if you put two electrolytic capacitors in series to double voltage the value is cut in half , example two 100uf @ 450 volts becomes 50uf @ 900 volts correct ? so my question is when doing this you just hook them in series and thats it ? you don't need to add a resistor ? just trying to get around a 475 volt issue , a 450 there might work but the surge would be why they put a 475 volt cap there in the first place.

thanks
mike

Yes just hook them up in series. No resister needed.

Unless an original application specifies it, resisters used across series connected capacitors were there as bleeders or loads. As mentioned, for what you're doing you are fine.

Okay, with some caveats:

http://electrochem.cwru.edu/encycl/misc/c04-appguide.pdf

My first ARRL book was more conservative, and recommended a resistive divider across the caps. Sadly, I've lost both copies. In general, if you use two high-quality IDENTICAL capacitors, you're probably okay. If you don't, you might have issues. I use them when restoring for others, just a nerdy thing with me.

If I undersand your question, I believe it would be a good idea to put equal value resistors across each of the caps to make sure the voltage is equally balanced across both caps (and to serve as bleeders).

---\/\/\/\/\-------/\/\/\/\/\----
| | |
-----||-------------||--------

Oh drat, my ascii art screwed up.. Anyway, there is a good discussion about the need for balancing resistors here:http://www.nmr.mgh.harvard.edu/~reese/electrolytics/ about half way down the page.
Paul

It all depends on the internal condition of the caps you are putting in series. If the unit has a bleed resistor in the circuit, then you may remove it, and add it to the caps as a series string, with 2 resistors in series to make up the original value, tying the center of the resistors to the center of the two caps, making an attempt to keep the right voltages appearing across the two caps. If as I began to say, the condition of the two caps, is that one has a different characteristic that allows a higher than rated voltage to appear across one of the two, then you run the risk of something going wrong, like a small explosion for example. Generally, although the numbers show that you can do it with a cap, its better not to do it with electrolytics unless you have no other options. But then know what can happen one day if one cap develops problems.

The series trick usually works fine without the resistors, but recently I had it fail in an application where I was using identical caps at slightly over the rating for a single cap...I was rather astounded by this, enough so that I may start using balancing resistors with lytics.

I think the problem is, that with newer type 'lytics, the dry-out and failure rate is very high. So one goes dry and starts becoming a resistor in stead of a cap, then the other one goes POOF! I think I'd probably go safe, and keep the voltage rating of each cap at 75% of the voltage being applied to the pair, or the pair being 150% of the voltage of the circuit. I don't think I'd use balancing resistors, as I don't think there's any need for it, and depending on the circuit, they might let a hum get through.

Charles

When using caps in series, bleeder resistors are sort of "insurance" the caps will share the voltage. Remember, caps have rather large tolerences so a cap with -10 to +50% tolerance can vary widely from one of it's siblings. Overvolting the same cap over and over on turn on will eventually hurt it and it may go leaky making a nasty mess of it's sibling. I've always used some value around 220 k ohm across each filter up to 600 volts and never encountered any problems. There was a speaker that had 6KV power supplies for it's amplifiers IIRC that had around 15 47MFD@450 volt electrolytics all in series and used bleeders on all of them (Beveridge 2's if memory is still ok). We changed HV diodes and power trannys but never an electrolytic.

Older Fender Guitar amps would use for the first stage of filtering two caps in series with 220K resistors across them too.

In this set the electrolytics are 475 volts , i have a feeling that a 450 volt would be ok its just the start up voltage may go at or above 450 volts and then drop thats why they have 475 volt caps in there , i could always order 500 volt caps but just for testing and getting the set up and running i was going to double the caps using two 450 volt caps making it 900 volts handling , so that would be way over whats needed but the caps would be ok .
the dumont ra-103 , ra-103D and ra-110 all use a couple capacitors in series with 100k resistors on them , designers put that in to avoid a higher voltage capacitor.

mike

In theory a pair of identical electrolytics will tend to balance the voltages across them fairly well due to leakage and the way the insulating layer is formed by the application of voltage. In practice as others have observed it can go badly wrong, especially if one cap dries out. I'd probably use bleeds even if this wasn's strictly needed. What value bleeds? The current needs to be large compared to leakage and small compared to what's being supplied to the load. This is total guesswork but I'd probably bleed about 1% of the load current.

If you are using non-electrolytics in series to increase the voltage rating then bleed resistors are essential. The problem in some case may be finding high enough values to keep the wasted current down. For example if you replace a 100n 5kV cap in a TV with mains EHT using 10x 1u 630V then 10M resistors will draw 50uA. This may be excessive.

In this set the electrolytics are 475 volts , i have a feeling that a 450 volt would be ok its just the start up voltage may go at or above 450 volts and then drop thats why they have 475 volt caps in there , i could always order 500 volt caps but just for testing and getting the set up and running i was going to double the caps using two 450 volt caps making it 900 volts handling , so that would be way over whats needed but the caps would be ok .
the dumont ra-103 , ra-103D and ra-110 all use a couple capacitors in series with 100k resistors on them , designers put that in to avoid a higher voltage capacitor.

mike

Or what I've done in certain cases with fast-warming B+ rectifier tubes or solid state B+ supplies, is to modify the B+ supply to include a delay relay, so as the filaments are all up before the B+ comes on. That shortens the duration (or eliminates) the unloaded high B+ surge on startup.

Charles