I currently use the Epcos Sikorels for full range, but for midrage upwards, would use the BHC T Network Caps (cleaner mids)
Which value of Sikorels are you using and can you elaborate on the sound differences you notice? My current ZEN V4 has a 10mF/63 Volt BC components 051 and the amp sounds pretty well balanced to my ears. I am planning to build an F3 and my options for output caps were either some combination of Sikorel cans (I have 2.2 & 10mF 100V and 4.7mF 63V) or alternatively a whole bank of either ELNA Silmics or alternatively small non-polar caps (Nichicon, or Silmics if I get them non-polar). Maybe I will even be able to compare the options in a prototype when I get round to it.
Last edited:
I'm use a version of John B's (E C Design) Charge Transfer supply cct to elimenate the diode (and line) noise altogether - makes the choice of power supply caps different again!
Could you post or pm me the link about this PSU? I remember it but cant find a bookmark.
I'm not able to find elna silk caps , just ELNA . Are all "ELNA" caps "ELNA silk" ?
If not where are you guys finding them , i don't see Silk caps on digi-key's website ..
If not where are you guys finding them , i don't see Silk caps on digi-key's website ..
Digi Key
On Digi-Key, their website does not specific Silmic II, but do a search for Elna, RFS series. The RFS series is the Silmic II.
Handmade Electronics also carries them at an affordable price.
On Digi-Key, their website does not specific Silmic II, but do a search for Elna, RFS series. The RFS series is the Silmic II.
Handmade Electronics also carries them at an affordable price.
Hi Telstar,
The thread is the "Building the Ultimate NOC Dac using the 1541A ..." and the first ccts appeared at post #2304 (page 58, i think) and #2347 (59) - there were a few later variations with transistors, not fets, but have buried them somewhere, will dig if you need them.
I only used the basic ccts for this as it's a much delayed project - MaxLorenz (Mercurio), and others, have done quite a bit more - works like a charm, especially with Torroids - much quieter background, music stands out from background - better small details, etc.
Martin,
I was afraid you might ask this - my listening requirements may/maynot be anywhere the same, so this is MY impressions only - and I use a technique of "tuning the transformers" (Prometheus Audio website) that makes all the supply variations much more audible.
To me the Siemens/Epcos Sikorels are the "best" for my F3 amp power supply and also for the O/P cap - This is the 22,000uF, B41550 ones - the older B41455 are not as good but with some of the 680uF B41590, B41684 bipasses (the only ones I can find) - not the cheapest caps and take forever to "break in".
so, for your proposed F3, suggest use either the 10,000uF but possibly better with the lower voltage 2 x 4,700uF - there's only 20 odd volts on the o/p - you couls also try the highly regarded system of adding a pair of the 2,200s nose to tail as a further bipass - play about with some of those POI caps as bipasses - amazing flexibility, stunning performance provided the xover components are GOOD quality.
I just found some NOS B41550 22,000uF/100v in Germany (eBay, would you believe!) and I think they still have some 33,ooouF/63v B41560 (about 1/2 way between the B41550s and the B41455s) - about 20 Euro, I think.
From memory (bit unrelible!) the BC 051 are the Phillips ones, yes? I found the bass to be a bit "fat" and a bit slow ( just not my taste) but pretty smooth, particularly on the Zen (jfet I/P?), The Roe have a similar sound to me, and the Mundorfs (F&TS, etc) are similar but more detailed and a bit quicker. You can see where my "hearing" fits.
Later, after the Xover comes together (looking at the B1 thread), I'll use the F3 to run the "full rangers" from >160 upwards and I'll use the T-Network caps (from DNM) as the mids are the best midrange of them all, but a bit "thin" low down (IMO, naturally!).
F5 for bass, no hesitation!
Hope this is useful ...
The thread is the "Building the Ultimate NOC Dac using the 1541A ..." and the first ccts appeared at post #2304 (page 58, i think) and #2347 (59) - there were a few later variations with transistors, not fets, but have buried them somewhere, will dig if you need them.
I only used the basic ccts for this as it's a much delayed project - MaxLorenz (Mercurio), and others, have done quite a bit more - works like a charm, especially with Torroids - much quieter background, music stands out from background - better small details, etc.
Martin,
I was afraid you might ask this - my listening requirements may/maynot be anywhere the same, so this is MY impressions only - and I use a technique of "tuning the transformers" (Prometheus Audio website) that makes all the supply variations much more audible.
To me the Siemens/Epcos Sikorels are the "best" for my F3 amp power supply and also for the O/P cap - This is the 22,000uF, B41550 ones - the older B41455 are not as good but with some of the 680uF B41590, B41684 bipasses (the only ones I can find) - not the cheapest caps and take forever to "break in".
so, for your proposed F3, suggest use either the 10,000uF but possibly better with the lower voltage 2 x 4,700uF - there's only 20 odd volts on the o/p - you couls also try the highly regarded system of adding a pair of the 2,200s nose to tail as a further bipass - play about with some of those POI caps as bipasses - amazing flexibility, stunning performance provided the xover components are GOOD quality.
I just found some NOS B41550 22,000uF/100v in Germany (eBay, would you believe!) and I think they still have some 33,ooouF/63v B41560 (about 1/2 way between the B41550s and the B41455s) - about 20 Euro, I think.
From memory (bit unrelible!) the BC 051 are the Phillips ones, yes? I found the bass to be a bit "fat" and a bit slow ( just not my taste) but pretty smooth, particularly on the Zen (jfet I/P?), The Roe have a similar sound to me, and the Mundorfs (F&TS, etc) are similar but more detailed and a bit quicker. You can see where my "hearing" fits.
Later, after the Xover comes together (looking at the B1 thread), I'll use the F3 to run the "full rangers" from >160 upwards and I'll use the T-Network caps (from DNM) as the mids are the best midrange of them all, but a bit "thin" low down (IMO, naturally!).
F5 for bass, no hesitation!
Hope this is useful ...
I agre that the Sikorels are fabuluos caps, and so are the Rifa PEH 169.
Both of them would be my first choises in most cases.
You have to consider though, what your amplifier is sensitive to. Some amps need smooth power with as low ripple as possible, that makes both the Sikorels and the PEH 169 the obvious choise.
Some amps though are not that sensitive to ripple, but might instead need low impedance. This calls for a different set of caps.
Low ripple means large capacitance, which normally works well with as low impedance as possible. No one delivers that in a "one can" solution as Epcos and Rifa.
Lately I´ve experimented a bit with a class D amp, and the PSU for it.
Both Sikorel and Rifa was tried out, but this amp is not at all sensitive to ripple, instead very low impedance was important. Therefor we tried Pansonic FC caps, and with only 18*660 µF instead of 4 * 10.000µF Epcos Sikorel, the amplifier really came to life.
Impedance was lowered a lot using the Panasonics, and to this amplifier this seems to be the most important thing.
So IMHO the amplifier design does in some way influence the best choise of smoothing caps.
Both of them would be my first choises in most cases.
You have to consider though, what your amplifier is sensitive to. Some amps need smooth power with as low ripple as possible, that makes both the Sikorels and the PEH 169 the obvious choise.
Some amps though are not that sensitive to ripple, but might instead need low impedance. This calls for a different set of caps.
Low ripple means large capacitance, which normally works well with as low impedance as possible. No one delivers that in a "one can" solution as Epcos and Rifa.
Lately I´ve experimented a bit with a class D amp, and the PSU for it.
Both Sikorel and Rifa was tried out, but this amp is not at all sensitive to ripple, instead very low impedance was important. Therefor we tried Pansonic FC caps, and with only 18*660 µF instead of 4 * 10.000µF Epcos Sikorel, the amplifier really came to life.
Impedance was lowered a lot using the Panasonics, and to this amplifier this seems to be the most important thing.
So IMHO the amplifier design does in some way influence the best choise of smoothing caps.
yeah, your right Kurt- different amps designs do often require different supply systems - another strange one is the Cmultiplier - some caps function very differently when used here for the same amp. Interesting, eh!
Well, looking at the specs, the Rifa peh-169 and the Epcos Sikorel 41550 are the best electrolytics of big capacity (4700uf and up).
I have the suspect that class A amps prefer bigger caps than a battery of small ones. Or a middle way (like using 6800uF caps). Prices are so variated that often one has to follow the better deal, rather than the rideal, unless cost no object.
I have the suspect that class A amps prefer bigger caps than a battery of small ones. Or a middle way (like using 6800uF caps). Prices are so variated that often one has to follow the better deal, rather than the rideal, unless cost no object.
Hello JH, I understand that listening impressions are system dependent and subjective, just was interested in your findings, it was not my intention to put you on the spot.
Generally, if you use a big fat cap like 22mF it may be worth trying to put the smallest cap from the same brand in parallel, like a 2.2mF; in the case of Sikorels/Rifa models mentioned the smaller caps have a resonant frequency 5 or more times higher than the big cap, so this might help bandwidth.
Generally, if you use a big fat cap like 22mF it may be worth trying to put the smallest cap from the same brand in parallel, like a 2.2mF; in the case of Sikorels/Rifa models mentioned the smaller caps have a resonant frequency 5 or more times higher than the big cap, so this might help bandwidth.
Nah Martin, no problems - just wanted to emphasize that we each have different "sound preferences" and that mine mayn't suit yours, is all.
I'm quite determined to get all the resolution, detail, etc that is available with the better oversampling systems from my prehistoric 1541A NOS system without any of the filter degradation (IMO) and going to a bit of trouble to get it at realistic playback levels - having a rather wide musical taste makes it a bit more difficult, but getting there.
Went a bit overboard on the cap bipass "thing" on the F3 amp - rather curious how far I could tailor the sound, improve the transient performance and also increase the control of the dynamics a bit, especially on complicated music.
For those not up on this No 3 special project of Nelson (First Watt amp F3), it's a single ended current gain classA amp running 42 volt rail @1.6Amps. It's also a bit of a good lesson in the use of electrolytics as it "puts out" current (about 1.5 amps), not volts and different caps at various cct points have quite dramatic effects.
My present power supply for the F3 is a R-C-R-C (0.1R and 33,000uF caps) fed via the 8ETH06 diodes ( Slightly cleaner than the BYW29) and each cap has a Sikorel bipass (680uF/63v) and the the first cap (ripple reduction) has a 'Rifa peg124 220uf+ 1.5R' snubber across it and the 2nd cap has a 330uF peg124 + 1.8R snubber across it
This then feeds a Cmultiplier which has Nichicon caps, not Sikorels, and again a Rifa CR snubber bipass - works quite well - push the resonances up, then kill 'em off!
The output cap has 21v volts on it, so the o/p cap is a dc blocking caps of 22,000uF/40v Sikorel with a 2 x 68uF/40 S&M B41590 (Siemens & Matsushita = Epcos) plus a Rifa peg 220uF/25v + 1.6R snubber across that - then a 10uF K75 rushian POI across the whole lot of them!
Bit of a mess, but marvellous sound - can easily add low mids, reduce upper mids, add detail to mid bass, etc, etc.
Curious thing - when I added the T-Network 10,000uF to the 22,000uF Sikorels, I couldn't get the 2 of them to settle down together, and no combination of either simple or complex bipasses/snubbers would fix it - some caps are just not meant to get along together - never did get to work out the problem!
This bipassing "thing" isn't recommended for people short of patience, but does teach you very quickly "how to listen" to specific things - good headphones are a short cut particularly with good quality piano music as a source.
Bye.
I'm quite determined to get all the resolution, detail, etc that is available with the better oversampling systems from my prehistoric 1541A NOS system without any of the filter degradation (IMO) and going to a bit of trouble to get it at realistic playback levels - having a rather wide musical taste makes it a bit more difficult, but getting there.
Went a bit overboard on the cap bipass "thing" on the F3 amp - rather curious how far I could tailor the sound, improve the transient performance and also increase the control of the dynamics a bit, especially on complicated music.
For those not up on this No 3 special project of Nelson (First Watt amp F3), it's a single ended current gain classA amp running 42 volt rail @1.6Amps. It's also a bit of a good lesson in the use of electrolytics as it "puts out" current (about 1.5 amps), not volts and different caps at various cct points have quite dramatic effects.
My present power supply for the F3 is a R-C-R-C (0.1R and 33,000uF caps) fed via the 8ETH06 diodes ( Slightly cleaner than the BYW29) and each cap has a Sikorel bipass (680uF/63v) and the the first cap (ripple reduction) has a 'Rifa peg124 220uf+ 1.5R' snubber across it and the 2nd cap has a 330uF peg124 + 1.8R snubber across it
This then feeds a Cmultiplier which has Nichicon caps, not Sikorels, and again a Rifa CR snubber bipass - works quite well - push the resonances up, then kill 'em off!
The output cap has 21v volts on it, so the o/p cap is a dc blocking caps of 22,000uF/40v Sikorel with a 2 x 68uF/40 S&M B41590 (Siemens & Matsushita = Epcos) plus a Rifa peg 220uF/25v + 1.6R snubber across that - then a 10uF K75 rushian POI across the whole lot of them!
Bit of a mess, but marvellous sound - can easily add low mids, reduce upper mids, add detail to mid bass, etc, etc.
Curious thing - when I added the T-Network 10,000uF to the 22,000uF Sikorels, I couldn't get the 2 of them to settle down together, and no combination of either simple or complex bipasses/snubbers would fix it - some caps are just not meant to get along together - never did get to work out the problem!
This bipassing "thing" isn't recommended for people short of patience, but does teach you very quickly "how to listen" to specific things - good headphones are a short cut particularly with good quality piano music as a source.
Bye.
You can make it even faster by putting some caps close (1-2 cm) to the output transistors 😉
Have fun, Hannes
Have fun, Hannes
from post #159
I would agree that "industrial grade" caps are often as good as the audiophile components. One exception I would make is the use of magnetic (steel) leads in industrial caps, and some audiophile manufacturers (ELNA and Nichicon) claim that steel leads can produce audible and measurable distortion. This is not surprising since the high magnetic permeability of steel means that the skin depth is only a tiny fraction of that of copper so that at higher frequencies, even at 10s of KHz, most current should only be conduted by the tin on the leads.
There is a great similarity to Kenwood's claim regarded the "KA-1000" integr. amp. Less metal and more plastic components by their enclosures was justified cause no magnetic permeability by plastic.
Unfortunately, I could still not comprehend, where these so-called magnetic distortions were to identify from in most cases exist bound of other imperfections. So I don't know about audible magnitude of this kind of distortion.
Whether the distortion differences in the same magnitude than the differences between e. g. air coils and coil with ferrite core (which creates clearly audible distortions in good hifi chains), I do serious doubt.
Attachments
Last edited:
Hello ROCKMANDRAKE,
Did you find to spare operation? It one capacitor fail, it would be difficult to find it 😉gorgeously fast on power supply...Rubycon CZ...Rock
gorgeously fast on power supply ...
Rubycon MCZ
Rock
This PCB is like a Porsche with bicycle tires. Very high PCB inductance and moderate resistance isolates the capacitors from the output stage.
You may be able to see the capacitor banks and feel that they are "fast", but believe me, the output transistors can't see the capacitors at all, electrically speaking.
btw: Anyway, perceived bass and kick "speed" is mainly related to speaker (crossover) design and room acoustics. I think that only the people that fears getting deep into these complex subjects tries to find a remedy with placebo amplifiers.
Last edited:
I have a comment on bypassing routines!
If you bypass an electrolytic cap, by which I mean parallelling an electrolytic cap with a smaller film cap, you create a complex electrical circuit with unknown specs.
I´ve tried this many times, lately in my class D amp, and everytime i like the bypasses at most, is when I put them away.
At first you´ll often think WOW, so clean, so relaxed, and then you´ll try to refine it even further, finally after a lot of experiments, you think you´ve made it. Then try to dismantle the whole ting.
To me this always has been the ultimate test, and until this day, no bypass caps has survived it.
Instead I´d go for decoupling in well thought out places. Decoupling near the component supplied can often do wonders, but be aware, that the type and value can be of great importance.
IMHO PPS caps are by far the best choise for that task.
If you bypass an electrolytic cap, by which I mean parallelling an electrolytic cap with a smaller film cap, you create a complex electrical circuit with unknown specs.
I´ve tried this many times, lately in my class D amp, and everytime i like the bypasses at most, is when I put them away.
At first you´ll often think WOW, so clean, so relaxed, and then you´ll try to refine it even further, finally after a lot of experiments, you think you´ve made it. Then try to dismantle the whole ting.
To me this always has been the ultimate test, and until this day, no bypass caps has survived it.
Instead I´d go for decoupling in well thought out places. Decoupling near the component supplied can often do wonders, but be aware, that the type and value can be of great importance.
IMHO PPS caps are by far the best choise for that task.
Hi Eva,
You can add to the list:
• speaker (crossover) design,
• room acoustics,
• managing vibration (box enclosure of audio equipment),
• power supply,
• ground management,
• electric energy,
• ...
We probably don't realize how this is true...and difficult to reach......perceived bass and kick "speed" is mainly related to speaker (crossover) design and room acoustics. I think that only the people that fears getting deep into these complex subjects tries to find a remedy with placebo amplifiers.
You can add to the list:
• speaker (crossover) design,
• room acoustics,
• managing vibration (box enclosure of audio equipment),
• power supply,
• ground management,
• electric energy,
• ...
Last edited:
gorgeously fast on power supply ...
You can make the cap bank much faster by using 2 planes for plus and minus - this creates a low inductance bus connection to the caps, see Rifa's "Electrolytic Capacitors Application Guide" page 5 http://www.evoxrifa.com/electrolytic_cat/electrolytic_appguide.pdf
There is a great similarity to Kenwood's claim regarded the "KA-1000" integr. amp. Less metal and more plastic components by their enclosures was justified cause no magnetic permeability by plastic.
I think there are two points to consider:
1. Being magnetic the skin depth of Steel is very low, about 1/38th that of copper (from a WEB article, but you should be able to calculate this for yourself if you mistrust the figures). At DC the conductivity of steel is ok, ca. 1/7th that of copper. But with rising frequency it will quickly stop conducting at all, so the impedance is not linear. In coupling caps I would consider this as no-no, as a power supply cap it "may" actually be ok. Actually I just got some non-polar caps that unfortunately turned out to have magnetic leads, I might do some listening tests to see how they compare to other caps (caveat is that I do not have the same caps with copper leads).
2. The measured distortion was referenced in a Nichicon article but I couldn't find it on the WEB. I still have a 2 page copy from an article but without a reference to point you to. One ould have to measure similar caps with and without steel leads to get a better picture, but then why take a risk if there is a potential of added distortion.
I am not sure if the same mechanism would actually apply to enclosures but at least one point should be critical, and that is eddy currents of signal carrying wires which may just be worse in steel enclosures. I would keep all signal leads away from metal surfaces anyway as I do believe this could otherwise cause audible distortion (you might actually be able to measure this ?). Note also that a couple of manufacturers, including Brinkmann, use metal enclosures except for the top plate, claiming that this makes a noticeable difference, some others do not use any metal at all (see DNM Design WEB pages for their principles). This would be similar to what Kenwood were doing.
Last edited:
- Home
- Design & Build
- Parts
- Best electrolytic capacitors