Electrolytic capacitors types for power supplies come and go. They attend different demands than stages interface or other more subtle applications.
Some are large and recommended for main raw filtering, usually before regulators. Some are small and have to take care of smaller currents.
Raw supply filtering usually requires low-esr and high temperature, even if large cans rarely fulfill any of those. That's why some people use smaller caps wired in parallel. Caps after regulators usually DON'T have to be low-esr.
The question is which brand types have worked for you in both places.
Some are large and recommended for main raw filtering, usually before regulators. Some are small and have to take care of smaller currents.
Raw supply filtering usually requires low-esr and high temperature, even if large cans rarely fulfill any of those. That's why some people use smaller caps wired in parallel. Caps after regulators usually DON'T have to be low-esr.
The question is which brand types have worked for you in both places.
I find that providing adequate MF and HF supply rail decoupling at the load removes the need for anything fancy/exotic at the main smoothing (or bank of) capacitors.
Standard commercial grade capacitors that meet specification are good enough.
Omit the supply rail decoupling and you have a completely different operating regime.
Transients at the load demand extra current and the long supply lines prevent most of that demand being met. This is audible. Fiddling about with capacitors has become audible, because they try to send transient current through the long inductive connections. Sort the decoupling and this dependance on the PSU dissappears.
Standard commercial grade capacitors that meet specification are good enough.
Omit the supply rail decoupling and you have a completely different operating regime.
Transients at the load demand extra current and the long supply lines prevent most of that demand being met. This is audible. Fiddling about with capacitors has become audible, because they try to send transient current through the long inductive connections. Sort the decoupling and this dependance on the PSU dissappears.
There may well be audible differences between different types of capacitors for a given power supply/amplifier setup, but it's far more likely to be due to differences in impedances and resonances which may be specific to that circuit rather than the magical properties oft claimed for some brands of capacitors.
Stick to solid industrial types, use lots of them, keep wiring short and thick and have good local decoupling as Andrew T suggests and you won't go far wrong.
Stick to solid industrial types, use lots of them, keep wiring short and thick and have good local decoupling as Andrew T suggests and you won't go far wrong.
Last edited:
visit as many as online store that you can find such pcx, hificoll,mpercy,farnell,mouser,digikey then find the one that fit your wallet
If you need caps lasting a lifetime, the kemet peh169 serie is great as reservoir caps due to a really robust construction and great ripple current ratings. But they're not very convenient (big, restrictive wrt positioning). Did I say pricey too ?
I tend to use panasonic fc/fr for most anything today (except when low esr can be a problem). They're compact and specs are good. Price is not outrageous.
5 paralleled 2200uF/50V pana fc will actually beat most 10.000uF/50V big caps for ripple current (12.4A for those 5 pana fc vs 9.9A for a 10.000µ/63v peh169 vs 5.6A for a 10.000uF/63V vishay 056 for example). Above 50V, big cans become mandatory if you don't want to // dozens of caps though. The vishay 056 serie is ok there or the kemet alc10.
btw: a note of caution against // smaller caps for 100hz filtering. Most datasheets for low esr caps now give the ripple current at 100khz. It's a good idea to check the derating table for 100hz operation. In my examples, the pana fc 2200µF/50V had to be derated from 3.1A @100khz to 2.48A @100hz.
I tend to use panasonic fc/fr for most anything today (except when low esr can be a problem). They're compact and specs are good. Price is not outrageous.
5 paralleled 2200uF/50V pana fc will actually beat most 10.000uF/50V big caps for ripple current (12.4A for those 5 pana fc vs 9.9A for a 10.000µ/63v peh169 vs 5.6A for a 10.000uF/63V vishay 056 for example). Above 50V, big cans become mandatory if you don't want to // dozens of caps though. The vishay 056 serie is ok there or the kemet alc10.
btw: a note of caution against // smaller caps for 100hz filtering. Most datasheets for low esr caps now give the ripple current at 100khz. It's a good idea to check the derating table for 100hz operation. In my examples, the pana fc 2200µF/50V had to be derated from 3.1A @100khz to 2.48A @100hz.
No, the -3dB point happens when the cap interacts with external resistance(s), not its own parasitics.
Certain. But also, interacting with external parameters.
I returned back to my audio hobby after 23 years, and I thought at the beginning that type of capacitors and rectifier diodes can con affect at sound a lot.
I built Erno Borbely Web 199/320 phono stage, and decided just to play with diodes/capacitors - as it was written a lot. My first setup was with the bank
of Panasonic FC 2200uf 50V - ten per rail, rectifier diodes IR HFA25PB60, 250W 2x44V transformer for each channel and 10H 0.5A 9 Ohm James Audio inductor for each rail (and, of course NTC softstart). All wiring is made with 2 mm2 silver plated OFC teflon wire.
Preamp sounds decent, but I have some big caps, and decided to check. First I replaced 10x Panasonic bank with Kendeil 10000uf x100V (bought from ebay seller umut) -
I really heard improvement - it is very hard to explain, but the sound became "transparent" and "filled". I tried to put one more Kendeil 10000 uf in parallel (for each rail, of course) - and it was disappointing - sound really became "flat", worse then with Panasonic FC bank. I have RIFA PEH200 100000uf 80V, that I purchased for my power amplifier, so I tried to use them - worse than Panasonic, better than parallel Kendeils. I also tried Kendeil 68000uf 50V - the same as RIFA, Panasonic FC is better. I still wanted to have 22000 uf as minimum, as I have 350 mA per rail power consumption (200 = preamplifier, 150 - shunt regulator), so I bought 4 Kendeil 22000uf 100V - I got the same sound as my first test with Kendeil 10000uf, I retested
really many times and asked my friends to estimate (of course blind), to be sure that it is real, - and 4 persons preferred Kendeil 22000uf variant, and one - Panasonic FC.
After all these tests I decided to check capacitors inside and after regulator - initially I used Elna Cerafine, as recommended, so after reading a lot about Black Gate, I bought some Black Gate capacitors from Michael Percy, "STD" and "FK" series, replaced all of them inside preamplifier and regulators, first "STD" - and it was really disappointing - the sound became "hard" and a bit "metallic", "FK" gave me the same result. I also checked Elna Silmic II and Panasonic FC inside regulator and as decoupling in the amplifier - I can not tell that Silmic is worse or better then Cerafine - they have very similar sound, I do not like at all FC after regulator - the sound became "flat". I returned back to Elna Cerafine.
I just want to tell about resistors (sorry, I know that we discuss capacitors) - first I used Vishay Dale, and I use to hear this sound. I read a lot about Caddock MK-132, so I purchased all set from Michael Percy,
and built another amplifier with all Caddocks - disappointing again, "metallic sound" at high frequencies - so it was second disappointment. Now I plan to purchase metal-foil and to hear how they are.
My conclusions about capacitors - try do not put in parallel capacitors after rectifier (Panasonic FC is exception), Elna Cerafine and Silmic II gave the most pleasant sound, I do not want to insult Black Gate fans,
but my opinion that I spent money for nothing (although get some experience).
I built Erno Borbely Web 199/320 phono stage, and decided just to play with diodes/capacitors - as it was written a lot. My first setup was with the bank
of Panasonic FC 2200uf 50V - ten per rail, rectifier diodes IR HFA25PB60, 250W 2x44V transformer for each channel and 10H 0.5A 9 Ohm James Audio inductor for each rail (and, of course NTC softstart). All wiring is made with 2 mm2 silver plated OFC teflon wire.
Preamp sounds decent, but I have some big caps, and decided to check. First I replaced 10x Panasonic bank with Kendeil 10000uf x100V (bought from ebay seller umut) -
I really heard improvement - it is very hard to explain, but the sound became "transparent" and "filled". I tried to put one more Kendeil 10000 uf in parallel (for each rail, of course) - and it was disappointing - sound really became "flat", worse then with Panasonic FC bank. I have RIFA PEH200 100000uf 80V, that I purchased for my power amplifier, so I tried to use them - worse than Panasonic, better than parallel Kendeils. I also tried Kendeil 68000uf 50V - the same as RIFA, Panasonic FC is better. I still wanted to have 22000 uf as minimum, as I have 350 mA per rail power consumption (200 = preamplifier, 150 - shunt regulator), so I bought 4 Kendeil 22000uf 100V - I got the same sound as my first test with Kendeil 10000uf, I retested
really many times and asked my friends to estimate (of course blind), to be sure that it is real, - and 4 persons preferred Kendeil 22000uf variant, and one - Panasonic FC.
After all these tests I decided to check capacitors inside and after regulator - initially I used Elna Cerafine, as recommended, so after reading a lot about Black Gate, I bought some Black Gate capacitors from Michael Percy, "STD" and "FK" series, replaced all of them inside preamplifier and regulators, first "STD" - and it was really disappointing - the sound became "hard" and a bit "metallic", "FK" gave me the same result. I also checked Elna Silmic II and Panasonic FC inside regulator and as decoupling in the amplifier - I can not tell that Silmic is worse or better then Cerafine - they have very similar sound, I do not like at all FC after regulator - the sound became "flat". I returned back to Elna Cerafine.
I just want to tell about resistors (sorry, I know that we discuss capacitors) - first I used Vishay Dale, and I use to hear this sound. I read a lot about Caddock MK-132, so I purchased all set from Michael Percy,
and built another amplifier with all Caddocks - disappointing again, "metallic sound" at high frequencies - so it was second disappointment. Now I plan to purchase metal-foil and to hear how they are.
My conclusions about capacitors - try do not put in parallel capacitors after rectifier (Panasonic FC is exception), Elna Cerafine and Silmic II gave the most pleasant sound, I do not want to insult Black Gate fans,
but my opinion that I spent money for nothing (although get some experience).
Bear in mind that there are many who say that Black Gates can take several hundred hours to settle, the sound quality yoyo-ing up and down in the meantime. I have experienced some of this myself.
About Black Gates - I also heard it (that they need 200-400 hours to "warm-up"), so I left
my preamp on for about 10 days with "STD" series - I can not say that the sound became better or worse, as 10 days passed, but after replacing them with Cerafine I heard real improvement (softness and transparency - again, very hard to describe). I had no power to repeat this test with "FK", but I have such feeling, that result will be similar.
my preamp on for about 10 days with "STD" series - I can not say that the sound became better or worse, as 10 days passed, but after replacing them with Cerafine I heard real improvement (softness and transparency - again, very hard to describe). I had no power to repeat this test with "FK", but I have such feeling, that result will be similar.
Alweit,
Your tests sound very interesting, and a part of it seems to agree with the findings this guy shares:
High End Audio - Electrolytic capacitors
You vaguely specify the regulator you used on the Borbely preamp, even if it seems to be a shunt type. What specific reg type did you use with the "just decent" original version?
In the next weeks I will design the pcb and build the first MM preamp Borbely designed for The Audio Amateur, where he mixed bipolars and FETs. On tha design Borbely implemented a version of Sulzer's regulator.
Building an all FET preamp presently is close to impossible, and most active parts in all-FET Borbely's designs are unobtainable nowadays. A pity, because of the hard labor EB invested on those designs.
Your tests sound very interesting, and a part of it seems to agree with the findings this guy shares:
High End Audio - Electrolytic capacitors
You vaguely specify the regulator you used on the Borbely preamp, even if it seems to be a shunt type. What specific reg type did you use with the "just decent" original version?
In the next weeks I will design the pcb and build the first MM preamp Borbely designed for The Audio Amateur, where he mixed bipolars and FETs. On tha design Borbely implemented a version of Sulzer's regulator.
Building an all FET preamp presently is close to impossible, and most active parts in all-FET Borbely's designs are unobtainable nowadays. A pity, because of the hard labor EB invested on those designs.
Last edited:
Crazy Caps
I am planning the power supply for a 200W/channel amp.
I like the idea of using one single large cap on each rail to minimise wiring and connections, and also I found that (cost no object) you can get better ESR in a single cap than even two smaller caps from same series.
So... I'm currently looking at the EPCOS Sikorel B41570 which I would mount horizontally off the heatsink in a 2U case. 5mOhm ESR and ripple rating off the scale. And lifetime is about a million years which is good because I plan to do this once.
I planned to use one Fairchild integrated full-wave bridge rectifier for each monoblock, but I realise that some snubbing (CRC) may be required between the AC side of the bridge and the DC rails (according to Hagerman (figure 7): http://www.hagtech.com/pdf/snubber.pdf)
I have often seen snubbers omitted (Power Supply for Power Amplifiers) or applied across the DC rails, but I have not seen snubbing done across the diodes and filter caps in practice. Since Hagerman seems to know what he's talking about, why is it not more common to see this snubbing arrangement?
I know there are a couple of smaller PCBs with a decent snubbing design built in (e.g. DIYAudio PCB) but I need caps >35mm diameter and +/-55 volts.
Any suggestions/ideas/comments? Many thanks. Pops.
I am planning the power supply for a 200W/channel amp.
I like the idea of using one single large cap on each rail to minimise wiring and connections, and also I found that (cost no object) you can get better ESR in a single cap than even two smaller caps from same series.
So... I'm currently looking at the EPCOS Sikorel B41570 which I would mount horizontally off the heatsink in a 2U case. 5mOhm ESR and ripple rating off the scale. And lifetime is about a million years which is good because I plan to do this once.
I planned to use one Fairchild integrated full-wave bridge rectifier for each monoblock, but I realise that some snubbing (CRC) may be required between the AC side of the bridge and the DC rails (according to Hagerman (figure 7): http://www.hagtech.com/pdf/snubber.pdf)
I have often seen snubbers omitted (Power Supply for Power Amplifiers) or applied across the DC rails, but I have not seen snubbing done across the diodes and filter caps in practice. Since Hagerman seems to know what he's talking about, why is it not more common to see this snubbing arrangement?
I know there are a couple of smaller PCBs with a decent snubbing design built in (e.g. DIYAudio PCB) but I need caps >35mm diameter and +/-55 volts.
Any suggestions/ideas/comments? Many thanks. Pops.