Cause of ultrasonic oscillation in a push-pull output stage ?

[Trevor]

I was wondering if the core material would have an effect re phase shift I can remember years ago trying to use transformer laminations in order to make inductors for a speaker xover their seemed to a lot of eddy current losses in the actual core
Trev

[Radio Wrangler]

There would be plenty of eddy current losses in a core. The job of laminations is to break up the path of circulating eddy currents by providing high resistance (well, poor connection would be a better way of putting it) between each lam and its neighbours. Then there are a lot of eddy current loops, but each one is constrained to be in the thickness of a lamination. Each loop gets less driving potential proportional to its area, but its resistance falls only proportional to its circumference.

So better transformers and inductors employ thinner laminations to cut losses to higher and higher frequencies.

The step beyond this to materials which are generally magnetically conductive without being very electrically conductive leads us to iron-dust cores and ferrites, but in these cases there is gapping in the direction of the magnetic flux and they are less conductive magnetically. Dividing iron up into slices is an attempt to synthesise a non-isotropic material which conducts magnetism in one plane, but does not conduct electricity in the third dimension.

With gapped inductors, you can get rather fierce field intensities which can induce circulating eddy currents in the cross-sectional plane of the wire they're wound with. Litz helps, but also tricks like spacing windings away from gaps helps as well.

Most engineers are strong on electrostatics and capacitors but weak on magnetics and inductors. Often these subjects are taught last in the curriculum and time runs out before much learning is done, then such things are avoided afterwards. Inductors are less perfect than most other passives and so are avoided, so experience isn't built which is a problem as the things really need understanding for the cases where they really are needed.

David

[kalee20]

With practically zero phase shift at 1kHz... And GJ getting the same results even of the scope probes are swapped, I think we can discount any errors in the measuring method.

Interesting that other transformers are OK with regard to anode and screen-grid being in phase, voltage-wise! It proves it's possible.

The Leak transformer really does look bad. Part of successful amplifier design is thinking about what frequency you don't want it to oscillate at, and ensuring that the design is such that it doesn't. An amplifier which oscillates just over two octaves above the upper audio limit is not good, and it really does look a marginal design. Good news is that there is scope for someone to redesign the transformer and make a much better job of it; not so good is that there's a dearth of people capable of doing it (though there are some on this Forum).j

[Trevor]

I wonder what the result would be on a linear /RSC transformer as used in the old 5-20 style I should imagine they would be awful as they were built to a price
Trev

[kalee20]

Depends how the economising was done. They could be inefficient through skimping on wire gauge; have bad LF distortion due to small core area and saturating; have bass cut-off due to cheap, low-permeability iron, and yet still have good coupling between the sections of the winding giving the screen-grid tap.

If I was designing the transformer I'd be inclined to wind the primary section bifilar, taking start wire 1 to HT+, finish wire 1 and start wire 2 as screen-grid tap, and finish wire 2 as anode. This would give super-tight coupling and minimal phase shift at the screen tap. But it would be constrained to have the screen tap at 50%.

For any other ratio, winding configuration gets much more complicated. Add that you also want good coupling between the turns for the OTHER valve, while also maximising coupling to the loudspeaker secondary, and the winding geometry quickly gets non-trivial.

It's a sad fact that what you do to minimise leakage inductance also tends to drive up winding stray capacitances - which might explain the opposite shift of phase that GrimJosef is seeing in some of his other transformers.

[GrimJosef]

I just want to say that I'm really grateful for all the effort that people have put in on this thread. I have learnt a lot.

I fear that as far as my particular issue with this amp and the iffy transformer goes there's not much more that can be done. The transformer is (perhaps only marginally) stable using its 8ohm tapping to drive an 8ohm load. How close to the edge it is with nominally 8ohm speakers I don't know. I would be unwilling to use its 16ohm tapping for anything and if this was my amp I'd be looking to replace the transformer altogether. Its owner, who I know is reading this thread from time to time, does have one amp which might be regarded as 'for spares'. So perhaps that's the way to go.

I'm aware that the issue with KT88s and output transformer stability troubled GEC. They wrote an appendix to their book An Approach To Audio Amplifier Design (1957) (downloadable here http://www.tubebooks.org/technical_books_online.htm under Audio) dealing specifically with output transformers.

For now I have to move onto the other 'feature' of this amp - the discrepancy between its specified and its actual distortion levels, mentioned briefly above. To that end I've started a new thread here https://www.vintage-radio.net/forum/...d.php?t=156357. Apologies if this is 'way too much about far too little' and thanks again !

Cheers,

GJ

[Radio Wrangler]

No need to apologise!

It's been fascinating reading. Unfortunately it's led to the worst outcome, that TL50 amps are decidedly marginal, and though there seems to be nothing atypical with the bad transformer, it may be that a combination of parametric spreads has made it a very iffy proposition in an amplifier.

The Radford transformer behaviour is dramatically better.

Shame. THe value of that pair of NOS KT88s in my attic may just have gone down once word gets around...

I think I'll avoid any TL50s that may be brought my way.

I wonder if we've now found a reason why the TL12 etc were more liked. We may never know for sure, but hifi is riddled with speculation (and speculators!)

We've all learned something, and I think this may be one of those threads that gets pointed to a lot in future and from other places.

David

[peter_sol]

Having established that the oscillation is being caused by the output transformer in the faulty amp.
Are you going to try and cure this? Bearing in mind that any new or other transformer may suffer from the same or similar problem.

[GrimJosef]

It'll be up to the amp's owner really. He has five of these amps altogether - two pairs (nominally in different countries, although right now all five amps are here !). He thinks of one of them as a bit of a 'spares donor' although it'd be even better as a working drop-in replacement if one of the regular four should go on the blink. One of the complications is that two amps are grey, with 240V mains transformers and the other three are what Leak called 'champagne' coloured with 120V mains trannies. So they're not exactly swappable. However I could take the output transformer innards out of the champagne 'spare' and put them into the grey shrouds of the duffer.

Cheers,

GJ

[bikerhifinut]

Originally Posted by Radio Wrangler

No need to apologise!

It's been fascinating reading. Unfortunately it's led to the worst outcome, that TL50 amps are decidedly marginal, and though there seems to be nothing atypical with the bad transformer, it may be that a combination of parametric spreads has made it a very iffy proposition in an amplifier.

The Radford transformer behaviour is dramatically better.

Shame. THe value of that pair of NOS KT88s in my attic may just have gone down once word gets around...

I think I'll avoid any TL50s that may be brought my way.

I wonder if we've now found a reason why the TL12 etc were more liked. We may never know for sure, but hifi is riddled with speculation (and speculators!)

We've all learned something, and I think this may be one of those threads that gets pointed to a lot in future and from other places.

David

I may be mistaken David, but as far as I can understand it, I believe the TL50+ was not really meant for the domestic audio/hi fi market, rather it was intended for the PA side of things. I seem to remember talking to a former Leak employee some years ago now and he reckoned that he could hear the output transformer saturating at high levels at low frequencies, suggesting that it had been cut to the minimum in terms of power handling.
In their day, even the TL25+ would have been considered to be a very high powered amp for use in a home, bearing in mind the efficient loudspeakers of the day?
I know my homebrew EL34 amplifier, and also the kit KT88 amp I assembled have much larger output transformers in terms of lamination density and they are only making 35 watts in Class A Push Pull. That could be an indicator?

Andy.

[kalee20]

Originally Posted by bikerhifinut

I may be mistaken David, but as far as I can understand it, I believe the TL50+ was not really meant for the domestic audio/hi fi market, rather it was intended for the PA side of things. I seem to remember talking to a former Leak employee some years ago now and he reckoned that he could hear the output transformer saturating at high levels at low frequencies, suggesting that it had been cut to the minimum in terms of power handling.

That would sound plausible, that the amplifier was designed for maximum watts-per £ ratio rather than minimum %THD x £ product.

I'd love to know the winding configuration so as to know what to avoid! I've designed and wound a few PP transformers but no PP UL ones yet.

Originally Posted by Radio Wrangler

No need to apologise!

It's been fascinating reading. Unfortunately it's led to the worst outcome, that TL50 amps are decidedly marginal, and though there seems to be nothing atypical with the bad transformer, it may be that a combination of parametric spreads has made it a very iffy proposition in an amplifier.

Absolutely! No wonder GJ was scratching his head, when faced with a faulty item, one looks for a duff component, when that fails maybe a build fault, but rarely a design fault in something from a highly-regarded manufacturer!

Has been a great thread!

[Radio Wrangler]

Maybe for the owner of that group of TL/50+ amps, some of the gilding has now dropped off.

It's a difficult concept for some to grasp, but just because something has valves in it doesn't guarantee it to be good, nor does a famous badge. It comes down to engineering and intent. Maybe that amp would have been a better design as a straight tetrode job and not ultra-linear for the PA market?

A friend's father had a 12 inch dual cone Goodmans speaker built into his lounge, driven by a TL12 and a varislope preamp, Chapman tuner, Garrard SL seventy-something with a Deram. It produced plenty of sound and didn't sound bad. A tweeter would have been helpful, but it didn't need more power... and his sons were into Deep Purple... ELP Floyd.

Eventually they built a Mullard 25+25W transistor amp from the blue book and a pair of Wharfedale 3-way kits, the one like the Dovedale. Pundits would cringe nowadays, but the counter argument is that I remember how much fun we all had.

If wind of this thread gets out then TL/50 amp prices might dip a bit and those KT88s in my attic as well.

Another guy in the village (another schoolfriend's father) ran a pair of KT88s - into a Woden UM3 transformer, modulating a Labgear LG300 150W AM shortwave transmitter. His mercury rectifiers were the best bit.... G4LJU

David

[GrimJosef]

I suspect the owner is pretty happy with them. Up to 15W or so their distortion performance is at least as good as the Quad II's and, to be honest, with music as opposed to CW sinewaves we don't spend much time at powers above that. They're relatively quiet too (1.4mV RMS hum and noise into 8ohms). They're a bit oversensitive (I got 50W out for 115mV in) which partly explains the hum and noise that they do have.

The Radford STA100 can put in a better performance (100W/ch from a pair of KT88s in each channel !) but it has individual bias pots and also a drive balance pot which need regular attention to keep it at its best. As a stereo amp it's also nearly unliftable, and they are as rare as hen's teeth and and not cheap if you ever do find them. The KT88s run hard though and in my opinion it's a safer amp with KT90EHs and a small bias circuit change to accommodate them.

Cheers,

GJ

[Valvepower]

GJ,

Here we go four graphs….

THD vs. Frequency for both TL50’s at 1 dB below clip. The clip level was 49.5W into 16 Ohm at 245V mains.

Input Level vs. THD vs. Power Output for both TL50’s. This graph is a tad confusing. Plotted on left y is axis power output (vs. the input level) and on the right y axis THD. I had to plot input level on the bottom x axis as the AP must step the gen amplitude, which is automatically displayed on the graph. I may have another go to tidy them up when I get a bit more time, but they should do.

0.1% THD was reached around 20 watts output for both amplifiers.

Hope these are of use.

Regards
Terry

[Trevor]

Here is a thought what about the properties of the nearly 60 year old lamination's ? Has the core changed its structure Would heating and cooling of the lamination's change the distortion figure
All other components show sign of aging
Regards Trev

[GrimJosef]

For the sake of keeping the oscillation (this thread) and distortion (the other thread https://www.vintage-radio.net/forum/...d.php?t=156357) discussions separate I've posted my response to the above post over there.

Cheers,

GJ

[Valvepower]

Hi,

I’ve attached a plot of the phase response of both UL taps on the output transformer used in the 140/13/61 amplifier. Note the notch (or dip) around 90K on the purple lead tap......

I would have posted last night, but there was some routine maintenance going down and instead using I-player I looked at The Repair Shop and Steve the Clock restorer restore a Tortoise 'shop counter' bell - at lest Tortoise bells don't hoot at 90KHz

Regards
Terry

[GrimJosef]

Great to see those phase plots Terry.

I only measured over the region 80kHz-100kHz in any detail but the plots for my good transformer look pretty much like yours - the phase on the purple lead tap (marked as green circles in my plot at post #84) peaks around 70 degrees and may be turning over at the highest frequencies. By contrast the phase on the white lead tap (green squares) is shifted more than the purple at 80kHz and that shift just gets larger and larger as the frequency increases.

My bad transformer is similar in some ways. But the big difference is that the primary's two halves are closer together in phase variation than they are either in your transformer or in my good one. Maybe that makes it easier for the two KT88s to co-operate, but not in a good way ?

Cheers,

GJ

[Valvepower]

Hello,

Whist looking through some stuff to find a circuit for the PYE Super Black Box and came upon the user manual for the Leak TL50 with original specification and remembered this thread I felt for completeness I should post it on the forum.

Regards
Terry

[Herald1360]

The little boxed paragraph about PA versions is interesting, especially given the subject of this thread!