CAPS - leakage and ripple: Of the two which is most important?

[lowvolt]

Mouser and Digikey can drive a man to tears.  The mountains of specs is maddening!

Sifting through capacitor choices, my first priorites are usually (in no particular order):

** Tolerance.
** Voltage.
** Load Life.
** Physical Dimensions.
** Lead spacing.

So then I'm left with a number of candidates, most times the differences are ripple and leakage.  For stompbox-type compression, overdrive, and distortion circuits which if the two are most important to keep an eye on?  And which is better .... "less" or "more"?  I would assume that less leakage and less ripple is more desireable than more of either one.

And out of curiousity, what about power filter circuits?  I just recapped the input power stage of my Hammond organ preamp recently (1962 A102).  I went the extra mile and used Nichicons, all load-life rated at 2khrs+ at 105c.  These were all 350 to 500 volt installations btw, I made certain to overspec the voltages a bit over stock.  But I kinda just threw a dart when it came to ripple and leakage.

[brett]

Hi
I'm no expert, but for Power filtering, it's good to have low ripple. But....the easiest way to lower ripple is to increase the size of the cap or put a little resistance (as low as 1 ohm) in-line ahead of the filtering cap.
A rough rule-of-thumb is that unregulated, power-hungry circuits (such as power amps) want at least 2200uF per amp of draw. For sensitive, low-power devices such as stompboxes, you will want about 1000uF per amp to get a reasonable supply (a few % ripple), then a regulator and smaller caps to achieve a tiny amount of ripple (a tiny fraction of a %).
I can't see leakage being a problem in the sorts of applications we have here (in any case it's very low in modern caps). But as I said, I'm no expert, and someone will probably know of somewhere where it's an issue.
cheers

[R.G.]

Mouser and Digikey can drive a man to tears.  The mountains of specs is maddening!

They have chosen to tell you everything you might want to know, not just a few "grabbers" to make the parts sound cool. There is no such thing as having too much information available. The devil is always in the details.

Sifting through capacitor choices, my first priorites are usually (in no particular order):
** Tolerance.
** Voltage.
** Load Life.
** Physical Dimensions.
** Lead spacing.

Hmm. I'm guessing you have already taken capacitance value into account, and all the caps you're comparing are the same nominal value. Taking this in the order you put them:
tolerance - matters for filtering accuracy or timing accuracy. Almost unimportant in choosing power supply filtering caps, where the idea is to just get enough. Aluminum electro caps are pretty incapable of being close tolerance, because they drift as they age. For power filtering, use electros and get as much capacitance, then voltage in the physical package that will fit.

Voltage: in all case, you just have to have enough. More than 30-40% high is wasted in many cases, even for surges. There is no virtue in getting aluminum electros at 2-3 times the necessary voltage unless (1) they don't make them that low, as in 1uF, where 50V is about as low as you can get, or (2) you have no good idea how high the voltage may be.

Load life: always good to know. Better to pick high temperature (105C instead of 85C) if this matters, but mostly effects DIY will never test the load life limits, because we don't work them hard enough or hot enough for them to be what fails.

Physical dimensions and lead spacing: important! it's gotta fit where you're trying to put it.

So then I'm left with a number of candidates, most times the differences are ripple and leakage.  For stompbox-type compression, overdrive, and distortion circuits which if the two are most important to keep an eye on?  And which is better .... "less" or "more"?  I would assume that less leakage and less ripple is more desireable than more of either one.

Ripple specs are probably "ripple current", which in turn is a way of stating how much AC current can flow through it without it heating up from the current going through the ESR, which is almost never specified. Ripple current does matter with filter caps right after rectifiers, and for precious little else in pedals. The first filters in the power supply of a DIY power amp need to be rated for high ripple, but not much else we do here. Leakage is the last thing I'd worry about unless I was building high voltage stuff (as I did recently in making an 800Vdc supply for a geiger counter) or ... well, that's about all. The leakage on a new cap of the right type is going to be so low it won't matter. If you have an application where it matters, you already know why. If you have to pick, go for high ripple rating (that is, high ability to withstand lots of current) and low leakage.

And out of curiousity, what about power filter circuits?  I just recapped the input power stage of my Hammond organ preamp recently (1962 A102).  I went the extra mile and used Nichicons, all load-life rated at 2khrs+ at 105c.  These were all 350 to 500 volt installations btw, I made certain to overspec the voltages a bit over stock.  But I kinda just threw a dart when it came to ripple and leakage.

High voltage power filter circuits are where leakage and ripple matter. However, the currents are so low in something like a tube preamp, the ripple is probably OK no matter what you get. Picking 105C already means other things are made better, so that's good. Load life rating is nice, but as with so many things, if I had to pick no cap or a cap not rated for long life, I'd buy the cap and use it while I looked for the perfect one in a more leisurely pace.

[lowvolt]

Mouser and Digikey can drive a man to tears.  The mountains of specs is maddening!

They have chosen to tell you everything you might want to know, not just a few "grabbers" to make the parts sound cool. There is no such thing as having too much information available. The devil is always in the details.

Sifting through capacitor choices, my first priorites are usually (in no particular order):
** Tolerance.
** Voltage.
** Load Life.
** Physical Dimensions.
** Lead spacing.

Hmm. I'm guessing you have already taken capacitance value into account, and all the caps you're comparing are the same nominal value. Taking this in the order you put them:
tolerance - matters for filtering accuracy or timing accuracy. Almost unimportant in choosing power supply filtering caps, where the idea is to just get enough. Aluminum electro caps are pretty incapable of being close tolerance, because they drift as they age. For power filtering, use electros and get as much capacitance, then voltage in the physical package that will fit.

Voltage: in all case, you just have to have enough. More than 30-40% high is wasted in many cases, even for surges. There is no virtue in getting aluminum electros at 2-3 times the necessary voltage unless (1) they don't make them that low, as in 1uF, where 50V is about as low as you can get, or (2) you have no good idea how high the voltage may be.

Load life: always good to know. Better to pick high temperature (105C instead of 85C) if this matters, but mostly effects DIY will never test the load life limits, because we don't work them hard enough or hot enough for them to be what fails.

Physical dimensions and lead spacing: important! it's gotta fit where you're trying to put it.

So then I'm left with a number of candidates, most times the differences are ripple and leakage.  For stompbox-type compression, overdrive, and distortion circuits which if the two are most important to keep an eye on?  And which is better .... "less" or "more"?  I would assume that less leakage and less ripple is more desireable than more of either one.

Ripple specs are probably "ripple current", which in turn is a way of stating how much AC current can flow through it without it heating up from the current going through the ESR, which is almost never specified. Ripple current does matter with filter caps right after rectifiers, and for precious little else in pedals. The first filters in the power supply of a DIY power amp need to be rated for high ripple, but not much else we do here. Leakage is the last thing I'd worry about unless I was building high voltage stuff (as I did recently in making an 800Vdc supply for a geiger counter) or ... well, that's about all. The leakage on a new cap of the right type is going to be so low it won't matter. If you have an application where it matters, you already know why. If you have to pick, go for high ripple rating (that is, high ability to withstand lots of current) and low leakage.

And out of curiousity, what about power filter circuits?  I just recapped the input power stage of my Hammond organ preamp recently (1962 A102).  I went the extra mile and used Nichicons, all load-life rated at 2khrs+ at 105c.  These were all 350 to 500 volt installations btw, I made certain to overspec the voltages a bit over stock.  But I kinda just threw a dart when it came to ripple and leakage.

High voltage power filter circuits are where leakage and ripple matter. However, the currents are so low in something like a tube preamp, the ripple is probably OK no matter what you get. Picking 105C already means other things are made better, so that's good. Load life rating is nice, but as with so many things, if I had to pick no cap or a cap not rated for long life, I'd buy the cap and use it while I looked for the perfect one in a more leisurely pace.

Thanks for the help.  I suppose I should have been a bit more clear.  I'm well aware of what the first few specs I mentioned address, and I usually pick the better spec'd caps in those areas, but ripple and leakage were the two that I wasn't sure about.  As I said the first few that I listed are the proority specs I look for when selecting caps for purchase.  I aim for the lowest tolerance variance available, the proper voltage rating, the longest load life/at highest temps, obviously physical dimaensions, and lead spcing.  I ALREADY KNOW WHAT THOSE SPECS MEAN AND I (forgive the capaital letters, my keyboard is trying to die .. it's very annoying!) use them for each batch of caps I buy.  But ripple and leakage were the remaining specs left over that were different between a given set of caps when comparing them for purchase.  You've answered that issue completely and thoroughly.  Thank you. 

[R O Tiree]

I found this document a while ago when I was trying to come up with a justification for buying more expensive (reliable?) electrolytic caps for the stomp-boxes I was building for people. Halfway down the first column of page 14, there is a relatively simple equation for predicting the expected lifetime of an Al electro. By far the biggest factor in extending the life quoted by a manufacturer is operating temperature compared with rated temperature (every 10 degree Celsius below rating more or less doubles the life) but de-rating the voltage has a further appreciable effect.

A quick play with the numbers reveals that a cap rated at 50V but only being fed 9V can be expected to last 3.5 times longer than one that is "just enough". It must be stressed that this was an empirical result from a study by Cornell Dubilier into the failure rates of their caps. Other manufacturers' mileages might vary.

[lowvolt]

Yea, I knew about the "double load life for every ten degrees below rating" thing.  That's why temp rating is one of my priority specs.  What my thing is, is that once that list of important specs I posted is used to sort out a return of possible contenders, the last two issues were ripple and leakage.  But from what I gather from replies here, those two specs seem to be of little consequence when used in these little tinkertoy circuits we play with for use with our electric guitars and basses.

As far as rated tolerance goes, I just checked about 250 caps I have by a ~better brand~ that list a 20% tolerance ..... not one of them exceeded 5% +/- of advertized capacitance.  When compared to caps rated at 10% tolerance, I found no better results with the 10%-ers over the 20%-ers.  In fact, out of 250 20% caps I tested all but one tested within 3% of rated capacitance.  It pays to stick with mo-gooder-stuffs I guess.