Partial, Intermittent Failure of Output Transformer

eolon

Gold Supporting Member
Messages
482
I am relating the following tale of woe because it is a very unusual failure mode for an output transformer. I had not seen or heard of it before.

I was recently working on an obstinate 1969 Vibrolux Reverb with an annoying problem. It presented as a weird, unusual distortion when loud chords were played – kind of a crackling breakup and nasty distortion that was not evident at lower volumes, or on most single notes.

A previous tech had tried to fix the problem, unsuccessfully, by adding huge capacitance to the first filter capacitor (it was 147 uF). The later filter stages had all been increased to 80 uF (the schematic calls for 16 uF). This is a vacuum tube rectifier amp. Also, an extra ground had been added between the brass grounding plate and the power transformer ground. The previous tech also bypassed the preamp cathode bypass capacitors with very small value ceramic caps, replaced all electrolytic capacitors, and used shielded wire on the inputs and the volume signal wires. All the filament wiring had been replaced. A three-conductor power cord was installed.

It seems the previous person was trying a lot of things to eliminate the problem, but none of the modifications were successful. I reverted all of the changes back to specification. The problem was still there.

The amp had one weak tube, (V3 12AT7), which was replaced. I couldn’t see how that contributed to the noise, however.

I did all the usual things. The bias was very hot for a VR, so I brought it down to dissipate 35 Watts, the Vibrolux Reverb's rating. The scope was not too helpful – the signal was fine all the way through the amp. I started wondering about the (original) output transformer, but it tested fine – no shorts or opens on either the primary or secondary, and no shorts to the core from anywhere.

I did not try R.G. Keen’s Secret Transformer Tester (https://www.premierguitar.com/articles/print/the-super-secret-transformer-tester-1), so I am not sure if that would have detected anything.

The transformer was not rusty or loose, so I discounted eddy currents as contributors to the noise.

On a hunch, I bought a new replacement transformer from Classic Tone (they are out of business, but you can still find some for sale).

I installed the transformer, and the problem was gone.

(I subsequently modified the circuit to Blackface specs. This is one of the Vibrolux Reverb amps with a tube chart that identifies the circuit as AA864, but this circuit doesn’t actually exist – the actual circuit in the amp is the Silverface AB568. It was converted to the Blackface circuit AA964.)

So…

I suspect that the transformer had a weak spot in the windings that worked fine with small signals at low power, but partially failed as demand was increased on the transformer, creating the crackly breakup. After the transformer surgery, the Vibrolux sounded great – bright and chimey, with solid bass and a typical “Fender” midrange.

Best Regards,
Don
 

Jeff Gehring

Gold Supporting Member
Messages
6,193
Let me guess: the OT was a 022848.

I have had MULTIPLE 022848 output transformers do this (always on speaker load, won't do it on a dummy load). The tell is that when you disconnect the NFB, the problem symptoms vamoose.
The fix is either a new, better OT, or plate-to-plate capacitance added to the PI, or grid to ground capacitance added to the power tube grids.

 
Last edited:

eolon

Gold Supporting Member
Messages
482
Let me guess: the OT was a 022848.

I have had MULTIPLE 022848 output transformers do this (always on speaker load, won't do it on a dummy load). The tell is that when you disconnect the NFB, the problem symptoms vamoose.
The fix is either a new, better OT, or plate-to-plate capacitance added to the PI, or grid to ground capacitance added to the power tube grids.

Hello Jeff:

Wow! Thanks for the information, and the scope shot. Fault is right on the peak. Of course I was using a dummy load when I had it on the scope.
Interesting.
Best Regards,
Don
 

trobbins

Member
Messages
371
Jeff, is there any wiring layout that could exacerbate that parasitic oscillation ? Eg. a 6L6GC anode wire running beside its grid wire (even though there is a 1.5k grid stopper on the socket), or the speaker feedback wire running beside a 12AT7 grid wire.

I guess another fix could be a zobel type RC at the speaker socket, with the feedback signal getting sufficient high frequency attenuation to suppress the loop gain at that oscillation frequency (which must be a nice resonant peak in the output transformer's response when appropriately loaded).
 

J M Fahey

Member
Messages
2,393
Interesting.
So this poorly designed/wound transformer causes bad enough phase shift to make amp unstable.

Weird they didn´t catch the issue at Factory.

Or maybe it was just one bad batch.

EDIT:
I guess another fix could be a zobel type RC at the speaker socket, with the feedback signal getting sufficient high frequency attenuation to suppress the loop gain at that oscillation frequency (which must be a nice resonant peak in the output transformer's response when appropriately loaded).
Speakers are quite inductive and show it more at higher frequencies, as you say a Zobel should compensate for that.

Another quick fix could be bypassing the lower leg of NFB, say with a .1uF ceramic, not for EQ purposes (it will work quite above the Audio band) but "to reduce high frequency NFB".
It has saved me sometimes.

In general NFB is good, but there are exceptions, where less (or none) is best.
 
Last edited:

Jeff Gehring

Gold Supporting Member
Messages
6,193
I should add that I've not really seen the problem in a totally stock wired amp, it usually surfaces when someone either blackfaces the PI, or removes the stabilization caps Fender added on some of those. It's a horrible sound, I can tell you that!
 

J M Fahey

Member
Messages
2,393
You bet.
WAY worse than clipping and even crossover distortion, which is saying something.

And yes, those "stabilization" caps must have been a desperate measure, they are certainly cheesy and not more than a bandaid.
They must have been stuck with a few thousand OTs, which they wouldn´t junk.
 

pdf64

Member
Messages
7,366
I suspect that the transformer had a weak spot in the windings that worked fine with small signals at low power, but partially failed as demand was increased on the transformer, creating the crackly breakup.
I've had a Weber OT with that issue / symptoms, happened to be their 4/8/16 version of the 022848. I pushed it from being intermittent into total failure, ie hard short across the primary, with some heavy overdrive.
What started it off was a 6L6 shorting, I guess that the fault current caused the primary to pass fault current, which caused a damaging degree of heat to build up in one region, It still managed to function at low-medium signal levels, but high levels were able to break through the compromised insulation and internally arc across to the other leg of the primary, the crackling being the audio manifestation of the arcing.
I also managed to push a Hammond 1650G into that failure mode too, though it didn't stay in the crackly zone for very long. No valve short, just heavy usage!
I did an autopsy on the Weber, and found the region of the primary where the windings had discoloured insulation, which as I wound down turned darker.
There are too many secondary windings, and epoxy potting, to bother with an autopsy on the Hammond.
 

Pete Cage

Member
Messages
1,455
Yes to the analysis by @HotBluePlates I've seen this dozens of times over the years with this OT. In most cases, it manifests most easily with a speaker load; sometimes it won't do it at all with a resistive dummy load.

The .002uF from g1 to ground that Fender added to the SF amp power tubes will eliminate this, with no sonic downside. I have to suspect that it was issues with the 022848 OT that caused them to start adding those.

Edit: I should add that the first time I encountered this, decades ago, I chased my tail for a week because it would only happen with an actual speaker load. The painful ones, you remember...
 

eolon

Gold Supporting Member
Messages
482
Some more info:

Maybe a bad lot of transformers in the Silverface circuit, but this one still had the 47K PI plate resistors, the pF caps from output tubes grids to ground, a big electrolytic capacitor on the output cathodes, etc. The modifications by the previous tech were amateurish - poor lead dress, lots of solder blobs, etc. I tried rearranging the lead dress to no avail. The "Hum Balance" adjust had been changed to a true bias adjust when I got the amp, otherwise it was the AB568 circuit, so I wonder why those elements did not cure the transformer behavior.

The speakers were original Oxford 10L5-3, and were tired, and one had a cone repair (that one was also noticeably weaker than the other). After the transformer replacement, the speakers were still flabby. I replaced both speakers with new Alessandro / Eminence GA10-SC64. The new speakers perfected the tone of the amp; I highly recommend them for replacements in this amp.

Best Regards,
Don
 

pdf64

Member
Messages
7,366
Hello Jeff:

Wow! Thanks for the information, and the scope shot. Fault is right on the peak. Of course I was using a dummy load when I had it on the scope.
Interesting.
Best Regards,
Don
Just to clarify, were you seeing Jeff’s parasitic oscillation issue?
 

pdf64

Member
Messages
7,366
I don't know for sure, but from the description I suspect so.
So with the old OT, it was still able to deliver a reasonable full power into a dummy load, or was the output wave collapsing above a certain level?
Sorry to go on, I'm just wondering whether the failure mode was parasitic oscillation, or arcing within the OT, across its primary. Or something else.
Not that it matters really I guess, the fix is the same :)
Do you still have the old OT?
 

Jeff Gehring

Gold Supporting Member
Messages
6,193
I also managed to push a Hammond 1650G into that failure mode too
I guess some people just can't have nice things, Pete! :D

Edit: I should add that the first time I encountered this, decades ago, I chased my tail for a week because it would only happen with an actual speaker load. The painful ones, you remember...
Yes, you do! Back when Bandmaster Reverbs were cheap, I had three of them that I was converting to combo config and blackfacing. All had the 022848 OT, which was just another OT to me then. I finished the builds and tested them as OK, admittedly at low volume settings. Shortly thereafter I wound one of them up to 'pretty loud', and this horrible noise appeared, sounding just like a badly rubbing speaker. I thought the speaker had shelled out for sure.

Savvy and insightful e-sleuth that I am, I thought: "How fortunate am I, that I've got three of these identical suckers to troubleshoot with! I'll just plug this speaker into one of the other chassis and if it then messes up I'll KNOW it is a bad speaker."

Well you can guess how that went. I plugged the PERFECT speaker into another chassis (which also had the 022848, and therefore had the same stability problem) and "CONFIRMED", when it sounded just as bad with the other amp, that the speaker HAD to be bad. Even shipped it off and paid to have it reconed.

When the recone came back, so did the horrible noise. That's when I finally chased it down. I hate that OT...
 

eolon

Gold Supporting Member
Messages
482
So with the old OT, it was still able to deliver a reasonable full power into a dummy load, or was the output wave collapsing above a certain level?
Sorry to go on, I'm just wondering whether the failure mode was parasitic oscillation, or arcing within the OT, across its primary. Or something else.
Not that it matters really I guess, the fix is the same :)
Do you still have the old OT?
Hello pdf64:
Unfortunately, I no longer have the transformer. I had a dummy load and injected about 100 mA into the amp at 440 Hz. I didn't see anything unusual. I played with the gain, but I might not have turned it up all the way. Probably should have. It didn't sound like crossover or other distortion modes, it was unique, which is why I thought it was a weak winding somewhere that only presented the symptoms when pushed, if such a thing is possible.
Best Regards,
Don
 

J M Fahey

Member
Messages
2,393
I suspect that the transformer had a weak spot in the windings that worked fine with small signals at low power, but partially failed as demand was increased on the transformer, creating the crackly breakup.
I find that very unlikely.
See no mechanism for a Zener type breakdown, where below a certain voltage nothing happens and with precision and repeatability, above a trigger voltage it shorts, time and time again.
Yes, there is insulation breakdown any day of the week, but then it never "self heals" or resets, it goes on worse and worse.

I very much suspect instability, doubly so by the clumsy caps added by Fender themselves to tame it, instead of solving the real problem.

Winding a transformer different ways does not change inductance very much, but capacitance is something else.

Something I suspect is that to save time ($$$$$$) they bifilar wound it, so ,say, 1200 turns need only 600 spindle turns.

I did that once for that very same reason (always the OCD optimizer guy) and results were disastrous.

You end with 2 x 120 meters of 0.28mm enamelled wire tightly pressed against each other.

The dictionary definition of a capacitor: 2 metal plates , close by, separated by an insulator.

FWIW I measured about .1uF winding to winding ... disastrous.

In principle such high capacitor kills sparkle, but much worse, it introduces resonances and **shifts phase** in an uncontrolled way.

Not that bad with no NFB, but a mess once you introduce it.

That transformer can still work, but in a non NFB circuit, such as some VOX.
 




Trending Topics

Top