[LASJayhawk]

Do you have a tube tester? It kinda sounds like the 80 has a short plate to filament. That would put a whole lot of AC across the cap and make it go puhpow.

[Jamie]

108.4k on one side, and 109.2k on the other - rectifier filaments to chassis, but these numbers creep upwards slowly, extremely slowly, like .1k every five seconds..

147.9 ohms on one side of plate to neg leads,
167 ohms on the other side of plate to neg leads

[Jamie]

LAS, yes, I tested all the tubes. The 80 tested out good with no shorts showing on my tester. I could search the tube stash and try a different one. Speaking of a short plate to filament, check out this photo...

   

The brown wire is running to plate. The black wire is filament. The brown wire is pinned up against the filament terminal by the black wire. You can see that I pulled it forward about 1/2 inch to see if the covering had worn off. I couldn't detect a short, but we are both on the same page.

[Mondial]

That could be a high resistance short which would arc and put AC into the cap as the voltage is raised.

From your resistance measurements, there is no B+ short and the transformer looks OK.

So it seems that its either the rectifier tube or the shorting wiring which may be the problem?

[Jamie]

Thank you for helping me and hanging in there. I guess I will reroute that wire, try a different 80, and see what happens. You can't tell it from the photo, but the covering looked solid. The distressed area is mostly from me pulling on it with the needlenose. It is pinned in there awfully tight though.. I don't like it. Thanks to everyone who helped, you guys (and gals) are the best.

[morzh]

Simple question: have you tried changing the rect. tube?

[Jamie]

Not yet, but that's on my list. I'll check my tube stash and see if I have another. The tube passed as better than good on my tester.

[EricS]

Rectifier tube is my suspect as well. Caps have a maximum rating for ripple (AC across the cap). If this is too high, the caps overheat and blow. My guess is that your rectifier tube is not really rectifying all that well and is passing a high component of AC along to the cap. Put your meter on the output from your rectifier tube and set it to AC and see what you get. Most run of the mill electrolytic caps can only handle ripple in the low tenths-of-a-volt range, any more and you will be severely stressing the cap unless you have specifically purchased a high-ripple capacity cap.

Eric

[Mondial]

Eric, as you mentioned, most of the high voltage replacement electrolytics sold today are not really rated for the amount of ripple they are subjected to in a vintage radio. The ripple current rating is the spec most important in choosing a cap, and most of the imported caps sold over the internet are not so rated.

I recently checked the power supply of a properly operating model 118 and found about 16V (around 50 V P-P as seen on a scope) of ripple across the input 8 uF filter cap and this level could be considered normal. This was a new replacement cap and was checked for capacitance and ESR before installing. The cap stays stone cold in operation.

The best way to check for AC on the cap as opposed to pulsating DC ripple would be to use an oscilloscope, but I don't believe PuhPow has access to one.

[BrendaAnnD]

Eric,

Is that different somehow from swamping (at least that's what I've heard it called), where the cap is smoothing the pulsating DC bumps by filling in the troughs?

[Jamie]

   

Update -
I replaced the 80 tube with a known good one, and rerouted the wire that provides plate voltage. I plugged the radio into the variac, and slowly added voltage while holding on to the electrolytic capacitor. By the time I was up to 55 volts, the capacitor was too hot to touch. I cut off the power. The overheating cap has the negative lead running to resistor 47, and the positive connected to the filament of the rectifier tube, and another wire leads to the speaker. This is how the previous cap was wired. I simply swapped in a new one.

The only other thing that looks suspect is a micamold cap (see photo) that runs from the ground lug on bakelite block no. 7, to the bandswitch. Don't know if this could cause the issue, but I figured it was worth mentioning.

So, I'm stumped.
Question is, would a total recap and resistor replacement possibly cure this problem? Or do you think is this something that is only caused by a wiring error or a possible (hard to find) short?

[Mondial]

Try removing the 80 tube and slowly powering up the set again. Connect your DVM set to AC volts across the problem cap and watch for any voltage rise. If you see any significant AC across the cap with no rectifier tube, then something is up with the power trans.

[EricS]

Brenda: It sounds like we are talking about the same thing here. My experience comes from high-current draw Class A solid state amps. In this arena, I use HUGE (250,000uF) caps in conjunction with a CRC filter to provide suitably smooth DC for an amp that draws 9A on a +/- 22V power supply. This amp draws nearly 400w out of the wall in order to deliver 100w of power the the speaker. Caps that are too small, don't have high ripple capacity, too old, etc. cause all sorts of problems with the power supply.

I have a few images of PS ripple on one of my web pages. Check here http://www.facstaff.bucknell.edu/esantan...tml#Output and page down 3-4 times to the section on "Measuring Power Supply Ripple" where you'll find pictures from my scope.

PuhPow: That mica cap should be easy enough to check. I've not seen many of these, but to me the shot of the underside looks a little strange. Can you disconnect one end of it and measure capacitance and continuity? This might provide an additional clue...

[Jamie]

Ok, will do. I think I found a wiring error. According to the schematic, both electrolytic caps positive leads connect via a 5k resistor (no 44 on schem) and a .05mf cap (no 43). The other electrolytic cap does run to a .05 cap (inside the condensor bank), but I can find no such connection for the overheating electrolytic cap. It just has two positive leads, one running to the speaker and the other to filament, which then continues on into the transformer. Whoever installed these old electrolytics made some changes.

[Mondial]

Resistor 44 and cap 43 were eliminated in later production so their being missing is not significant.