PS help needed - warm caps & sagging voltage (w/no tubes

[smzinno]

looking for help with some strange behavior in a new power supply I'm testing, no tubes currently installed other than the rectifier. after the supply is up and running for about 15 minutes, the first few filter caps are getting fairly warm and the voltage begins to drop. I haven't spotted any kind of short or other current drain that might explain it so far, and the PT/OT never even gets warm; 6.3v supply is not connected at all.

here's what the string looks like:

PT is 415-0-415 (unloaded), rectifier is 5U4GB

47/450v cap (plates), bypassed w/ 220k
stand-by switch
200 ohm/5w
47/450v cap (screens), splits into two parallel strings:
first string - 2200/2w into 22/450v (fx loop, reverb)
second string -
2200/2w into 22/450v (PI)
10k/2w into 22/450v (v2)
1k/1w into 22/450v (v1)

bias supply is off of main leg, 220k into reverse diode, 22k pot, 33k to gnd. this still needs some work, as I'm only getting about -32vdc max.

the first power up did have a short that cooked the first cap, but that has been replaced. now I'm using the variac through a series 40w bulb to power the circuit. with a supply of 125v I get 430vdc at the first cap on stand-by, and about 410vdc off stand-by. after about 10-15 mins, the off-stand-by voltage will begin to drop a couple volts per minute and the larger caps get warm. the on-stand-by voltage is fairly constant, but the first cap still gets warm. the light bulb is bright at first when the power is applied, then goes pretty dim orange and stays that way.

if I don't use the light bulb limiter, I can get around 450vdc with the variac at about 100v. but the voltage drop still happens after about the same amount of time.

again, all of this is with only the recto tube installed and the 6v heater supply disconnected; PT/OT stay cool the whole time.

any thoughts or suggestions greatly appreciated !

[drz400]

I know it sounds stupid but you sure you dont have a filter cap in backwards?

You have 450V caps in series or just one handling that? that might be pushing it if it sis just one

[smzinno]

the caps all appear to have correct polarity. none are in series at the moment, so it is running pretty much right at the limit for the first couple. I was thinking of changing the first one for a series pair, but wanted to try and determine if there's some other problem first.

is the 200r between the first two filter stages enough to get them decoupled? 47uf is technically already high for the 5u4gb (rated 40uf), so of course double that would be way over. I was thinking of trying a higher resistance between them.

[drz400]

the caps all appear to have correct polarity. none are in series at the moment, so it is running pretty much right at the limit for the first couple. I was thinking of changing the first one for a series pair, but wanted to try and determine if there's some other problem first.

is the 200r between the first two filter stages enough to get them decoupled? 47uf is technically already high for the 5u4gb (rated 40uf), so of course double that would be way over. I was thinking of trying a higher resistance between them.

What makes caps hot is overvoltage and being backwards
two 50's in series will give you 25uf and twice the voltage capcity essentially since you have half the voltage across each one (use 220K bleeders across them).
I would make 100% sure you dont have one in backwards

[VacuumVoodoo]

Are the caps old or new? If old, did you reform them? You may have a cap with pretty high leakage current. A reverse polarized electrolytic would most certainly blow up in your face.

[tonelab2]

Usual 'rule of thumb' safety factor for first filter cap is 40% over( for voltage/current surges).
Most circuits you see traditional totem set up ie: 2 x 100uF/350v in series
= 50uF/700v.
415-0-415 x 1.3 = 539.5v.
On this trans with 5u4 recifier I'd want to use 100uf/450v is series just for peace of mind. YMMV.

[smzinno]

the caps are new, although they came from a local guy with a really large warehouse of stuff, so I can't be exactly sure how long they may have been on the shelf. it's possible they could be old stock in need of being reformed.

I'm definitely going to try and double up for at least the first node, I have some 68u/400 that would get me into a good range.