Putting together the BOM and ordering parts for yet still another huge PA amp build. I’m considering replacing electrolytics in a couple of places - just to remove the main points of failure years down the road. My experience has been that it’s all the little 1 and 10uF electrolytics in the support circuitry that go out and not the power amplifier itself. None would affect the sound in any way, but are mission critical. The timing caps in the soft start and DC protect, hold/averaging capacitors on the limiters, and the coupling and boot strap caps on the rail switches. In any of these places, if the cap gets flaky the amp starts having problems. Normally I just use electrolytics, but considering going to something different for reliability. Is X5R up to the task? COG isn’t practical in that cap range, and neither is polypropylene. High voltage, mechanically stout Z5U discs and such have gone the way of the dinosaur. X7R is a bit more pricey but doable. Are these low voltage ceramics ok ESR and reliability wise, or would I just be better off planning on recapping every 5 years? I never seem to have problems with the big computer grades lasting 20 or more years, but the stupid little 10 mikes even go bad on the shelf.
Go X7R if you can and consider that the capacitance varies significantly with voltage applied. 10uF of X5R is going to be functionally less than the same electrolytic. You might look at tantalum if you want long life, or solid polymer caps.
There is not much difference between X5R and X7R except that X7R can work up to 125 degrees C and X5R only up to 85 degrees C, that X7R is bigger and more expensive and only slightly less non-linear.
Apart from the non-linearity (like 50 % capacitance loss at half the nominal voltage, 80 % at the nominal voltage), a +/- 10 % to +/- 20 % initial tolerance and max. +/- 15 % capacitance variation over the temperature range, there is also a long term drift. The nominal value usually applies 1000 hours after the last time the capacitor was heated above its Curie temperature (as happens during manufacturing and can also happen during soldering).
The capacitance reduces by a certain percentage for each factor of ten increase of the time since the last time the capacitor was heated above the Curie temperature. For example, with - 5 %/decade, the capacitance would be 10 % higher than specified 10 hours after heating and 10 % lower than specified 100 000 hours after heating above the Curie temperature.
Apart from the non-linearity (like 50 % capacitance loss at half the nominal voltage, 80 % at the nominal voltage), a +/- 10 % to +/- 20 % initial tolerance and max. +/- 15 % capacitance variation over the temperature range, there is also a long term drift. The nominal value usually applies 1000 hours after the last time the capacitor was heated above its Curie temperature (as happens during manufacturing and can also happen during soldering).
The capacitance reduces by a certain percentage for each factor of ten increase of the time since the last time the capacitor was heated above the Curie temperature. For example, with - 5 %/decade, the capacitance would be 10 % higher than specified 10 hours after heating and 10 % lower than specified 100 000 hours after heating above the Curie temperature.
If you can stand 1000 buy, Aerovox & Kemet make COG 1,2.2,4.7,10 uf caps. But nobody stocks them. I got some Aerovox gold 10 uf 50 v COG on closeout from Newark years ago and they sounded really good as coupler caps in my organ. Were over $5 each though. replaced dried out electrolytics and shorted tantalum, there was no room for 63 v film caps.
Maybe we ought to do a group buy. I'd take a dozen 10 uf.
I've used 4.7 uf 100 v x7r as input caps to my power amps. But I won't brag about it. My theory is the higher the rated voltage, the less of the voltage curve you are using with a 2 v signal. Put some in last Thursday for 2.2 np electrolytic in a PV-4c amp. I hate doing jobs over & over again. The speakers I'm using are used Toshiba 6" projection TV whizzer cones in ported newark parts boxes, so those would cover up 2% distortion, anyway. Sounds waaaay better than the built in speaker in this flat screen Samsung TV.
Maybe we ought to do a group buy. I'd take a dozen 10 uf.
I've used 4.7 uf 100 v x7r as input caps to my power amps. But I won't brag about it. My theory is the higher the rated voltage, the less of the voltage curve you are using with a 2 v signal. Put some in last Thursday for 2.2 np electrolytic in a PV-4c amp. I hate doing jobs over & over again. The speakers I'm using are used Toshiba 6" projection TV whizzer cones in ported newark parts boxes, so those would cover up 2% distortion, anyway. Sounds waaaay better than the built in speaker in this flat screen Samsung TV.
Last edited:
Tantalum... something I hadn’t thought about. I originally dismissed it because the coupling caps I was worried about in the rail switches might see reverse bias at startup, and do see an amp of peak current twice per cycle. But they could probably deal with a 50% cap change. As long as it holds much more charge than the Ciss of the hexfet being switched. The 555 timers and compressors would prefer the cap value be stable with voltage - but there is no possibility of reverse bias or large transients. Guess there is no reason they all have to be the same type.
Every tantalum in my 1960's equipment has been bad. !!!! That is dozens of them. By date code, some were changed by factory service after 2 years. The replacement tantalums I bought in 1985 had popcorn noise. So I am no tantalum fan. The new ones I bought didn't even have a quoted service life on the datasheet. So they are supposed to work how long? I've put in over a hundred 10000 hour nichicon & rubicon 4.7 and 10 uf aluminum electrolytic . Some of them are over 6 years old & doing fine. The ones in the FM radio get maybe 30 hours a week of use.
David of NC on organforum said the green epoxy sealed CDE electrolytic caps in Fender amps of the seventies never fail. He also said the 4 uf sardine can electrolytics in Hammond tone cabinets of the 1940's are all still good. So we are buying short lived trash because that is what the market wants!
David of NC on organforum said the green epoxy sealed CDE electrolytic caps in Fender amps of the seventies never fail. He also said the 4 uf sardine can electrolytics in Hammond tone cabinets of the 1940's are all still good. So we are buying short lived trash because that is what the market wants!
Last edited:
I’m not sure 1940s or 1960s capacitor technology is still worth considering. MnO2 tantalum and polymer tantalum both should generally last longer than a normal wet aluminum electrolytic if used and rated appropriately. Tantalums are used in high reliability applications often, whereas you will not see many aluminum electrolytics. I have read several places that MnO2 tantalum caps don’t have a wear-out mechanism. This may not be true, though. This paper has some good info:
http://www.kemet.com/Lists/Technica... of High-Voltage MnO2 Tantalum Capacitors.pdf
also https://ec.kemet.com/wp-content/upl...ircuit_with_wrong_decoupling_capacitor_R1.pdf
Solid polymer aluminum electrolytic caps may be good too. I know the hybrid caps have higher ratings but 20 years from now I’d put my money on a true solid polymer rather than a cap that still has a liquid electrolyte.
Last edited:
If you’re going to expose them to reverse bias or high ripple currents I’d probably not use MnO2 tantalum caps, or be very careful in derating. Polymer types should be safer.
Ceramic is very reliable but they do lose capacitance with age as well. Ceramics should be fine if you can tolerate the huge voltage and temperature coefficients of capacitance.
Last edited:
I avoid tantalum capacitors whenever I can because tantalum ore is a conflict mineral (see Coltan - Wikipedia ) and the datasheets almost never mention where the tantalum comes from. Unfortunately the same holds for tin and gold, but I don't know any good alternatives for those.
Kemet claims to source 100% conflict-free tantalum, but I don't know the exact details.
conflictfree
I’m thinking ceramic is the way to go in the pulse coupling position. I ordered 22 uf, to go where 10 uf aluminum electro used to be. It has to deal with charging 13000pF worth of hexfet input capacitance thru a 15 ohm resistor. And has to do it at the audio rate. Mostly subwoofer trequencies, but I don’t want any stupid behavior at 20kHz either. I was somewhat concerned about the 25 rating on caps that big, but it seems the rating is more driven by capacitance loss than by “it will blow up if you hit it with 30 volts” (which is 2X what it would normally see) or “one amp for 200ns at a time is bad news”.
The other target applications won’t see much stress, but I really don’t want to deal with caps getting leaky over years of use. A 1 megohm charging 1 to 2 uf on a 555 timer for instance - works pretty well initially. Then it takes longer and longer for the relay to kick in, and eventually it just won’t anymore. I’ve had to replace too many of them, both in DIY and store bought amp supervisor circuits and I’m sick of it. I doubt a tantalum would catch on fire, but if it’s just as unreliable as aluminum I’d rather use something else. Stay with the ceramic (with only 5 volts on it where it ought to have well over half it’s capacitance) or go to a fractional-uf green mylar chicklet and up the resistor to 10 meg?
The other target applications won’t see much stress, but I really don’t want to deal with caps getting leaky over years of use. A 1 megohm charging 1 to 2 uf on a 555 timer for instance - works pretty well initially. Then it takes longer and longer for the relay to kick in, and eventually it just won’t anymore. I’ve had to replace too many of them, both in DIY and store bought amp supervisor circuits and I’m sick of it. I doubt a tantalum would catch on fire, but if it’s just as unreliable as aluminum I’d rather use something else. Stay with the ceramic (with only 5 volts on it where it ought to have well over half it’s capacitance) or go to a fractional-uf green mylar chicklet and up the resistor to 10 meg?
Ah, yes, timer circuits. I've been using ceramic 1 uf up in timers since the late seventies. Those circuits are still working, whereas the power up silence circuit in Allen organs pretty much silences all of them by 25 years due to the electrolytic cap they used.
Voltage modulation of the capacitance is not critical in non-amp circuits.
My use of 50 v ceramics in amps for couplers for 2 v signals indicates to me the fear of them is oversold. Any affordable speaker covers distortions in the 2nd decimal place and probably the 1st also.
Voltage modulation of the capacitance is not critical in non-amp circuits.
My use of 50 v ceramics in amps for couplers for 2 v signals indicates to me the fear of them is oversold. Any affordable speaker covers distortions in the 2nd decimal place and probably the 1st also.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.