Question about series filter caps

[Nixon101]

I'm about to recap one my my old traynor mkii bassmaster's. For the plate and screen nodes it calls for an 80uf in series with a 2x40uf can. Well as I was shopping for caps, I put the 80's in my shopping cart, then went to get the can's and could only get dual 50's. All F&T caps rated for 450v and 500v for the cans. Now if I realized I couldn't get 2x40 cans, I would have just got 100uf caps to match the cans, but I'm not sure if it's a big deal at all to mismatch capacitance values in series. I'm no expert but off the top of my head I can't see why it would matter as long as the voltage ratings are well within spec, which they are. And I realize that they have a manufacturing tolerance of -10% to +30% so it is likely they wouldn't be matched anyways. But I guess I'm just looking for some reassurance this is totally safe. I suppose I could wire the 2 cans in series for the plate node and the two 80uf caps in series for the screen node to keep each set matching if there really is a problem with mismatching, but I'm guessing I'll be totally fine.

[martin manning]

The voltages will be split in inverse proportion to the capacitances. I'd increase the 100k across the 80u to 100/80*100 = 125k.

[sluckey]

Probably the simplest thing to do would be to buy four axial 80s.

[Nixon101]

I already bought and received the caps. I did my ysr-1 by leaving all the old cans in but just disconnecting them and making a new board inside with all the new caps. For some reason at the time I wanted it to appear "stock" from the outside lol. This one I wanted to just replace the cans and caps like they were originally.

Regarding the voltage sharing.... I kind of had the same idea. But wasnt 100% sure that the voltage would be shared unevenly, but if it would, that it would be so minimal It wouldn't really matter. This thing has about 570v on the plates and just slightly less on the screens. I think I'd be safe just using 100k's as I know I have plenty 3 watt resistors in that value. I have nothing between 100k and 220k, but I do have 82k's. Would a 100k/82k work in a similar fashion to 120k/100k? I've read you need atleast 200k minimum between hv and ground so I don't want to go too low. I'll probably just stick 2 100k's and call it a day.

Edit.... turns out I have a brand new 100uf cap in the parts drawer I can use with the 50/50. But still one short.

[Ten Over]

Stick the 100k balancing resistors in there. They are way overkill, so no need for fine tuning.

[martin manning]

I'll probably just stick 2 100k's and call it a day.

A couple of 250k 1W in parallel would be fine. The current through the balancing resistors wants to be somewhat greater than the cap's internal leakage current. You can work that out, but 100k seems on the low side. My thought was to go up a bit rather than down on the total resistance.

[Ten Over]

The voltages will be split in inverse proportion to the capacitances. I'd increase the 100k across the 80u to 100/80*100 = 125k.

If the voltage drops are inversely proportional to the capacitances, then the voltage drop across the 80uF would be greater than the voltage drop across the 100uF. This would mean that the 80uF capacitor would have a greater resistance than the 100uF capacitor. In order to balance the voltages, the resistance of the resistor across the 80uF capacitor would have to be less than the resistance of the resistor across the 100uF capacitor.

[martin manning]

The voltages will be split in inverse proportion to the capacitances. I'd increase the 100k across the 80u to 100/80*100 = 125k.

If the voltage drops are inversely proportional to the capacitances, then the voltage drop across the 80uF would be greater than the voltage drop across the 100uF. This would mean that the 80uF capacitor would have a greater resistance than the 100uF capacitor. In order to balance the voltages, the resistance of the resistor across the 80uF capacitor would have to be less than the resistance of the resistor across the 100uF capacitor.

The resistors are chosen so that they have the same voltage split as the caps. The voltages remain unbalanced, but matched to the dissimilar capacitances. Since the 80u has more voltage drop, it gets the larger resistor.

[Nixon101]

I'll probably just make do with what I have and put 2 100k's in for now. Next time I make an order I'll get the one capacitor I'm short and redo it. But I'm always looking to learn and soak up the information from all you guys, so keep the comments coming. You guys that also commented on my other thread where I was tossing around the idea of adding the post phase inverter cathode follower to my other traynor amp..... don't think I disappeared. I hoping to still get to that during the Christmas holidays and will update that thread when I do.

[Ten Over]

The voltages remain unbalanced, but matched to the dissimilar capacitances.

For what purpose?

[martin manning]

Same reason as always… so the caps share voltage in proportion to their capacitance.

[Ten Over]

Same reason as always… so the caps share voltage in proportion to their capacitance.

But for what purpose?

[ChopSauce]

What for? Increase max voltage handling, I suppose.

[martin manning]

Same reason as always… so the caps share voltage in proportion to their capacitance.

But for what purpose?

Normally if say 500V rating is required then two equal-valued 250V caps could be stacked. Due to differences in their performance (leakage current when new and over time) they may not share voltage equally, and one of them could be operating over its rating. Balancing resistors force the voltages to be shared equally to avoid that. If two different values are used then they will want to share the voltage inversely with their capacitances, and the voltage ratings and balancing resistors would be chosen accordingly. Here the voltage ratings are more than adequate to survive a large discrepancy. I don't know what the down side of the resistors trying to force a voltage split that is different from the capacitances might be. Ultimately they will settle somewhere in between, which means there will be some transient behavior related to the difference in time constants introduced. It's a trivial matter to adjust the resistor value, so why not? The easiest and cheapest way to avoid any of this is to get another 100u, which will bring the total capacitance of each filter up to 50u as opposed to the 40u it had originally. Probably not a big deal.

[Ten Over]

Normally if say 500V rating is required then two equal-valued 250V caps could be stacked. Due to differences in their performance (leakage current when new and over time) they may not share voltage equally, and one of them could be operating over its rating. Balancing resistors force the voltages to be shared equally to avoid that.

The leakage current manifests itself as a resistance in parallel with the balancing resistor. If the leakage currents of the two capacitors are different, then the parallel combinations will be different and the voltage will not be shared equally even though the balancing resistors are equal. However, the difference in the voltage drops will be reduced since we are dealing with a parallel combination instead of a single resistance. Under these circumstances, a 250V rating is not sufficient for a resistor-balanced 500V system and that could well be why Leo used 350V capacitors. On the other hand, a bridge rectified system with the PT center tap connected between capacitors will force two equal-valued 250V capacitors to share the 500V equally.

What we are interested in is how the voltage is shared when one capacitor has its maximum leakage current and the other has very low leakage current. Obviously, we are going to have to start with a WVDC for each capacitor that is significantly greater than 250V. Then we can work out the value of the balancing resistors so that the WVDC is not exceeded under the worst case scenario.