Caps in series/parallel

[TerribleT]

Hello,
I posted this at another site but got no help. Hope someone here can answer a few questions.

I got 200 Rubycon GXB 2200uF 16V caps that I want to connect 2 in series to increase voltage capability to 32V and then parallel 25 pairs to get 23kuF.
To obtain correct voltage sharing between the capacitors,
I read that it is a good idea to use voltage sharing resistors. According to a formula I found, I would need to connect a 30k ohm resistor across each cap. Here's the link :http://www.evoxrifa.com/electrolytic...c_appguide.pdf
and an excerpt:

'The voltage sharing resistor is calculated:
Rvsr = 1000/{.015 X C(uF)}
Example: C = 4700 μF
Rvsr = 14 kΩ'

Has any one ever built such a capacitor bank? I would need to wire up 50 caps and 50 resistors. Also can someone look at the second method of connecting a voltage sharing resistor on page 5 of the linked document? It uses only one pair of Rvsr reistor across the series /paralleled caps. Looks much simpler than method I. Soldering in two resistors instead of 50 is much easier task. Also, would I have to increase the resistor's power rating? Finally are there any advantages for either method?
Thanks,
David
http://www.evoxrifa.com/electrolytic...rolytic_appgui http://www.evoxrifa.com/electrolytic...rolytic_appgui
Thanks,
David

[donpetru]

Hello,

Please post exactly schematic how you plan to connect the 50 capacitors and resistors? As you say, you will not get 23kuF but 27,5kuF yes, because you have 25 pairs (with two 2200uF series) * 1100uF = 27500uF.
For example, if your resistor is 30k, capacitor is 2222uF.

Would you like to have each capacitor in parallel to resistance or a group of parallel capacitors to be connected to one resistance? Where you want to use this group or bank of capacitors? The two methods of page 5, do the same, with the observation that one metod need to increase the resistor's power rating.

Nice day

[TerribleT]

Hi,
Thanks for the reply. I want to make some cap banks for a Pass F5 build. I think the voltage rating of 32 volts would just be enough for the F5 power supply. I have 200 of these caps and I can series as many pairs as needed. I'm thinking somewhere between 20k to 30k uF banks. As seen in the picture in the document I've linked, there are two method in connecting the caps. The first requires a resistor across each cap in series to increase voltage capacity and then this pair is treated as one cap that is connected to another pair in parallel and so on. The second picture shows the caps connected as two large banks of paralleled caps and these two banks series together to increase voltage capacity. The picture shows only a one resistor across each bank of paralleled caps. I hope this makes sense, but what I want to know is which arrangement would be the best. And what would the power rating of the resistors be for each case.

http://www.evoxrifa.com/electrolytic...c_appguide.pdf

Regards,
David

[donpetru]

Both arrangements with capacitors and resistors are equally useful (no major differences exist).

However, I recommend to use of each capacitor with its resistance, because of better uniformity voltage as each capacitor and resistance using dissipate less power (preferably use - chemical resistance).

Regards

[Steerpike]

One bonus of using many resistors (i.e, one resistor per cap) is that IF a resistor fails, it will only destroy one cap alongside it. If you use only two resistors total, and one fails, it will destroy ALL the caps as well.

[donpetru]

A tip:

If you use more capacitor in parallel with a resistor, to not fail IF resistor, you can mount a 100nF capacitor in parallel with capacitors high capacity and resistor. However, the best solution remains: one resistor per cap.

Capacity bank depending on the maximum current consumed by the audio amplifier F5. You need calculate this current. After thar, calculate equivalent bank capacity (kuF or uF) and then resistors.

[woody]

If you look at the F5 owners manual the power supply uses clc filters. Each chanel has
it's own +-- power supply using 15,000 uf x2 then .47ohm X4 in parallel then 15,000uf x2. For a total of 120,000 uf per chanel!

So your 200 caps would just make it for a single chanel ... But what happens if you have
one of those caps fail? Finding a bad cap would be rather problimatic.

[TerribleT]

Hello Woody,
Thanks for your input. I can get another 200 caps for $5, so this would be a cheap experiment. As you pointed out in the case of a cap failing, it would be a task to locate the part. But, this brings up a serious concern. What if a cap does fail? It may short or open. How would each of these situation affect the power supply/amp? Would it just blow a fuse or something worse I would hate to replace output transistors!
Regards,
David

[wwenze]

200 for $5 of what qualilty?

Chance of failing increases with quantity used.

[TerribleT]

Hi,
This is surplus stuff. The caps have these markings: Rubycon GXB 2200uF 16v. 105deg C 61. Each cap is 20mm tall and 15mm D.I couldn't find any information on the GXB type. Obviously it's been discontinued from their line. Any information on this cap would be nice.
David