Parallel vs. series vs. one filter cap

[Chris333]

Is there any electrical advantage to using either two parallel or two series filter caps instead of one, assuming the total capacitance and voltage handling is equal for all three options?

For example, 2 X 100uf-250V series vs. 2 X 25uf-500V parallel vs. 1 X 50uf-500V (please correct me if these don't each come out to 50uf-500V!).

[Rob S]

Is there any electrical advantage to using either two parallel or two series filter caps instead of one, assuming the total capacitance and voltage handling is equal for all three options?

For example, 2 X 100uf-250V series vs. 2 X 25uf-500V parallel vs. 1 X 50uf-500V (please correct me if these don't each come out to 50uf-500V!).

Hi there,

This is a view and an opinion only and I'll probably contradict myself...

Electrloytics used to have a tolerance of something like + or - 50 to 100 per cent or +150 -50 per cent or something like it - don't know what it is now but say it's much better...

To me, that still means that two capacitors together in series or parallel would combine to be in the worst case nearer to the values they're supposed to be.

Two series capacitors might share an unequal voltage load because of differences in tolerance which could put one of them under overvoltage stress.

Two parallel capacitors would appear to be the best solution but...
then you have two components with twice the possibility of failure and twice the expense.

On a build or a repair, I would use what I've got - on a design you would use what fits or what's cheap and if it's high frequency then you'd need to use low e.s.r. things.

Presumably the capacitors are for a valve amp power supply and in this case (here's the opinion) I don't think it matters - stick a .1 cap bypass across with a 100k resistor and the audiophiles & electricians will be happy

I make it up you know,

Rob.

P.S. On restoring an antique piece of gear I wouldn't use an antique capacitor cos to me it's daft.

[Enzo]

The reason we use two caps in series is to get higher voltage rating. If you can even find 40uf 600v caps, they will be expensive, but a couple 100uf 350 volt caps will be a snap to source at reasonable cost. That is why an amp maker does that. And those two caps might be smaller than the one 600v cap.

I see no particular advantage to doubling them in parallel unless you need more capacitance than is readily available at a price.

Pretty much keeping it simple works best. One cap instead of two works for me. WHen you see multiple caps where one would work, it is almost always done for economic reasons.

Look at Marshall and those 50+50/500v can caps. They but them by the truckload and no doubt get a great price on the volume. They can probably jump the two halves of one of those together for 100uf/500v cheaper than they can separately get a 100/500v cap. Plus the size of the holes and stuff would remain the same as the other caps, keeping tooling cheap.

Even in my shop, if I need a 40/500 axial cap, I might use two 22/500 caps instead becasue I already stock them and get them at a good price. That of course depends upon what I am doing.

[markbul]

In TUT3, K. O'Connor claims that "lead inductance, internal resistance and leakage capacitance" work against the suppression of noise and amplifier stability. "These commodities are smaller in lower value caps than in high capacitance units, so several low-C caps will outperform one high-C cap of the same net value."

His words, not mine. Discuss among yourselves.

[kg]

generally, he's right--the equivalent ESR of multiple, smaller value, paralleled caps will be lower than the ESR of a single larger cap.

whether or not this makes an audible difference i will leave to the ear of the beholder.

[KevinOConnor]

Hi Guys

Whenever the voltage does not force it you should use non-series caps. In the series connection, the bad parts of the cap add together to make a much worse net capacitor but of higher voltage rating than either cap itself. Bleeder resistors MUST be used when series connecting caps as this will ensure proper voltage sharing.

The use of parallel caps works to reduce the bad facets of capacitor performance and thus improve the sound. Remember, the power supply is the other half of the signal path and it is these filter caps that audio currents go through.

NEVER use Illinois Capacitor electros as they are total junk. The next worst type are F&T and any multi-section can cap. All of these are old technology and obsolete for a reason.

Truly excellent long-life low-leakage high-voltage caps are not expensive thanks to the SMPS market.

As I saidin another thread here, the cap voltage rating is "working" voltage, so using a 450V cap at 450V is perfectly fine.

Have fun

[Mike Sulzer]

Really? The cap with 2C value that you must use two of to make value C with higher voltage rating could have a sufficiently low ESR so that the two in series are equivalent to the C with the higher voltage rating. I do not see how you can make a blanket statement, but rather yuo have to look at the ratings.

Hi Guys

Whenever the voltage does not force it you should use non-series caps. In the series connection, the bad parts of the cap add together to make a much worse net capacitor but of higher voltage rating than either cap itself. Bleeder resistors MUST be used when series connecting caps as this will ensure proper voltage sharing.

The use of parallel caps works to reduce the bad facets of capacitor performance and thus improve the sound. Remember, the power supply is the other half of the signal path and it is these filter caps that audio currents go through.

NEVER use Illinois Capacitor electros as they are total junk. The next worst type are F&T and any multi-section can cap. All of these are old technology and obsolete for a reason.

Truly excellent long-life low-leakage high-voltage caps are not expensive thanks to the SMPS market.

As I saidin another thread here, the cap voltage rating is "working" voltage, so using a 450V cap at 450V is perfectly fine.

Have fun

[KevinOConnor]

Hi Guys

Mike, other than the "blanket statement" portion of your post, the rest seems like jibberish and I have no idea what you are commenting on.Or maybe...

If you take typical capacitors that people use in amps, the series ones have their ESR add together. There are sufficiently bad high-voltage caps that could in your example have a same or higher ESR than the two in series, but it is not likely. The OP asked if there are technical reasons for series/parallel/single/doule caps and there are, as I cited above.

The best amps made mostly use parallel caps, and in hifi it is more often than not.

I guess I understand jibberish afterall.

Have fun

[Mike Sulzer]

Really? 13 100 micro f 450 v capacitors found in a catalog have an average ESR of 1.88 ohms. The two 220 v 220 micro f caps I found had .905 ohms ESR. Seems like a series connection of two can be just as good. Or maybe not if you choose the wrong capacitors.

Hi Guys

Mike, other than the "blanket statement" portion of your post, the rest seems like jibberish and I have no idea what you are commenting on.Or maybe...

If you take typical capacitors that people use in amps, the series ones have their ESR add together. There are sufficiently bad high-voltage caps that could in your example have a same or higher ESR than the two in series, but it is not likely. The OP asked if there are technical reasons for series/parallel/single/doule caps and there are, as I cited above.

The best amps made mostly use parallel caps, and in hifi it is more often than not.

I guess I understand jibberish afterall.

Have fun

[Raybob]

I just finished a 'Vibrolux" based build. For the heck of it, I tried something different with first stage filter caps and ended up with the quietest idling amp I've built. I use 100@100 for bias filtering, but the big difference was trying series 80s with 220K resistors, as in a Super Reverb. Schematic for Vibrolux calls for same parallel 16s to get 32 that is in Deluxe Reverb. I usually use 2x22s parallel but tried 80s in series. My usual Vibrolux is around 51db measured at front of speaker, Deluxe around 49db, but this Vibrolux w/parallel 80s reads 46db! Quietest I've ever seen except smaller builds as Princeton or Champ styles.

[Enzo]

Try measuring those combinations - OUT of the circuit - for value and ESR. Cap tolerances are loose, so your series 80s may have a larger net capacitance than the parallel 22s. For example.

[soundmasterg]

The next worst type are F&T and any multi-section can cap. All of these are old technology and obsolete for a reason.

Are you saying the F & T axial caps are old technology just because they are bigger than some others out there? I've had good reliability with the F & T's in repairs and they aren't so outrageously expensive like the Sprague Atoms are so that many people can afford to use them. I often like Panasonic caps and the Webers are actually pretty good too, but I do use the F & T on occasion and haven't had one issue whatsoever with them.

Greg

[KevinOConnor]

Hi Guys

I wouldn't use the Spragues or the F&Ts or any other multi-section cap. Being old technology designs they are not as good as new-technology electrolytics - with electros newness is a benefit in their design and manufacture. Every electro will last a long time if it has regular application of voltage, as I stated earlier. This just means they will last a long time but does not mean that their DA does not rise over time - which it does - making them less ideal as they age.

The multi-section caps when used as intended compromise good grounding. If you are doing a repair and it is a multi-section cap that is at fault, snip its connections to the circuit and install a new cap inside the chassis. This maintains stock appearance but provides much better filtering.

In Raybob's experience above, the new caps just by being new can sound better than the old ones regardless of how either pair was/is wired. If all the caps are new, then you have to consider the specs of each cap as Enzo pointed out. There might also be some minor change of grounding that occurred that was not noticed, and that could be the aspect of the mod that made the improvement. As TUT shows, simply changing the orientation of the rectifier connection to the first filter and the power take-off from that cap can reduce hum or buzz by 10dB.

Have fun

[Raybob]

...In Raybob's experience above, the new caps just by being new can sound better than the old ones regardless of how either pair was/is wired. If all the caps are new, then you have to consider the specs of each cap as Enzo pointed out. There might also be some minor change of grounding that occurred that was not noticed, and that could be the aspect of the mod that made the improvement. As TUT shows, simply changing the orientation of the rectifier connection to the first filter and the power take-off from that cap can reduce hum or buzz by 10dB.

Have fun

Actually, I'm speaking of new amp builds. I have been using new F&T electrolytics with no problems but I'm referring to a Vibrolux build, which I've done 15 or 20 new bulds, most getting 2x22uF caps series as per schematic. I did the last build using 2x80s in series with 22K resistors as per AB763 Super Reverb, and it was definitely 3db quieter than any of the other builds measured using 2x22uFs parallel. All new caps, all using same grounding scheme using star for 1st stage filters/tubes, buss bar.for preamp grounds/filters I will be seeing this amp before it leaves for Ventura and will at least measure total uF at 1st stage.

[J M Fahey]

definitely 3db quieter than any of the other builds measured using 2x22uFs parallel. All new caps, all using same grounding scheme

Measuring beats imagining any day of the week