Looking at power caps I see that I could pay more for higher ripple current values. I know it’s an important spec, but what’s a good value and what’s unnecessary?
It is not so much 'what is a good value' but rather a case of what do your requirements actually need.
You can always say higher is better but what if the part will not physically fit.
A Class AB amp at low output (which is what we listen to most of the time) will see only a fraction of the ripple current that would occur if the amp was delivering say 100wrms into a load.
Also look at cap lifetimes and temperatures because sometimes there could be a 10 to 1 difference in lifetime (which is usually quoted at max ripple and temperature).
So no easy or quick answer... you need to know what currents and temperatures are expected before making a choice.
You can always say higher is better but what if the part will not physically fit.
A Class AB amp at low output (which is what we listen to most of the time) will see only a fraction of the ripple current that would occur if the amp was delivering say 100wrms into a load.
Also look at cap lifetimes and temperatures because sometimes there could be a 10 to 1 difference in lifetime (which is usually quoted at max ripple and temperature).
So no easy or quick answer... you need to know what currents and temperatures are expected before making a choice.
Depends on average and peak current draw and mains frequency.
Also the metric for peak current in an AC waveform doesn't necessarily mean you will have the current at the point of demand. For example electrolytics and the rate of their chemical reaction may mean you see a delay. However measuring that is overkill given the remainder of domestic power issues.
A large cap will start heating from the inside out. Compare that with a set of smaller parallel caps that total the same capacitance, the result is higher current capacity, lower ESR and cooler running. If you have the space.
Also the metric for peak current in an AC waveform doesn't necessarily mean you will have the current at the point of demand. For example electrolytics and the rate of their chemical reaction may mean you see a delay. However measuring that is overkill given the remainder of domestic power issues.
A large cap will start heating from the inside out. Compare that with a set of smaller parallel caps that total the same capacitance, the result is higher current capacity, lower ESR and cooler running. If you have the space.
My application would be one if the First Watt clone Class A amps. So, fairly high temperature and current I think.
There's a couple of thing I think about w.r.t temperature:
* if you imagine the chassis of the amp - cooler at the bottom and gradually getting warmer you move up the amp chassis to the top. I would look at positioning the caps lower - also if there's air vents at the bottom, a cap over the top of the vent helps cool it.
* radiated heat - tubes radiate heat, so do resistors so putting caps next to tubes, resistors or heatsinks will reduce lifespan and increase noise.
* if you imagine the chassis of the amp - cooler at the bottom and gradually getting warmer you move up the amp chassis to the top. I would look at positioning the caps lower - also if there's air vents at the bottom, a cap over the top of the vent helps cool it.
* radiated heat - tubes radiate heat, so do resistors so putting caps next to tubes, resistors or heatsinks will reduce lifespan and increase noise.
SoaDMTGguy, I'm guessing you won't know what the ripple current level is in any of the amps, and you won't be able to measure (or otherwise simulate) the ripple current in an amp you may end up making?
If so, then imho it doesn't matter for you as you have no metric for assessing the expected service life of a cap in your amp, so either have to wait for it to ooze or show distress, or disconnect and measure its ESR and capacitance at periodic points in its life (including at the start). Even if you somehow determined the ripple current, and use a cap with a known service life in hrs (making a few assumptions along the way), you then should really add that info to the amp's documentation, and add an entry in your calendar to measure or replace the cap after a certain number of hrs operation or years of your life.
The caveat is that the original designer has chosen an e-cap value that for any common cap it will have a ripple current rating in excess of the ripple experienced in the amp.
If so, then imho it doesn't matter for you as you have no metric for assessing the expected service life of a cap in your amp, so either have to wait for it to ooze or show distress, or disconnect and measure its ESR and capacitance at periodic points in its life (including at the start). Even if you somehow determined the ripple current, and use a cap with a known service life in hrs (making a few assumptions along the way), you then should really add that info to the amp's documentation, and add an entry in your calendar to measure or replace the cap after a certain number of hrs operation or years of your life.
The caveat is that the original designer has chosen an e-cap value that for any common cap it will have a ripple current rating in excess of the ripple experienced in the amp.
Last edited:
It is very rare for 50/60Hz *audio* amplifiers to have trouble with ripple heat.
We design for very LOW ripple so we do not have buzz in the sound. We pick very very large capacitors, both value and physical size, and the size helps shed the heat.
For most audio power amps, AB design makes for very LOW idle current and thus low ripple current. In speech/music, the time spent at MAXimum power is very very small and will not add much heat. (Concert amplifiers audibly clipping on peaks still average only 10% of rated power.)
The capacitor makers know what you are likely to do. Even "junk" caps can work very well in a DIY world where it is all for fun anyway. (Or self-repaired.)
The ripple ratings matter more for "big ugly DC power" like welders and lighting, where ripple is much less important and cost of large capacitor banks can be high.
We design for very LOW ripple so we do not have buzz in the sound. We pick very very large capacitors, both value and physical size, and the size helps shed the heat.
For most audio power amps, AB design makes for very LOW idle current and thus low ripple current. In speech/music, the time spent at MAXimum power is very very small and will not add much heat. (Concert amplifiers audibly clipping on peaks still average only 10% of rated power.)
The capacitor makers know what you are likely to do. Even "junk" caps can work very well in a DIY world where it is all for fun anyway. (Or self-repaired.)
The ripple ratings matter more for "big ugly DC power" like welders and lighting, where ripple is much less important and cost of large capacitor banks can be high.
The NP amps are class A. One example write-up (Aleph J illustrated build guide) indicated the output stage idles at 1.7A, and the first filter caps are 2x 15mF (per polarity) or 1x 33mF, so caps could idle with a few amps ripple current each. I just checked one 33mF e-cap identified in the diyaudio thread (Panasonic T-UP) and it has a 7A ripple 85C rating at 3khrs, so not a large margin and well worth keeping the cooling air to the first filter caps and setting your calendar if you are going to butt caps up against each other and use a closed chassis (ie. based on assuming the heatsinks on the outside are all that matters).
Last edited: