Push Pull Class A OPT design

[6A3sUMMER]

Good push pull output transformers can and sometimes do have equal turns for the 2 halves of the primary. But doing that in a simple fashion will give different DCRs for each half.
Multiple layers of each half of the primary can be used in such a way that the DCR of each half is equal.
Transformer design and build is both an art and a science.

Single Ended transformers can be wound simply, or with more complexity. But one major difference is that the E and I core is usually air gapped. Unlike the push pull transformer, there is not the issue of balancing currents to make it work properly. The tradeoffs are in other factors.

 

[6A3sUMMER]

In some of the literature on Toroids, Menno van der Veen specifications are based on the assumption that the (Quiescent DC) primary currents from the output tubes need to be balanced.

Because it is easy to balance the currents, and later check and rebalance if necessary, why not do so?

My car's engine can run for a while without too much maintenance, but it runs better when properly adjusted.

"Your Mileage May Vary"

 

[trobbins]

MagicBus, well made OPT's can and do have equal half-primary turns, and effectively the same DCR and leakage inductances, but may have noticeably different shunt capacitances. The capacitive imbalance does not practically degrade performance for audio amps, as any effect is well out in frequency and most amps will be managing their stability out at that extreme of frequency by other means.

As an example, in the mid-1950's one of the Heathkit's Williamson OPTs appears to have been 'balanced' during manufacture with the addition of an embedded shunt capacitor across one half-primary. The view is that there was a continuing development of experience with making OPTs during that time, and subsequent experience (a la Patrick Turner) was that such an effort to 'balance' shunt capacitance was of no benefit in practise.

PS. The OPT with the added shunt capacitor, was also a design with a few laminations of a high-nickel steel - visually very noticeable - but not viewed as a great design innovation by many.

 

[MagicBus]

So, DC and AC balance may or may not meet dependent on OPT quality. If we adjust for equal plate currents does this ensure DC cancelation regardless all transformer and tubes imperfections? Any AC imbalance will mostly affect CMRR I suppose.

 

[6A3sUMMER]

AC imbalance can result in more 2nd harmonic distortion. It is like "PUSH pull", or
push PULL". That is Asymmetry, like you can see in Single Ended (POSITIVE alternation, negative alternation; or positive alternation, NEGATIVE alternation).

I prefer to use good power supplies so that I do not have to rely on Common Mode Rejection.

 

[DF96]

The is a loose statement of lenz's law which put another way says that current delivered to the load does not add appreciable flux to the core and indeed all transformers (power, PP and SE) rely on this. However this does not relate to the discussion at hand about the DC offsetting and AC summing in a PP transformer.

It relates precisely to the question which was asked:

Are you sure that is applies to all push pull amps? When operating in say the Class B section of an AB amp the current is only going though one winding - the other is 'off'. Hence there will be a flux 'imbalance', leading to an increase in net flux, possibly reaching saturation.

Given that the OTP primary windings most probably won't be absolutely matched then if a pair of unmatched tubes is used how they should be biased; for hum cancelation they have to be at different idle currents.

You should be able to assume that the two halves of the primary have the same number of turns each. In a good transformer the exact arrangement of windings should have good symmetry so equal currents should result in equal and opposite flux. In a cheaper transfomer the former (equal turns) should still be true but the latter (equal flux) may not be - althoiugh it should be close.

Good push pull output transformers can and sometimes do have equal turns for the 2 halves of the primary.

Good transformers will always have equal turns. Even most bad transformers will have equal turns. Any PP OPT with unequal turns should be returned as faulty.

 

[LinuksGuru]

Good push pull output transformers can and sometimes do have equal turns for the 2 halves of the primary. But doing that in a simple fashion will give different DCRs for each half.
Multiple layers of each half of the primary can be used in such a way that the DCR of each half is equal.

Unequal number of primary halves turns = defective product.
Unequal DCR resistance is absolutely normal, 10 - 20 Ohm is nothing compared to typical 2K - 8K primary impedance.

 

[JMFahey]

You should add bias balancing circuit, and set it so both halves receive exact same current, period.

Some amplifiers, notably Fender "Silverface", did not even have bias adjustment circuit, only bias balancing.

But you can have both, just use two bias pots, one for each half.

 

[TonyTecson]

Unequal number of primary halves turns = defective product.
Unequal DCR resistance is absolutely normal, 10 - 20 Ohm is nothing compared to typical 2K - 8K primary impedance.

i make my OPT's with equal turns on each half to within 1 or 2 ohms dcr of each other.....

i have an antique sound lab, OPT that used dual bays,
both halves of the primary had exactly the same dcr....

 

[6A3sUMMER]

It sounds like you have a good method to keep the DCRs within 1 or 2 Ohms of each other.

Are you using many interleaved winding layers?
Are you using side by side windings? i.e. dual bobbins?
Are you using double C cores?
Are you using larger wire diameters for the outer layers?

 

[TonyTecson]

It sounds like you have a good method to keep the DCRs within 1 or 2 Ohms of each other.

Are you using many interleaved winding layers?
Are you using side by side windings? i.e. dual bobbins?
Are you using double C cores?
Are you using larger wire diameters for the outer layers?

there are threads that discussed about winding OPT's....http://www.diyaudio.com/forums/tubes-valves/109165-opt-design-assistante-el84.html

i use 4 section primary, and 3 to 5 sections secondary..
single bay bobbins, no C-cores, EI-M6.Z11, same size wire for primary...
primary sections 1 and 4 in series and sections 2 and 3 in series..

 

[JMFahey]

Well, that should about average total resistance per primary half, although each successive layer will still have longer mean turns so higher resistance.

When you mentioned "OPT that used dual bays" I thought you had used split bobbin forms, which of course allow winding 2 mirror image primaries side by side if you wish.

 

[laya]

My focus is guitar tube amps at present. It's not uncommon and potentially desirable to have one half of the signal clipping and the other half not at all. My rather weak understanding of Fourier analysis suggests this effectively creates a DC component in the signal - Ie the AC signal imbalance creates a (variable) DC offset component. With that in mind... can that create or reflect in a similar DC offset in a push-pull audio output transformer that could contribute significantly to core saturation? If so, does class A operation have any function in eliminating that (assuming both tubes have identical class A bias current?)

 

[pieter t]

Well, that should about average total resistance per primary half, although each successive layer will still have longer mean turns so higher resistance.

Not "about average"; with equal thickness of isolation between layers, primary sections 1 in series with 4 and 2 in series with 3 will have equal DCR.
Capacitance will be a bit "unbalanced" but in practice no problem.

 

[20to20]

If so, does class A operation have any function in eliminating that (assuming both tubes have identical class A bias current?)

Class A can be overdriven by an input signal, too. It will clip and sound just as bad. Class A is just a bias amount that keeps a normal input signal from dropping into the cutoff range of the tube on the negative signal swing.

 

[6A3sUMMER]

TonyTecson,

Thanks for sharing the number of layers and how you connect them.

For output transformers I would prefer your method; versus the one transformer you had from Antique Sound Labs that used the dual bobbin method.
I would not expect a dual bobbin to couple as well at high frequencies.

I think the dual bobbin method may be more suited for 60Hz Power Transformers,
and other applications that require high isolation.

Years ago, I hand wound a trifilar powdered core toroid transformer to connect unbalanced-to-balanced circuits. I used a Vector Network Analyzer to test it. Then it interfaced the balanced DSL signal to the unbalanced spectrum analyzer input.

Using the same Vector Network Analyzer, I was also able to test interstage and output transformers. I measured frequency response, impedance, phase, capacitance, and leakage inductance. It taught me a lot.

 

[pieter t]

TonyTecson,

Thanks for sharing the number of layers and how you connect them.

For output transformers I would prefer your method; versus the one transformer you had from Antique Sound Labs that used the dual bobbin method.
I would not expect a dual bobbin to couple as well at high frequencies.

I think the dual bobbin method may be more suited for 60Hz Power Transformers,
and other applications that require high isolation.

There is nothing wrong with dual bobbin push-pull output transformers, as done by Lundahl and me.
It gives a real centertap DCR wise; I don't see (and measure) worse coupling at high frequencies when the winding scheme is good.

 

[6A3sUMMER]

pieter t,

I stand corrected. Thanks!
There ARE "dual bobbin" audio transformers that work well.

I have used Lundahl interstage transformers.
They had 2 C-core laminations facing each other (2 air gaps). There was at least one primary and one secondary over each gap. They Did work well.

Also, they have a good measure of hum pickup rejection of external fields, because of the 2 air gaps and the phase of the windings on the gaps. In certain applications I did notice the Lundahl interstage models did have less hum pickup than some air gapped EI interstage models I also used.

I used them both as single ended to push-pull, and single ended to single ended.
The single ended to single ended performed better.
Now I am thinking that push-pull to push-pull might be even better.

But until now I never thought of the Lundahl as dual bobbin. . . . Hmmmm.
I was thinking of dual bobbin as what was pictured in post # 32 above.
Now I have to re-think about how good the performance of a properly designed dual bobbin as in #32 could be.

I have never used a Lundahl push-pull output transformer, I may have to try some of them next.

You mention you have done some dual bobbin push-pull output transformers, did you wind them yourself, or are they commercially available?

Thanks.

 

[TonyTecson]

There is nothing wrong with dual bobbin push-pull output transformers, as done by Lundahl and me.
It gives a real centertap DCR wise; I don't see (and measure) worse coupling at high frequencies when the winding scheme is good.

this was described by Norman Crowhurst at vertical sectioning, reduces capacitance of the windings and so helps at hf...

Lenard Audio has more of this...Valve Amps: Output transformers

 

[soulmerchant]

I have followed this discussion with interest.

Here is another wrinkle for you to consider. Most older PP amplifiers I get to restore benefit greatly from de-gaussing the OPT's. I have done more than a few now and it has real merit. Of course I have told relatively few people that the stunning improvement in their amp's fidelity is chiefly due to this.... high-enders prefer to hear that it was due to exotic caps or wire, etc.

Its a treat to hear those older OPT's sing as when they were new. After de-gaussing, I ensure optimal amplifier function, etc. which usually means the tubes /valves need replacing and re-bias and other associated checks/replacements.

Ian