The Transformer That Couldn’t Add

This was driving me nuts. I had a power transformer from an old Dukane power amp that I wanted to use for a project. The transformer tested fine, except for the HV secondary winding, which is center tapped.

With the xfrmr completely disconnected and with no load, between the CT and each leg there was ~350 VAC, but between the legs of the winding there was about 0.6 VAC. Yeah.

There were no shorts anywhere between any windings or any windings and the core. The DC resistance of the secondary HV winding was ~ 92 Ω, and between leg 1 of that winding and the CT was 50 Ω, and between leg 2 and CT was 42 Ω.

I confirmed the wire colors with the schematic, and then measured the resistance about 7-8 times to be sure I wasn’t seeing things. The windings were exactly where they should be. The HV secondary winding also has a bias tap. It has the appropriate resistance and voltage when powered.

The primary has a tap for 117 VAC and 127 VAC.

When I put 122 VAC to the primary, all the secondary windings are about where they should be, unloaded, except for the weird HV reading of less than 1 volt.

I performed R. G. Keen’s Secret Transformer Test for shorts, using a neon bulb, and it tested OK – no internal or external shorts.

I kept thinking it was a polarity issue, but there is no change when I swap leads of either the primary or secondary.

Oh, by the way; Inductance measurements are prone to error with iron core transformers, but I didn’t read anything out of the ordinary on that winding.

This was the key observation from Jeff Gehring at The Gear Page:

Yes! I think that is it. I was wondering if polarity was involved. But why would they do it that way, unless it was a mistake, or for a particularly odd application? Beats me.

Thanks everyone for helping me think this through. I ended up running the Variac down to 60 VAC, and the secondary in question showed 212 mV, about half what I saw at 120 VAC – so it’s not meter-related.
If it is indeed polarity (phase) related, I can confirm it by hooking up the xfrmr like a buck and read the resulting voltages.

This xfrmr came out of a Dukane 1D460A, a PA amp with a single 5U4GB rectifier, and the HV leads to the plates, as normal, with CT grounded and the bias tap to a normal bias circuit.
I have to assume that the amplifier worked properly at some time in its life, maybe with half-wave rectification.

Here are some cartoons showing what is going on.

FIG.1 is a normal “in-phase” secondary – it shows no phase shift of the voltage from the primary to the secondary.

FIG. 2 has 2 secondary windings, each with the same number of turns. We read the same voltage on each, but the top winding is 180 degrees out of phase with the bottom one because the polarities are reversed.

FIG. 3  is what the Dukane transformer is doing: It has the same 2 windings as in FIG. 2, but now they are connected by a “center tap”. Each winding reads the same voltage to the center tap, but because they are o-o-p, reading across the series winding gives zero Volts; the voltage curves are exactly opposing each other.

Oh, and something else: In the case of a “real” center tap, as in FIG. 4, as is the case with every other transformer I’ve ever worked with, The secondary dot can be additive or subtractive (in phase or out of phase), it doesn’t really matter to the rectifier. BUT the center tap is exactly in the center of the winding and thus voltage read at Vb will be exactly 180 degrees out of phase with Va. You will read the combined voltage Va + Vb across the whole winding, because it is just one winding and has just one polarity.

In the case of this Dukane xfrmr, there are 2 windings, each with a polarity, and out of phase and subtractive, so you should (and do) read zero volts across the two windings in series and out of phase.

The 1D460A circuit is completely conventional.  The same transformer was used, I think, in their dual-rectifier designs, which I suspect may have something to do with it. Two separate windings may make sense – each for its own rectifier. Just a guess, but I would think that 2 windings in two different directions around the core would be more expensive to manufacture.

Dukane made dual rectifier amps, like the schematic segment below, and, unlike most other dual rectifier circuits (Mesa), they ran one leg to both plates of each rectifier.  This is unusual, but it works exactly the same as the Mesa circuit, where one leg of the HV went to one of each of the rectifier plates, and the other leg went to both rectifier’s remaining plates.

With the Dukane design, if you lose a rectifier, you get only half-wave rectification.  With the Mesa design, you would still have full wave. This is where I question whether the two identical subtractive polarity windings might influence the “dual rectifier with one rectifier missing” scenario, but I just can’t see how it would be of any benefit; the single remaining rectifier would only be seeing one of the windings on both of its plates – only one sine wave, voltage and current, and it could only rectify half the sine wave (sigh).

I didn’t think it was possible for me to hate an inanimate object, but this lump has changed my mind.

I hooked some diodes up to the HV winding, grounding the center, and yep, half-wave rectification.

So, Jeff, in his wisdom, was right. I have to assume that this was a mistake during manufacture. The thing that blows my mind is that this transformer came out of a supposedly working amp. How is that possible?

I excavated the center tap from the transformer.  You can barely see them, they are so small, but I reversed one of them, soldered it back together, lots of corona dope, and now I’m waiting for it to dry. Tomorrow I will test it and see what happens.

736 Volts across the whole secondary!

The re-wiring of the transformer was successful. Which is good, because a new replacement is over $100. Yay!

I suspect, at the transformer factory, it was Joe’s first day.  I wonder if he made any other transformers like this.