When OT Diagnostics Don't Help

[bob p]

I've got a little P-P EL84 amp that's silent. Voltages everywhere are OK. There's signal on the plates that isn't getting to the speakers. The usual suspect in a case like this is the OT, but in this case the OT tests look pretty good:



In this case, I ran the voltage tests to determine the OT primary impedance to determine its suitability for a conversion, before I bothered to check out the functional status of the amp. It was only after checking out the OT for another project that I realized the amp was having problems.

This is just an example that when the numbers look OK, it doesn't mean that the OT is good.

[Jazz P Bass]

Yeah, the proof is in the pudding.
As Enzo has recommended, it is easiest to simply have a spare transformer for that purpose.
Usually, if the plates have voltage , the cathode is properly connected & you have a real healthy signal at the grids, that is all the output section is meant to do.
If there is still no voltage at the transformer secondary, it is bad.
Sometimes you luck out.
I had an old Ampeg something or other with the shiny chassis.
I could see the transformer arcing in the reflection.
Bad OT!

[bob p]

What surprises me about this is how good the OT numbers look, when the OT is clearly not working.
If anyone sees anything obvious in the numbers that I'm overlooking, please speak up!

[km6xz]

Is there no signal at all on the secondary? If just a little, and mostly odd harmonics, is it possible that the problem is with cancellation due to grids being in phase? Use your dual trace scope to verify the grids being out correctly out of phase. If they are in phase it verifies the transformer is very well balanced if there is no output.
Or you can disconnect the transformer and send it a signal from your signal generator to verify ratio and output.

[bob p]

No signal on the secondary, other than some very tiny illegible garbage that's at/below the limits of my scope's sensitivity.

Just for grins, after starting this thread I made a rough plot of the frequency response of the OT, by inserting ~500mV at the 16-ohm secondary and looking at the plate voltages in the relevant audio band. Again, this gave me the impression that the OT was fine.

Your phase cancellation idea is a good one. I do remember dual tracing the outputs in differential mode. I'll re-check that just to make sure I haven't buggered up the diagnostics. If I find that I missed a bad phase inverter, I'm going to kick myself.

[trobbins]

I'd be amazed if the OT was a problem, given you have nominal DC resistances.

I usually do a high voltage resistance test to confirm no insulation breakdown between windings and to core. That would be my only other static test to typically apply.

Are you using UL configuration? Have you tried using the other primary taps just in a PP config? Have you checked if AC signal on primary side plates (perhaps with two high resistance dividers) , and on taps?

[leydenjar]

I've used R.G. Keen's method of testing OTs for a few years.

Small voltage on secondary tap then read both individual halves of the primary.

They should be very close.

I'll let you define "very close".

[Leo_Gnardo]

I've found shorted secondaries in everything from old Ampeg OT-151A's to brand-new Marshall OT's in 2000 series amps. When the measurements look OK from plate-CT-plate but there's still no (or not much) signal coming out, 5 minutes spent tack-wiring in a known good OT proves where the fault is fast and without further doubt.

[bob p]

Thanks everyone for your insights.

Are you using UL configuration?

Yes, the original OT is UL. The original circuit is a circa 1961 mono HiFi amp that is biased into Class A.

I've used R.G. Keen's method of testing OTs for a few years.

Small voltage on secondary tap then read both individual halves of the primary.

Then I'm sure that you'll recognize the calculations that are in the table. That's the first test that I did!

5 minutes spent tack-wiring in a known good OT...

In the spreadsheet that I posted there are measurements of the secondary inductance, and measurements of the secondary R that were taken with a 4-wire, 6-digit, Kelvin Resistance Meter. I think those results pretty clearly establish that the OT isn't shorted. For the sake of redundancy I went ahead and wired up a spare O50JM OT. That didn't help either. Aargh.

This is just an example that when the numbers look OK, it doesn't mean that the OT is good.

Well, maybe this is really an example that when the OT numbers look OK, they really do mean that the OT is good.

I gave up on the OT hypothesis and started chop-sticking the amp. Still no improvement. I almost bought into the phase inverter hypothesis... until I realized that the dual traces appearing to be in-phase was an artifact of my having left the Channel 2 Inversion button engaged on my scope. Aargh #2.

After hooking up a variety of leads and loads for what must have been 100 different setups, the amp "magically" started working.

So I spent a couple of days banging my head onto the bench as I tried to figure out what was wrong with this little 50-year old vintage HiFi amp. The problem turned out to be stupidly obvious, and something that you might not think about right away if you're used to working on fairly modern/recent gear; it was something that you could miss if you don't think about all the little things that can plague a 50+ year old chassis. The problem? Oxidized zinc screws on the speaker terminals. The white oxidization was nearly invisible on the zinc screws, and even when I thought I had scratched it all away, the terminals just weren't clean enough. D'Oh!

Now I feel really dumb, having spent my time chasing a wild goose!

[Leo_Gnardo]

Oxidized zinc screws on the speaker terminals. The white oxidization was nearly invisible on the zinc screws, and even when I thought I had scratched it all away, the terminals just weren't clean enough. D'Oh!

Now I feel really dumb, having spent my time chasing a wild goose!

Glad to hear you solved it! Perseverance pays off. Well, usually.

And that goes to show us, rust never sleeps, oh how true. Corrosion on whatever kind of metal. If I could have back all the hours I've spent chasing that wild goose... Now where can I find one of those 4-wire Kelvin 6 digit meters like you have?

[bob p]

The oxidation problem kicked my butt for a while. Why? Because the insides of those old-fashioned HiFi flat screw-type speaker terminals are hard to clean. I had no problem cleaning the exterior surfaces, or getting good readings by digging into the terminals with a sharp pointed probe that could penetrate the oxidization, but the flat insides of the screw were much harder to clean, and provided effective insulation to the speaker wires that went to my test load. Couple that with my silly mistake of taking readings off of the daisy chained load resistor, and you can see how one bad terminal connection could cause a cascade of bad readings. Now I'll have to remember to just replace all of those old terminals because they're troublemakers.

4-wire 6-digit meters are commonly available from companies like Fluke in benchtop packaging, but they're kind of expensive.

Fluke - 8845A 120V - Multimeters - Bench - Allied Electronics

8845A/8846A 6.5 Digit Precision Multimeters | Fluke Calibration: United States

[Leo_Gnardo]

The oxidation problem kicked my butt for a while. Why? Because the insides of those old-fashioned HiFi flat screw-type speaker terminals are hard to clean. I had no problem cleaning the exterior surfaces, or getting good readings by digging into the terminals with a sharp pointed probe that could penetrate the oxidization, but the flat insides of the screw were much harder to clean, and provided effective insulation to the speaker wires that went to my test load.

Thanks for the links bob p. Over a grand, hokey skokies! I'll have to keep an eye open for a used Keithley 4-wire milliohm meter from the 70's, more in line with the rest of the obsolete test equipment on my bench. I guess I'll have to do without 6 digit accuracy.

FWIW I've used folded or rolled bits of 600-grit sandpaper, sometimes even glued a little slice of 600 onto a wood stick, to clean those difficult spots. Also I make my own mini wire brushes from bits of guitar string .009 or .010, soldered into brass tubing, broom-style. Haven't seen anything like 'em made commercially. Those "soldering aid kits" sold by Radio Shack include a lame version of the ones I make. Frustration with those led to cobbling together a better tool. And of course there's nothing like a brand new shiny bright connector, if you can find the proper one. Another example of improved output connectors: a customer dropped off a pair of Dyna MkIII with the output screw-strips removed and a pair of 5-way binding posts in their place, gold plated. Me like!

[bob p]

6.5 digits of precision is WAY more than we need for amps and such. I ended up buying the 4-wire meter and the LCR meter for a different project where the precision was needed. But now that I have them, they're really handy for looking at anything that involves a low resistance measurements -- like transformers.

In addition to abrasive cleaning, I guess I should be thinking about chemically cleaning those posts, or just replacing them altogether. Thanks again.