YAMA (Yet Another Mystery Amplifier)

[PJL]

My understanding is it is a paraphase splitter. Each phase then goes to a separate driver that uses a triode constant current source to deliver a low output impedance. Not sure if I am correct nor if it delivers any real benefit. I would also be surprised if the screen voltage drop using an OA2 in the UL tap does not have consequences! PS: Feel free to dispute any of my ramblings...

[Richardgr]

Ramble away! It is why I posted the details here, to get some valued input to help me assess the way forward.

I bought this thing because I thought it had some special significance because it had been developed by a noted Swedish amplifier designer. The circuit is elaborate and points to someone who had expertise. I have not got the simulation going yet, so can't say how it looks in that domain, but I am expecting it will probably look fine.

The execution is confounding, no sign of finesse. Some of the unsupported connections, which are not crimped together, are a hair's breadth apart. The power transformer does not appear to have the grunt for a quad of 6550As, at least the heater windings are marginal. There is an additional small transformer under there which only appears to be there because of the pilot light. The burrs on some of the metal. Most, not all, holes have grommets, but some are too small so wires are tightly bunched. No chassis earth and a fixed power cable.

So what do I have. Two regal output transformers. They are UL, with secondaries for 4, 8 and 16, and yet there is one more wire from them that is just cut after it enters the chassis. Another secondary option? I will have to remove one to find out.

With all of those secondary options I could build with CFB, then I could clone my Michaelson and Austin TVA-10 amplifier and have a bootstrap driver (utilising all the sockets).

The stressed heater winding is not shorted, so I think I can lay it upside down and power it up and at least listen to it; experience how it was meant to sound. But it does not feel like I can simply repair it.

[Diabolical Artificer]

A paraphase, yep I was wondering where the tail was. 550v HT for the OP section is a bit mad. The 807/6L6 beam tetrode family are more linear at higher voltages but if your running in UL the screen grids get hammered. The 150C2 solution is less than optimal, better to reduce HT if poss.

To that end I'd remove the paper outer of the scorched tfmr to have a look at any possible damage and if needed rewind the htr winding & possibly pull some turns off the HT winding. Leaving the scorched paper on looks gash & could confuse future owners of the amp. I've plenty of Nomex 7 can pop some in the post if you want any Richard.

Andy.

[Maarten]

While probably not the cause of the smoke in this case, picture 4 in post #1 shows a RIFA paper capacitor lurking in the top right corner which can definitely cause smoke and even fire and should be replaced on sight (once the actual fault is found).

[Richardgr]

I managed to peel most of the protective outer paper from the PT to survey the damage of the overheated heater winding. On the back of the transformer it has got hot, but looks salvageable, but to me on the front it looks like the winding has got too hot. The paper I removed was quite carbonised. One of the leads for the other winding was overlaid on the winding that got hot, melting the insulation of that.

This is new territory for me. What are the options?

I suppose in a perfect world I would remove it, get the laminations apart (I was 10 when I last did that), remove both heater windings, remove more paper and hope the next secondary is undamaged. Then redo the windings.

Writing it helps a bit - I suppose I have nothing to lose. Any tips for loosening the laminations? It looks to be varnished. The Es and the Is are alternated as pairs.

Plan B is to use a PA transformer that was used for a 60W EL34 PA. It is slightly bigger, but with a less wide core. It has to 5A heater windings and interesting secondaries (all unloaded): 372V, 139V and 42V. In the original amp the 139V was added to the 372V for a high voltage EL34 (close to 600V I would think), and the 372V also used for the screens. So I think I could buid a 30W amplifier, using the 372 for B+, 139V for regulated screens.

It is annoying to be at this crossroads. Another option would be to add a dedicated filament transformer, tidy the wiring to improve the quality, and keep it as it was conceived.

[Techman]

So what happened when you powered the transformer with no load and everything disconnected? With or without lamp limiter to actually 'prove' whether the transformer is faulty or not. It's got hot, but it may not have shorted turns. On the other hand the primary may also be damaged.

[Radio Wrangler]

V2B, the second triode in Richard's schematic is a unity-gain inverting feedback amplifier.

It loads the anode of the first triode somewhat, and has an equal anode resistor, lacking that loading factor itself. But the loading on the first triode while reducing the drive to the subsequent stages, means that the phase inverter gets a similarly reduced drive, so it should all come out in the wash. A bit of resistor value fiddling is neaded to balance the thing up because the second triode, within its feedback loop isn't infinite gain.

The grid of the second triode is working as a virtual ground.

At higher frequencies than audio, delays and roll-off don't make this a good arrangement, but at audio they can be fine. My favourite phase splitter has to be the Bailey/Radford triode-pentode mod of the long tailed pair.

David

[Richardgr]

Originally Posted by Techman

So what happened when you powered the transformer with no load and everything disconnected? With or without lamp limiter to actually 'prove' whether the transformer is faulty or not. It's got hot, but it may not have shorted turns. On the other hand the primary may also be damaged.

There was no noise or heat from it. Voltages looked OK.

If I am careful with the insulation on the leads from the windings I think I can do a controlled experiment so I can give it a listen. I cantacted the guy who supplied it and he said it had been on a shelf providing good service for some years. Maybe it got left on too long one time.

There is a matching preamplifier that I missed out on when the amp was sold in the auction. They have had it returned and said it was available, but now I am not sure sure I need an addition to this project.

[Techman]

I'd get that preamplifier anyway - if it's cheap enough there's nothing lost. It could be the preamp which has caused the overheating.

It could be a short circuit in the external circuitry rather than a long term overload that's caused the overheating, which is what I think has likely happened - have you checked for this yet?

That transformer is probably ok and just needs a bit of a tidy up and clean with a bit of protective lacquer applied, after all, it's not as if there's any significant potential between those very low voltage windings. You could always separate them slightly and carefully with the use of a blunt tool.

[Diabolical Artificer]

This is new territory for me. What are the options?

Valve amp design starts with, for me at least with the OPT's; what OP valves will suit the OPT, from that, everything else flows. What valves you have, how much time & money your willing to spend all comes into it, but for me the OPT dictates the design.

Next select the OP valves you want to use, from that, looking at valve datasheets you will know what power supply you will need, and conversely what power tfmr you need.

If that tfmr was mine I'd strip it down to make 100% sure it's ok. I use an old butter knife (to separate varnished laminations), a hammer etc to take apart lams. Once you have the first 5 EI lams off it's pretty straight forward.

Lastly I saw a few posts back you intended to bolt on a gain/driver stage from another amp, I'd advise against that. Design something that will drive the actual OP stage, that meets it's needs etc.

Andy.

[Techman]

Andy is going for the all or nothing approach - fair play!

I'd take the easy option and do as I said above, after all, what's the point in stripping and trying to rewind a transformer that you already have a suspicion isn't really up to the job in the first place?

Just get it working with what you've got and run it for a while to see what happens and then assess what needs to be done - if anything.

[Techman]

Originally Posted by Techman

It could be the preamp which has caused the overheating.

I've changed my mind about this as it looks like the preamp has its own dedicated power supply.

Getting that preamp will add value to the whole setup if you ever wanted to sell it, even though valve preamps aren't generally that well thought of. It depends if you can get it cheap enough, which you may be able to do for the previously mentioned reason and the fact that it's not a lot of use to anyone else on its own.

[Richardgr]

Originally Posted by Diabolical Artificer

Valve amp design starts with, for me at least with the OPT's; what OP valves will suit the OPT, from that, everything else flows. What valves you have, how much time & money your willing to spend all comes into it, but for me the OPT dictates the design.

...

Andy.

The complication with this amplifier is that the OPTs are overspecced in relation to the PT. They are 12cm x 10cm, comparable in size to 100W OPTs. They would have plenty of overhead for a pair of 6550As running at 70W a PP pair. That corresponds to over 300mA of current per tube pair with max signal, so the PT would need to be specced at at least 500mA to cope. And then that doesn't chime with the screen grid protection with OC2s when potentially the screen could consume more than 30mA.

There is a nice operating point at 400V on the plate and 270V on the screen with low distortion, 60W per channel, but then the PT voltages are not correct.

Not knowing why there is signs of something burning by one valve base, close to the shorted heater wiring is annoying. Could this thing have been cathode biased at one point? The bias supply is a separate 24V transformer, was it an 'improvement'?

I have a very nice pair of ABB power transformers with matching chokes; all the makings for a pair of monoblocks.

Between Techman and Diabolicial Artificer I am Indecisive Dave.

I need to do a controlled test to verify the operating point.

[Diabolical Artificer]

That corresponds to over 300mA of current per tube pair with max signal, so the PT would need to be specced at at least 500mA to cope.

Those figures are insane, 75mA quiescent current per OP valve is over the top for a hifi amp unless you live in a loft conversion in New York.

If you stick to using 6550 OP valves then yes you need a mains transformer with a high current capability. What's missing from your calculations is the OPT primary impedance, primary L, Z ratio, power specs etc. Unless you know those your stumbling blind.

The first step in designing an amp which I missed out in my previous post is what do you want it to do? EG do you want max power OP or pootle away at 3W whilst listening to ABBA. In your case though you have to work with what you have, the OPT specs, size of chassis & what have you.

Your idea of using 6L6's has merit, it means you can use a smaller mains tfmr, lower HT/B+, and a lower open loop gain, which means a less critical design & better stability.

Looking at the 6L6GC data, attached, each OP valves quiescent current is roughly 50mA, That's 200mA total so you need a power tfmr/supply 1.5 times that = 300mA for a stereo amp. Add a bit on, say 50mA for the previous stages.

All this is academic though, if you can get the amp working as is for a bench & listening test that should tell you which is the best way to go.

Andy.

[Richardgr]

Originally Posted by Diabolical Artificer

...
The first step in designing an amp which I missed out in my previous post is what do you want it to do? EG do you want max power OP or pootle away at 3W whilst listening to ABBA. In your case though you have to work with what you have, the OPT specs, size of chassis & what have you.
...

Andy.

I would think I am not the only person on this forum who has too many plates spinning on sticks and bites off more than he can chew. I bought this thing because it had some provenance, and the cost was less than the sum of the parts (about the same as a pair of equivalent OPTs).

I don't need a 70W per channel amplifier. My daily listener is either a Tubelab SPP (EL84/20W) or an original Baby Huey ECL86(15W). I have bookshelf speakers in front of my desk, and they work fine with them.

However, there is fun to be had taking the Lamborghini for a spin around the block, and I suppose that was part of my motivation. I have a SS Marantz PM8000 paired with Canton Ergo DC92 speakers, that have a power handling of 300W, 20Hz to 30KHz, and that is a whole different listening experience where you can feel the music as well as listen to it (with The Beatles and Come Together you can hear the points where the master was spliced). At some point I shall pair the 6550A amp to those speakers and see how the designer did.

[PJL]

If you want to work with the transformers you have, I am sure you can run the 6550 at lower HT (400V), scrap the 0A2's, and set the bias for quiescent current of maybe 70mA. You could also look at upgrading the heater transformer and moving more of the valves onto that?

[Richardgr]

I am working through the stages of Vintage HiFi grief: Exultation in the purchase, disappointment on the initial inspection, reflection, more reflection, then remediation.

Sometimes there is another 'reflection' stage in that process, mostly after remediation has started.

It seems to me that there is a fatal flaw in this design which is the screen voltage limiting via regulator valves. They have a limited current range, typically 5-20mA, which is perfect for low power circuits, or as references for controlling higher power circuits. They are not suitable for controlling the screens of 6550As in a powerful amp where the screen current can swing from 9mA to 40mA, and the regulator has to have a minimum load of at least 5mA so it continues to regulate.

Much as I would like to restore it to the original circuit, it does not feel like there is any point.

Therefore, I have to do some re-design to make something sensible out of this project.

I do have a replacement transformer that is a similar rating to the current one, but different windings. That transformer would also struggle with the needs of 6550A valves. There is a chassis with holes for 2 octals, 3 B9G and 2 B7G sockets.

The 6550As test good, so I think I will put them to one side, maybe even sell them to offset some costs. I don't foresee the need for that sort of output power at home, they belong in a guitar amp.

I have long had an ambition to build Miles Prowers Vixen amplifier. I really like 5B/255M valves, and they are a shoe-in for 807s. I make my own adaptors for them so they work with octal sockets, they are excellent (and look beautiful) in my Quad IIs. The driver and splitter are 6SL7 and 6SN7, so 5814/ECC82 and 6CG7 would be the same, and use 2 of the B9G sockets.

The Vixen has a valve regulator for the screens. The design is 0A2 + 6AQ5 + 6CB6. They are all 7-pin valves, but I think I can sub the 6CB6 for an EF80, so that would fill all the spare holes on the chassis.

Very much interested in alternate strategies, if anyone has a suggestion? The circuit for the original amplifier was novel. It is a shame not to make use of it but I don't see a simple way forward in that case.

[Radio Wrangler]

That psychology is generally termed "Buyer's remorse" because it happens in far more fields than hifi

That mains transformer is clearly not adequate. The core and bobbin are too small for the power and currents asked of it. Either too much heater current is being taken or else the heater winding wire is too thin for good efficiency. Either way, if you had a replacement to hand, a new one of the original part, you could fit it and have an amplifier as good as when it was new, and it would just go down the same path.... an overheating mains transformer followed by the same failure.

The choice of output valves, their HT voltage and their quiescent current yield the anode-to-anode impedance which the output transformers create from the speaker impedance. Hence the turns ratios of the output transformers will no longer be optimal if you go to gentler output valves consuming less HT (and heater) power. This means that the available power will be reduced, but it sounds like you do not need its full power capability, so maybe a change of output valves and re-setting the bias resistors to suit would be a solution, allowing a re-wound transformer to be used.

Otherwise you have to address the inadequacy of the existing transformer. If re-wound without heater windings, then the gauges on primary and HT wndings could be thickened a little, reducing ohmic heating and improving efficiency. You'd then need to add a second mains transformer to power the heaters.

Well-designed transformers are usually very reliable components. For one to fail, something has to have gone seriously wrong in overloading its outputs. But not all things are designed as well as they need to be. Various famous names in audio have done things with proven inadequacies. That someone has risen to pundit status in audio doesn't seem a reliable indicator that their arithmetic is up to standard for sizing power supplies and components. There is a strong suspicion that the mains transformer in this unit was undersized at the design stage, or that someone used a power supply design already done for a smaller amplifier and just pushed it too far... well it seemed to work at the time sort of thing.

I can understand you wanting to get it going to hear what things sound like with it, but I don't think, even repaired, that it's a good candidate for a long-term user. If your repair involves changing the output valves to something less demanding, then what you would hear would be no longer representative of this amplifier design, so as a means of satisfying your curiosity, it wouldn't be worth doing.

New transformers or re-winds are expensive things, and I worry that you may just have a repeat performance. Transformer sizing and design is a well-trampled task. Seeing one gone wrong this way that hasn't been killed by a faulty load is not a good indication that an easy repair is possible.

David

[Diabolical Artificer]

Very much interested in alternate strategies, if anyone has a suggestion?

I agree that the original design especially with the iffy OP valve screen grid circuit is a waste of time. I think your idea of dialing the power OP down etc is a sound one, but I don't like the Vixen design.

It's hard to see from your schematic but it looks like a Class AB2 design with the cathode followers direct coupled to the OP valves. Class AB2 is hard to get right, I know this because I've had issues with my latest amp build driving the OP valves at HF, I hit slewing distortion & the CF's drifting bias wise with higher OP power - they go more negative driving the OP stage more into class B. A 3 stage design - voltage gain stage, LTP & CF's is also problematic when applying NFB too. Lastly why drive the OP valves into AB2 if you don't need the OP power?

Attached is a schematic for a amp, in that case it uses 5881 OP valves, but I recently used the same gain stage & LTP driver on a 807 amp & got excellent results, see - https://www.youtube.com/watch?v=XIVDh90_Zvk It means you do need a negative supply of approx -80v but you'll see it uses a LTP with a CCS same as the vixen amp.

I'd also recommend using a regulated screen grid supply for the 5B/255M's. It's well worth reading the STC 807 datasheet in regards to this. I used a mosfet regulated supply, the supply you've shown will do if you want to fill holes but I'd advise compromising your design just to fill holes in the chassis.

Andy.

[Marconi_MPT4]

Out of curiosity what is the output transformer specification regarding step down ratio and UL tap percentage? Manufacturers typically specify 40% for a 6550 but will be different for other valves as might be the optimum load. It may be the case these turn out to be better suited to use 5881, 6L6 et al.

Regarding original schematic is R19 really 100 ohms? C11 400pF appears to be a bit of an oddity in that there is not another capacitor wired to V7 pin 5. It might be there to balance out strays.

It would be interesting to see how the amplifier performs as-is, time to liberate that Änglagård CD

Rich