Opinion on filter cap rating please

[jojokeo]

I've got new tranny coming to raise B+ and wanted to ask how high of a capacitor voltage rating is high enough? My unloaded voltage should be around 458v w/ ss rectification on a 325v PT. My B & C nodes' are rated 500v already and should be fine but the main resevoir cap is only rated 450v. I'll replace it likely w/ two series caps rated at least 350v each (w/ bal resistors) and things will be fine but I wanted to know how high is high enough just for general knowledge?

[phsyconoodler]

Do the series cap thing definitely.Most Filter caps can go over their rating for a short time,but tempting fate is not really adviseable.
  Ask yourself this question:do you run the amp unloaded for longer than a minute to check bias and heater voltages?Yes?Then you re at risk.No?Then you have less risk.
  Safety margins are best exceeded in my experience.

[bluesbear]

"Safety margins are best exceeded in my experience."

Absolutely right. If you build it right from the first, you won't have to go in and do it over. Building (or modding) is fun. Repairing mistakes sucks!
Dave

[jojokeo]

Exactly. But I believe that there is a "rule" or equation where there's a maximum peak voltage that hits the resevoir cap above the loaded "equalized" voltage that the amp settles at when turned on. Also, the amp's voltage will raise and lower while playing, especially at high or max output, from the measured non-playing voltage. So you want ample "headroom" above a certain voltage of your B+ so as to not strain your resevoir cap which would then lead to premature failure and/or poor regulation.
I'm pretty sure I've read about this from Weber, Blencowe, or Aiken somewhere?

[phsyconoodler]

That is doubtful.Every cap manufacturer would have different ideas about what their capacitors can handle.A blanket formula would not be useful.

[jojokeo]

Okay, here's the suggested cap voltage "rule" on Merlin's site:
Voltage ratings: All the capacitors used MUST have a higher DC voltage rating than the HT, and ideally not less than 25% higher, also taking into account any rise that may happen due to variation in the mains voltage. Capacitors with ratings up to 450Vdc are readily available since they are still used in switch-mode power supplies in modern appliances.
However, if we were worried about running a capacitor too close to its voltage limit, we can place two or more capacitors in series to increase the voltage rating of the whole. Equal value resistors should be placed in parallel with each capacitor to ensure the HT is divided equally between them.
The resistors should be large so as not to pass significant HT current, but should pass at least five times the expected leakage current of the capacitor, which is typically in the region of 0.05mA. Values of 220k to 470k are typical, and remember to check their power dissipation. They can also act as a 'bleeder' network that will allow the power supply capacitors to discharge when the amp is switched off.

[HotBluePlates]

What I think you were looking for is: input Vac x 1.414

Ignore rectified loaded, unloaded, etc. The maximum voltage that could exist is the peak of the incoming a.c.

You could then ask yourself the effects of higher wall voltage, etc.

[jojokeo]

What I think you were looking for is: input Vac x 1.414

Ignore rectified loaded, unloaded, etc. The maximum voltage that could exist is the peak of the incoming a.c.

You could then ask yourself the effects of higher wall voltage, etc.

Thanks HPB. That helps in conjuntion with what I was wanting to know that I most recently posted (I answered my own question ultimately). If I use the lower 115v input wire instead of the 120v, I will expect to get appx. 478v off of a 325-0-325 tranny based on the new voltage at it's highest input of 120 mains voltage which then gives ~339v (from the original 325v). Then 339v * 1.41 = ~478v (unloaded B+),    478v * 1.25 (for cap safety margin) = 598v for minimum cap rating to be used. So If I use two 350v caps in series I will be in great shape yielding a single cap rating of 700v.

I was going to do this anyways as noted earlier in this thread but wanted to do the math and gain the understanding along the way for myself and maybe be helpful for others too down the road?

[PRR]

> Every cap manufacturer would have different ideas

They do. And different in different times.

Old-old ratings were "Working Volts" and "Surge".

Fire-up the amp. When rectifier warms, first cap shoots to say 420V. when 6V6 warms, votage drops to say 360V. You need 425V Surge, 375V Working.

Modern cap specs get very exacting. There is a "life" for a certain Voltage and Temperature, with derating for some other stresses. Often the Life is only 1,000 hours. However this is at MAXimum rated temperature. Info is given for estimating life at other temperatures.

Life doubles for every 10 degrees C cooler than rated temperature. If you can stay below 50 deg C, life is long; above 65 deg C you should consider better location, hi-Temp caps, or occasional replacement.

But what about voltage?? Well, voltage is good but excess voltage is excess leakage, which is heat, see above.

And that's why "Surge" rating has vanished (aside from most stuff not needing warm-up today). The turn-on overvoltage hits a cold cap. Temp won't rise much in a few seconds, even a minute or so.

However the life computed from temperature may be extended if you don't need the full voltage.

http://www.niccomp.com/Products/General/Alumlyticlifeexpect.pdf -- Fig 1

Time to failure may double if you only need 0.7 of the rated voltage. Triple at 0.4 rated voltage.

> My unloaded voltage should be around 458v w/ ss rectification on a 325v PT

450V cap is fine for breadboard. The B+ will probably sag in all normal use. We all have amps with 430V-444V on 450V caps giving everyday service.

500V cap may be a better bet. In part because the few 500V cap makers know what you are doing (tube amps) and value your business. You aren't the far fringe of a billion-cap market to them.

2*350V is also fine, but violates the K.I.S.S. rule.

> to raise B+

Why?

[jojokeo]

> to raise B+

Why?

http://www.el34world.com/Forum/index.php?topic=11968.0
That thread sort of explains it. I didn't seem to get much feedback on my theory/question of getting max output power from el34/kt77/6L6 tubes w/ the way the amp was and also a tranny change may allow the PT to run a bit cooler?

Today I had the amp apart and took extensive voltage readings w/ all three tubes - which I should've re-done before the last "button-up" (& cathode resistor bias change). I suppose I jumped the gun and got the high speed wobbles?! Bottom line is the HT wasn't quite as low as I thought and the total mA's for the whole amp wasn't as high as I thought. All three totaled very close btwn 84mA-87.5mA on a 100mA tranny. I was thinking it was maxed out providing all she had (and why it seemed to run so hot) and that I wasn't getting the total output and maybe not quite getting the headroom either? This is my first SE large tube amp so I've got nothing to compare it to which doesn't help either. The tubes are all running at ~22watts Pdiss so that's pretty close. Plus all of my other voltages are exactly where I want them when I designed this using Merlin's load line plotter program, even throughout the VVR's range (controlling only the power tube's plate & screen).
I thought of you saying "what the hell would you do that for? & you likely won't hear much difference if I were to change things?" and also tubenit's comment of "sounding great is a pretty cool description" so I put her back together again and am going to leave it be. So yeah, I got the high speed wobbles, backed off, and regained control of my senses again. It really does play and sound great for what it is meant to be. Maybe another time I'll hit the bigger output tubes w/ 400+ volts and see what kind of difference it makes or not?

[PRR]

> cathode resistor bias change

Cathode-bias power output is (in almost every case) limited by Dissipation.

Raise the voltage, you must lower the current. Now you must re-ponder the load impedance. You get essentially the same power 250V to 600V.

22W Pdiss is not high for these tubes (yes it IS hot).