Capacitors: Who is Right? What ages them faster?

Power supply electrolytics - the main thing audiophiles worry about getting old and having to replace.

What's the truth here?

Manufacturers can be quoted as saying the life expectancy of caps is anywhere from 10 - 15 years. Is this true? Why is it that Krell amps seem to need capacitor changes in 8 years, while people have receivers from the 70's that never had caps replaced?

Another controversy, with three viewpoints:

(1) I was told that caps age faster in amps that are left on continuously. The rationale here is that the lack of a discharge cycle is not good for the caps since they are in a constantly charged state.

(2) Others claim that leaving a unit unused in a box for years also shortens the life of the caps, as they never see a charge.

(3) Others say the lifespan is the same whether on the shelf, or in the unit.

Who has the technical background or the knowledge to sort this out?
My educated guess is that none of 1, 2, or 3 are correct, and that what really causes cap aging is thermal cycling. Krells tend to run hot (never owned one, but they have huge heatsinks for a reason), so this might shorten the life.

Leaving equipment off for a long period is probably OK, leaving it on for a long period is probably OK, but switching on and off every day is probably not so good due to the thermal cycling, which causes mechanical stresses (expansion and contraction).

The lifetime will also vary with the quality of the parts.

So if I were you I'd leave all solid state stuff on all the time to minimize temperature variations and condensation problems.

Another thing to bear in mind is that some manufacturers and repair houses make a lot of money swapping caps, and so it pays them to put fear into the audio community that their caps are bad and need changing. By analogy, try to find a car mechanic who'll tell you that oil really doesn't need to be changed every 3000 miles.
I have seen data sheets that give operational life as much as 300,000 hours (34 years), with a shelf life as high as 10 to 15 years "...without deterioration of quality". What will affect the life span is the ambient temperature (if too high, the caps will start leaking current), the ripple current frequency and the initial temperature rise (on/off operation). If you leave the amps off, then you increase the life span by virture of zero frequency but lose it when you turn on the amps (high initial temp rise). If you leave the amps on, then the reverse is true - so, to me, it's a wash whether you keep the amp on or off.

But 8 years to a ps filter cap? Doesn't sound right to me, nor have I heard that Krell caps have such a problem; however 34 plus years for a cap is, IMO, a bit on the optimistic side. I would guess that 15 to 20 yrs would be in the ballpark. Then again, my ML 20.6 pc boards were replaced twice in 15 years. Because of bad caps.
What does this mean with regard to those amps that use few large value caps as opposed to those amps that use more smaller value caps?
One of my collegues does research on capacitor reliability (for aircraft systems) and he agrees that thermal cycling is the main problem. I will ask him again now that he has had time to look more into the issue and let you know. In my experience with using caps in designs is that 10 to 15 years is typical in 55C ambient conditions (for electrolytics).

As for the on/off bit, I have thought about it at length since I am a thermal guy by education and have decided that off once a week and when I leave town is the best balance.

For large value caps, the thermal lag time is longer so perhaps they normally last longer than smaller (physically and value) ones. I will ask my buddy tomorrow about this too if I see him. Arthur
I spoke to Nelson Pass today. He said that, over time, the fluid in the cap slowly dissipates from the cap, until it eventually dries out. He believes that if the amp is left on, the higher ambient temperature is accelerating the process, but by how much, he didn't seem sure.

Your colleague is probably more knowledgeable than Pass is on this issue, since this is his profession. I'll be very interested in his answers......
Kevziek: Most of Nelson Pass' amps are of a high bias design and run hot. Leaving them on exposes the caps to a greater amount of heat, so they will end up "cooking" themselves and drying out. Then again, if one turns the amp off and on as needed, the caps are exposed to big temperature swings from the amp being stone cold and then coming up to operating temperature. Doing this repeatedly stresses the caps more than leaving them turned on all the time. That is, so long as the amp has proper ventilation both internally and externally to begin with.

This is not to mention that leaving them on reduces the amount of exposure to high levels of in-rush current. It is in-rush current that damages equipment the most due to the huge surges involved and the rapid internal temperature changes that take place when this occurs. This is why light bulbs typically blow when turning them on, not once they are already on.

Other than that, Seandtaylor did a nice job. GS's comments are right in the ball park too, but one has to realize that not all caps are made the same or have the same spec's and reliability. For that matter, caps from the same manufacturer and of the same product line may differ due to variances in the dielectrics used from batch to batch. Some batch's of products are much better / reliable than others.

As a side note, just a while back i pulled some caps from an amp that Nelson designed and released in appr 1990. While this amp is 13 years old and idles at about 130* at the heatsink, it has been on 24/7 for over three years. With all of that in mind, the electrolytics in the power supply actually tested above their rated values by almost 20%. Needless to say, i didn't replace them as there was no need. Sean
Sean: I don't know if you are correct about the caps being exposed to high heat. Nelson said that he has always run the caps considerably below the rated temps. He mentioned 55 degrees C, while ratings are 85 C. The heat sinks may be absorbing the bulk of the heat, while the internal components remain cooler. I'm not sure on this, though.
Hi guys, very interesting subject. My system uses four Linn Klouts in an active mode (electronic crossover inside the amp), each amp has eight 63DCV/10,000uf capacitors. This amps run very hot, I have always wondered if the heat or age would damage or wear them. Could someone please advise me on how to check them? Can it be done with a digital multimeter? Thanks.
All the time, and to allow a minimum of 48 hours for settling prior to listening??

This is in the Threshold owners manual. Just a thought.

Kevziek: While the heatsinks and output devices will obviously get hotter than anything else, the ambient temperature inside of the amp is also raised as the amp comes up to temperature. I'm not claiming that ANY amp runs the caps too hot, because if they did, that would be an underdesigned product. We all know that we don't have any of those floating around and that i would be the last to comment on something like that : )

What i am getting at is that the caps "cooking to death" will be a long, slow process. While the caps are heated, they aren't heated enough to exagerate their failure rate. In effect, Nelson verified this with his comments saying that their operating temp was measurably below their rated specs.

As such, you've now heard the answer to the question, albeit in a confused and round-about manner, directly from Nelson himself. That is, the caps aren't run up against their thermal limits when turned on. This means that they aren't stressed. At the same time and using common sense, leaving them on keeps them stabilized in terms of thermal shock AND reduces the potential for blowing holes in the dielectric via removing in-rush current from the equation. Sean
Sean - 130* at the heatsink is WAY too high for any amp to operate. Most transistors can't operate above 120*C without burning up. Including thermal impedance effects, 130* at the heatsink means about 150*C at the junction (in a good design) which is impossible. Your temp measurement needs to be verified.

Capacitors are even worse: many can't function over 85*C. This is a problem I face nearly every day in my applications. the darn caps are typically the thermal limiter in many designs and so they must be kept as far away from the transistors as possible. Caps have a low loss character so they aren't too affected by the load - time is the bigger problem since the heat present around them exacerbates the drying process.

Also, the temperature rise caused by cranking the volume cannot be overlooked as may have here. The initial temp rise at turn-on is only half the battle (less than that even in class B designs) - the other half is when you turn the volume up due to I^2*R conduction losses in transistors. Since current is squared, the volume control results in non-linear heating which can overcome initial turn-on effects rather easily. A much lower time constant driven by high power supply energy in this case is also to blame.

Anyway, I haven't seen my buddy yet but I will start a thread after I talk to him. Cheers! Arthur
Ah, perhaps Sean meant 130*F... I had long forgotten about Fahrenheit so I didn't think of it till now. Otherwise, see my reply above.
Aball: Celsius is for use by "fureners", not us folks here in the USA : ) Sean
Sean, in my conversation with Nelson, he indicated that the caps would probably dry out faster when the amp is on all the time, probably due to the heat, as you suggest. Yet, he seemed to feel that you shouldn't worry about life span for probably 15 years, in his designs, anyway. How much of this is conjecture on his part, I don't know. That's why I eagerly await Aball's buddy's opinion.