nice central heating substitute. And that's just one pair for one channel of the line driver. Two pairs plus another two pairs (or quads) for the power amps and the heating will be quite significant.

Tube amps for the colder months, mosfets for the summer

With your high quality input step-up transformer this gets back to a more idealistic input loading, however the output current is diminished.

Maybe this is where a power bipolar might now be correctly introduced...

... snip ...

Certainly looks an interesting hybrid, I will see if I can simulate this. I do have some NPN power devices from my early experiments.

The point with the pure tube triode implementation is to show that the Zeus follower topology is perfectly usable with vacuum devices.

Also I am seeing if I can implement the audio path entirely with 1920's style technology. The 6C33C has an indirectly heated cathode, but the same circuit can be accomplished with pairs of directly heated tubes such as the 300B with the heaters in series.



Next is to try out a 6C33C power output stage with higher B+ of about 140 volts and something like a 10:1 step down output transformer.

Best wishes,
Susan.