Why did my filter cap go poof?

[jewishjay]

I'm building a 2204 based on the Mojo British 800 schematic. After triple checking the schematic I plugged it into the lightbulb limiter, measured voltages (550) and watched it for a few minutes, then I put tubes in and measured voltages (367) again. It was so quiet, no hum or anytning, so I plugged in a guitar to make sure it was alive. It was VERY ALIVE. It sounds so good, you guys. I couldn't believe it. So I decided to plug it directly in to the outlet. When I switched it on I heard a faint hisssssssss. So I switched it off ...BANG! poooooooof. The first filter cap gave up the ghost.

These JJ cap cans are rated for 500 volts. I admit...it read 550 without tubes, but once it was under load the plate voltage dropped to 367...why did the cap can explode?

[thetragichero]

i would double up on the first node to account for inrush current

[sluckey]

You had 550V on a 500V can when you were plugged into the bulb limiter. When you plugged straight into the wall that 550V increased, probably significantly. That can was doomed. Doesn't matter that the voltage decreased when the tubes warmed up. The B+ appears on the diodes/caps almost instantly when you hit the switch. Boom boom!

With that much voltage you need to use two caps in series for the first and second B+ nodes.

[jewishjay]

ok, do i need "strapping resistors" in parallel with the new caps?

[sluckey]

Yes. I would use four 100µF @ 350V and four 220K @ 3W.

[jewishjay]

Thanks. So to be clear...100uf@500v + 100uf@500v in series = 50uf@1000v

[DummyLoad]

only one diode per 1/2 primary? are those 2.1K PIV diodes? if those are 1N5308, with the kind of B+ you're reporting, then you should have at least 2 1K PIV diodes in series for each 1/2 primary. you are really putting a stress that silicon otherwise.


--pete

[sluckey]

only one diode per 1/2 primary? are those 2.1K PIV diodes? if those are 1N5308, with the kind of B+ you're reporting, then you should have at least 2 1K PIV diodes in series for each 1/2 primary. you are really putting a stress that silicon otherwise.

Another Mojo snafu!

[jewishjay]

only one diode per 1/2 primary? are those 2.1K PIV diodes? if those are 1N5308, with the kind of B+ you're reporting, then you should have at least 2 1K PIV diodes in series for each 1/2 primary. you are really putting a stress that silicon otherwise.


--pete

they're   1N5408  1000V

[sluckey]

they're   1N5408  1000V

Yes, and you need four of them unless you want them to explode too.

[jewishjay]

ok, no I can do that. and in Mojo's defence I only stole their schematic, they didnt sell me any parts. I sourced most of the parts from Antique Electronic Supply and this whopper of a transformer was NOS from eBay.

what about adding a dropping resistor between the two halves of that first cap can? is that a good idea? cool things down a bit?

[DummyLoad]

suggestion to correct?   


--pete

[Mike_J]

Thanks. So to be clear...100uf@500v + 100uf@500v in series = 50uf@1000v

Sluckey mentioned 100uF @ 350V in series with another 100uF @ 350V which would yield 50uF @ 700V. Use the three watt 220K resistors to balance the draw across both caps reasonably evenly. Don't want most of the draw to happen to just one of the 350V caps or it could be another bad experience. Resistors are important. Do the same thing on each of the first two B+ stages. Wouldn't think you would have a problem after that. That is how I had my 5f6a Bassman set up before on the first B+ stage. Changed my ecaps and now will try two 22uF @ 500V caps in parallel which will make 44uF @ 500V so there is some potential for some excitement but a lot of tweed Bassman amps were made that way without incident.

[jewishjay]

Ok. Here's what exists, and what i'm thinking of doing. Sorry for the sloppy drawing, it's really hard on a small mobile. The blue lines are the additional caps, and since JJ doesnt seem to make 350v cans i went ahead and ordered 100uf @ 500v. the red "butterfly wings" are the 220k 3W resistors. and the red squiggle is possible a low ohm high watt resistor just to split the stages....what do you think of that idea?

[sluckey]

You can't use cap cans to do this. You could use a mixture of cans and individual caps but it would be easier and more straightforward to just use four individual caps. And put a resistor across each cap. Like this...

[PRR]

> put a resistor across each cap.

Are those the invisible resistors?

[mresistor]

LOL

[Mike_J]

The two 220uF @ 350V ecaps with 220K @ 3 watts resistors across each cap would yield 110uF @  700V. This is okay at the first B+ filter stage because you are using solid state rectification (the diodes). If you used a rectifier tube like some earlier Marshall's you would not want to use that much filtering. Each rectifier tube has its limits. You may already know this but I mention it because if you are anything like me someday you may decide to put a switch in that switches between tube, after you added a socket, and solid state rectification and it is not always as simple as just adding a tube. This post may not be necessary for you but it would have been for me at one time.

[sluckey]

> put a resistor across each cap.

Are those the invisible resistors?

The resistors are on a separate layer. You don't see them?

[acheld]

No see.

They went poof!

[PRR]

The resistors are on a separate layer. You don't see them?

 

[sluckey]

Gotcha back!  Hahaha... You got me , I got you.   

Merry Christmas... Steve

[jewishjay]

UPDATE: alright, I fortified the filter section to cope with the inrush voltage, and all seems well there. i ran the amp for a few minutes with the light bulb limiter (it did glow more brightly than any of my other amps) and it sounds great, dead silent at idle, and plenty of volume, gain, and tone available.

However, I next plugged it in without the limiter to take some measurements and after a few minutes one of the el34 began to redplate, so I quickly turned it off. The plate voltage was reading 557 and i could smell the heat and hear a little noise, I observed the cherry glow and switched it off without taking any further bias readings. My IR thermometer says that tube was over 400f !!! What are some possible causes?

All the coupling caps are rated 630v but maybe one is leaky? Theres -43v of bias between the two 220k resistors, but i didnt get a chance to measure it at the actual grids. What else might cause this crazy high current condition?

[sluckey]

Increase the negative bias voltage.

[jewishjay]

Increase the negative bias voltage.

ok, thanks. I changed the 220k bias resistor to 180k to make more neg voltage available. so now the bias is under control, but its not very balanced. at 569PV im seeing .044pc and .036pc works out to 20Wpd and 25Wpd. not sure what to do about that. and the el34s get HOT like 400 degrees...normal?

also i suspect the PV in the preamp might be too high, so i'm thinking of changing the 10k 2w dropping resistors? 
V1a 293      V1b 351
V2a 210     V2b 407!!!
V3a 279     V3b 295

[sluckey]

V2b 407!!!

Nothing wrong with that.

[sluckey]

As for your generally overall high voltages, I suspect that "whopper of a transformer from eBay" is not a good choice for this amp.

[jewishjay]

V2b 407!!!

Nothing wrong with that.

As for your generally overall high voltages, I suspect that "whopper of a transformer from eBay" is not a good choice for this amp.

the datasheet for 12ax7 says 300v max. if you dont feel 407 is too high, why would this transformer be a poor choice?

[mresistor]

My take is because it has too high of voltage and current rating and also does not load down with the circuit like a smaller less powerful and more appropriate transformer would. 

[sluckey]

V2B is a cathode follower with the plate connected to a B+ node. There will be a high voltage on the plate. But there will also be a big voltage on the cathode, probably about 212V. So, the voltage that's actually ***ACROSS*** the tube is only about 407 - 212 = 195V. And there's nothing wrong with that. You have to look at the schematic to evaluate these circuits. 407V may seem excessive on a 12AX7 until you take a closer look at the circuit.