Weird, Weird stuff with the PT of a Twin, Secondary Voltage.

[wizard333]

I went over this thing about 6 months ago, unsoldered the PT from everything, soldered it back, and it worked, for no apparent reason. The chassis is pretty dirty, as far as grime on the metal, and while I cleaned it up electrically, it has a fine layer of yuck on all the metal. Since unsoldering for testing then re-soldering magically fixed it and I couldn't get it to screw up anymore after that, I assumed that the soldering process burned off some semi-conductive crap and solved the issue.

Fast forward 6 months, it is back and doing the same things, and I can't explain why. Here's the deal:

With or without all other circuits unsoldered/disconnected, including the heaters and the bias, and anything past the rectifier diodes, it does this:

AC from each leg of the secondary is 340, so I'm getting 340-0-340.. That should give me 476vDC ish after the rectifier diodes (1.4x340v). But it doesn't. With nothing else connected the end of the diode chain reads 306VDC with about 4VAC ripple. Ok, so some of those old silver diodes test funny, I can't think of why else I'm not getting 1.4x340 at the end, so I replace the diodes. Nope. Same readings. Huh? I pulled apart the eyelet board for the rectifier (mains and bias), cleaned it up, tested for conductivity (none) at the same time I replaced the diodes. No difference.

Checked the center tap connection. It was gimpy, so I got out the big gun and re-soldered it. Tight and solid. No joy, same results.

Now here's the kicker. With no power tubes installed, standby off so nothing but the first bank of caps is in place, the voltage measures: 320VDC and ......120VAC.

WTF?

Keep in mind, neither heaters nor bias are now connected.

1) So why am I not seeing 1.4x340 at the end of my diode chain?
2) How the hell can the VDC go up with the 1st bank of caps connected AND the VAC shoot from 4VAC ripple up to 120VAC??

Doesn't make any sense at all. Any ideas?

That 120VAC showing up with the 1st cap bank connected makes me think maybe a primary leg came disconnected somewhere, but everything seems tight unless somehow inside the PT it came disconnected and I can't see the break. No idea how to confirm that short of replacing the PT, and I don't have a spare Twin PT laying around.

[R.G.]

AC from each leg of the secondary is 340, so I'm getting 340-0-340.. That should give me 476vDC ish after the rectifier diodes (1.4x340v). But it doesn't. With nothing else connected the end of the diode chain reads 306VDC with about 4VAC ripple. Ok, so some of those old silver diodes test funny, I can't think of why else I'm not getting 1.4x340 at the end, so I replace the diodes. Nope. Same readings. Huh? I pulled apart the eyelet board for the rectifier (mains and bias), cleaned it up, tested for conductivity (none) at the same time I replaced the diodes. No difference.

Checked the center tap connection. It was gimpy, so I got out the big gun and re-soldered it. Tight and solid. No joy, same results.

Now here's the kicker. With no power tubes installed, standby off so nothing but the first bank of caps is in place, the voltage measures: 320VDC and ......120VAC.

WTF?

Keep in mind, neither heaters nor bias are now connected.

1) So why am I not seeing 1.4x340 at the end of my diode chain?
2) How the hell can the VDC go up with the 1st bank of caps connected AND the VAC shoot from 4VAC ripple up to 120VAC??

I think that the leads may be loose/high resistance in their connections to the magnet wire inside the transformer, and perhaps shifting around and touching things.

Resoldering not only makes a good connection, but may well involve moving the transformer lead, which could cause the issues noted.

I would get out the ohmmeter, and ohm it out, looking for high resistance primary or secondaries, and less than infinite resistance between windings when the ohmmeter is connected between them and the leads are wiggled.

The 120Vac bothers me a lot. You may have an internal short to a primary lead, and that's quite dangerous.

[wizard333]

All the solder connections on the primary have been checked. Nothing is bad or loose, in fact I take the white lead directly from the courtesy outlet, where it is soldered to the white lead of the 3 prong. Only the black lead sees the fuse and switch, and there are no loose connections.

Similarly, there is *nothing* connected to the secondaries except the rectifier diodes, and that connection is solid.

If it is a bad connection, it is internal.

Riddle me this: Why does the 120VAC not show up until I connect the first cap bank, which theoretically should shunt it to ground anyway? With nothing connected to the end of the rectifier diodes, AC is about 4VAC.

Makes no sense to me whatsoever.

[Enzo]

If I saw those numbers, my immediate thought would be that your DC is not filtered. I have to think the 120vAC of ripple is coincidence to the mains voltage, rather than related. If 120v mains was somehow connecting to the HV secondary, I don't think it would simply impress 120vAC on top of a low DC voltage. SCope it.

If your PT HV CT is well grounded, then find a 10uf or 22uf 500v cap in your drawer and connect it from the + end of the rectifiers to ground. If the voltage wakes up, then your old main caps are either bad or not really connected.

As to why it would measure one thing one time and another thing later, well, I surely couldn't count all the times a meter probe made poor contact with old solder and gave a funny reading. That is why needle sharp probes are important to pierce the oxide surface of joints. Or better yet, clip wires that have been carefully seated.

[wizard333]

As to why it would measure one thing one time and another thing later

It didn't, it consistently gave those readings.

There was high DC that had to be bled if I hooked up the filter stage (which my meter confirmed), and the filter caps are a few months old, so I didn't think bad filter cap. Visually, they looked beautiful, with nice, neat tight soldering. Turns out + the lead on the front one in the series chain was only intermittently connected to the inside of the cap, so it looked great from the outside, but a good tug on the lead yanked it right out of the rubber. Nice. Golf-clap for IC QC.

My question remains however, about how I could be getting a reading of 303VDC and 2VAC at the end of the rectifier. 303VDC and 120VAC I could understand, but if the 120VAC showed up with the bad cap connected, why did I see 2VAC with it not connected?

[Enzo]

306VDC with about 4VAC ripple

320VDC and ......120VAC.

I guess I misunderstood what those referred to then, they don't look consistent.

I don't know other than the things I suggested. But if you corrected a wiring defect and it now works, there you go.

[wizard333]

306VDC with about 4VAC ripple
320VDC and ......120VAC. I guess I misunderstood what those referred to then, they don't look consistent.

The former is with nothing at all connected to the rectifier, the second is with the 1st filter stage connected. Different test conditions. But given one or the other, quite consistent.

I still don't get the 306DC with 2VAC showing up at the end of the un-loaded rectifier though, becoming 320vDC and 120VAC when loaded with (half) a 1st stage filter. Perhaps one of the theoreticians can chime in and explain it, I'm very curious about the technical reason for those readings.

[Enzo]

Really, If you measured unfiltered DC out of the rectifier and came up with only 2VAC ripple, I would write it off to a slipped contact with a meter probe, or a meter selector switch not quite in the notch selecting AC volts, or some other situational anomaly.

Or set up the conditions so you can repeatedly get unfiltered DC with almost no ripple, leave the meter clipped on and then scope it to see what is really going on.

In terms of the circuit or schematically if you prefer, I wouldn't expect to see any explanation.

[wizard333]

It consistently measured that though, over a period of hours, being clipped in and out, off and on with the power, etc etc. Wasn't like I read it once and went "hmmm". It was like that the whole time, which is what really baffled me. Everything else I read with that meter under those conditions was exactly what I expected to see, so I know the meter isn't bad, the ground clip was grounded, etc.

It thought there might be some math heavy theoretical reason why that reading would occur, something I forgot from tube electronics classes now 30+ years ago that would explain why an unloaded rectifier with no filter attached would read like that between the end of the diode string and ground, where the PT center tap was. Still seems odd to me. What I do recall is "Full wave solid state rectification will yield a voltage of 1.4x one leg of the center-tap-grounded PT". Doesn't seem to be the case unless a load or (fully functional) filtering is attached though. Very odd that half a filter would do that. What was attached, when anything at all was attached, since one cap was out, would be a 220k series resistor to an 80uf cap bypassed by another 220k resistor. That gave the 320VDC 120VAC reading. That makes more sense, since if you add those up you get close to the 1.4 number, but 306VDC with 2VAC given no filter and no load is just....odd.

[Tom Phillips]

... but 306VDC with 2VAC given no filter and no load is just....odd.

I don't recall you mentioning the meter model that you were using. Some meters give strange readings with signal that has high DC and simultaneous high AC present. Scope it as Enzo suggested to see the waveform. Glad you found the cause of the problem.

[Enzo]

yes, the scope would reveal any mystery filtration.


One thing comes to mind, Is the connection at the rectifiers completely removed? Or are we relying on the circuit to isolate it?


yes, rectified AC will yield 1.414 times the RMS voltage, in other words the peak voltage. But without a filter it only gets there 60 or 120 times a second and only for an instant. SO yes, a filter is absolutely required to get 1.414 DC on your meter, a load is not.

Now then, a filter cap only has to be large enough to hold up under the current draw from the load. If you have voltage dividers and tubes and stuff all drawing current, then you need several ufs of cap to hold the voltage up. If you have just a bare rectifier sticking out into the air, even a small cap like a .5uf will charge up to voltage, and absent any current draw, it will filter well. SO for example if you removed a 40uf filter cap, but left a 0.1uf parallel cap thinking it didn't matter, well it could matter. And that is an example of what I mean by was the rectifier completely disconnected from the circuit.

If we remove the main caps, but leave the standby on, then the later B+ filter nodes can still act as filters and smooth the supply. Just hypotheticals.

[wizard333]

Is the connection at the rectifiers completely removed

Yes completely, unsoldered, gone. That was one test condition. The other was with only the 1st stage of filtering connected, as controlled by the standby switch, so the later stages show 0 VDC. All this with bias and heaters disconnected (PT leads unsoldered), power tubes out. Although I found it really didn't make any difference whether the heaters and bias were or weren't connected, or tubes were present or not.

I didn't scope it, but I'm going to pull apart another Fender I have here just do to that this weekend. The meter I'm using is the digital Radio Shack one that runs about $80. Laugh if you want, but my $500 Fluke died within 2 years, this thing has lasted 10 and been dropped countless times. It has never given me odd or inconsistent readings before.

So probably the 220k resistor across the 1st cap isolated the 2nd enough that it wasn't filtering much, even though it was charging and showing high VDC.