Last year, after building up my second XRK971 Pocket Class A amplifier board, I did some extensive output cap rolling in a quest to find the best-sounding electrolytic caps for coupling the amp to my 50 ohm headphones. I was surprised at just how different the various caps sound in the signal path.
Some of this was posted in the Pocket Class A forum and some on Head-Fi. Now I'm rounding it all up in one place with the hope that it might be useful to somebody. No, these weren't blind tests. You don't have to believe my ears. Try testing for yourself and see if you agree!
The majority of my listening tests were performed through ATH-M50 headphones, which I have found to be invaluable for giving an honest rendering of sonic differences. These findings were confirmed through further comparisons using my Sennheiser HD598. I also auditioned the caps while using the Pocket Class A as a preamplifier to feed two different Class D-powered nearfield speaker systems. Sources were a wide variety of FLAC files played both from my XDuoo X3 DAP and my ECHO Audiofire soundcard.
This test was intended to compare caps of similar physical size that would fit into the available space on my PCB for this particular build. This turned out to be 10x20mm, though a few (marked with asterisks) were 12.5x20mm, and then the OSCONS are smaller than all the rest. I believe that controlling for size vs. specs likely reduces the accuracy of some of my comparisons somewhat, since voltage and capacitance ratings varied between caps tested.
Voltage rating varied from 16V to 25V. Theoretically a cap rated for higher voltage exhibits less distortion, per Cyril Bateman. I tested some types at multiple ratings and found that in reality, this effect was difficult to hear. Different capacitor brands and series had much more audible effect than different voltage ratings.
Capacitance varied from 220uF to 1000uF. Of course, this affects the low corner frequency. All of these caps provided satisfying bass quantity...in truth even a -3dB rolloff at 15Hz is only noticeable with critical listening comparisons. This chart shows predicted bass response of each cap value with my headphones...both pairs are 50 Ohms, and in parallel with the amp's 1K load resistor, the total load is 47.6R.
Here are my listening impressions:
Nichicon KW
value tested: 25V 1000uF
Reasonably pleasant, but the treble is quite dulled. Nicely engaging, slightly grainy mids. Bass extends low but is a little uncontrolled and surprisingly quiet for a 1000uF cap. Subpar detail and dynamics.
Nicicon FW
value tested: 25V 1000uF
Brighter than KW. More balanced mids sit on par with the rest of the spectrum. Clearer treble, with some sparkle, but still lacking air up top--stuffy. Good presence in mids, with just the right amount of bite. Bass quantity is better than KW but a bit distorted.
Nichicon Fine Gold FG
values tested:16V 470uF and 25V 330uF
A small step closer to KZ sound in resolution and overall engagement. Smoother than KW or FW, with a balanced frequency response. Deep, punchy and controlled bass. Still a bit "closed in" due to being a bit short on high treble. These can sound strained in louder passages. The 25V may be a hair smoother, with a little more treble, but not enough to avoid a similarly small, stuffy soundstage. Where is the high treble, Nichicon? In the KZ only, I guess. Sheesh.
Nichicon KA
values tested: 25V 470uF and 25V 1000uF
The KA actually stands out from the typical Nichicon "house sound" IMO. It has a slightly mid-centric signature with enough treble to sparkle some, though the treble is a tad hot. Midrange and vocals come forward in the mix nicely, but without grain--they are the smoother than even the FG. Bass has decent weight and control. Somehow I find the KA more natural in timbre than all the other Nichicons. The sound is clearer, though not particularly airy or exciting.
Panasonic FR
value tested: 25V 1000uF and 25V 1500uF*
FAR more neutral and revealing than all of these Nichicons. Maybe its the ultra low ESR…it just gets out of the way and lets the music come through. It gives the cleanest and most spacious presentation, with excellent detail and real air on top for a beautifully open soundstage. The bass is awesomely powerful with deeper extension than any of the Nichicons, and tight control. Blackest background. Fastest transients. Need I say more? If I had to nitpick, I'd wish for a touch more warmth and presence in the mids, but I suppose that might just be a coloration that suits my taste. These caps simply sound REALISTIC.
Panasonic OSCON
values tested: 25V 390uF and 16V 1000uF
I found these to be extremely well balanced thorough the frequency range, with excellent bass weight and particularly great bass control. The bass is just effortless. Overall these have a full, warm and creamy sound signature. But it has a soft/dull one-dimensionality to my ears, and fails to engage me. These fall short on detail compared to all other caps I have tried.
Elna Cerafine
values tested: 16V 330uF, 25V 220uF, 25V 470uF*
I can see why a few people like these more than Silmics. They have more midrange forwardness and bite to them, and at first blush seem to offer a more engaging sound. However, their treble can be a little harsh after a while, and the response is rather poorly extended on both extremes of the frequency spectrum, resulting in a less detailed, somewhat boomy and congested sound, with noticeably less bass control than the Silmic. Not my cup of tea.
Elna Silmic II
value tested: 16V 220uF
Bass has flawless texture, speed, and extension, even with a marginally undersized 220uF cap. A polite, soft midrange and treble makes you want to crank the volume, and the overall sound is nice and full. It hides the flaws of harsh recordings, but also seems to push vocals farther away and hide some harmonic texture, resulting in an odd hollowness in midrange timbre. I'd say these sound spacious due to decent extension and a somewhat exaggerated treble sparkle, but the high treble is missing. I particularly dislike the way these do metal percussion. Cymbals and hi-hats do not sound realistically metallic. Other than that, these sound great, if slightly veiled in parts of the spectrum.
Nichicon Muse KZ
value tested: 25V 330uF*
Nice. These are quite detailed, with a beautiful, jangly midrange and sparkly treble that gives an incisive "electric" sound to the upper ranges. Vocals really jump out and grab my attention so that song lyrics have excellent intelligibility. The overall presentation is ever so slightly on the hot/forward side of neutral, with a hint of upper mid harshness, but the overall timbre is natural, with enough treble to give an open soundstage. Metal strings and percussion sound right. This size gives impactful, fast, and detailed bass with my 50R cans. With these, I never feel like I'm missing any detail whatsoever, and I wouldn't say the soundstage is huge, but certainly bigger than either Elna. They're lightning fast, and darn revealing.
------------------
Conclusion
After all this, I now believe that NO electrolytic cap on the face of the planet is totally transparent in the signal path. My favorite in this application turned out to be the Panasonic FR, and I believe these to be the most transparent of the bunch, although I can absolutely understand how others might prefer the other "flavors"--particularly the KZ and Silmic, which each have their strong points and are justifiably loved by many.
Conspicuously missing from this lineup is the Nichicon Muse Bipolar BP and the Panasonic FC. I have heard good reports about both and hope to test them eventually.
Some of this was posted in the Pocket Class A forum and some on Head-Fi. Now I'm rounding it all up in one place with the hope that it might be useful to somebody. No, these weren't blind tests. You don't have to believe my ears. Try testing for yourself and see if you agree!
The majority of my listening tests were performed through ATH-M50 headphones, which I have found to be invaluable for giving an honest rendering of sonic differences. These findings were confirmed through further comparisons using my Sennheiser HD598. I also auditioned the caps while using the Pocket Class A as a preamplifier to feed two different Class D-powered nearfield speaker systems. Sources were a wide variety of FLAC files played both from my XDuoo X3 DAP and my ECHO Audiofire soundcard.
This test was intended to compare caps of similar physical size that would fit into the available space on my PCB for this particular build. This turned out to be 10x20mm, though a few (marked with asterisks) were 12.5x20mm, and then the OSCONS are smaller than all the rest. I believe that controlling for size vs. specs likely reduces the accuracy of some of my comparisons somewhat, since voltage and capacitance ratings varied between caps tested.
Voltage rating varied from 16V to 25V. Theoretically a cap rated for higher voltage exhibits less distortion, per Cyril Bateman. I tested some types at multiple ratings and found that in reality, this effect was difficult to hear. Different capacitor brands and series had much more audible effect than different voltage ratings.
Capacitance varied from 220uF to 1000uF. Of course, this affects the low corner frequency. All of these caps provided satisfying bass quantity...in truth even a -3dB rolloff at 15Hz is only noticeable with critical listening comparisons. This chart shows predicted bass response of each cap value with my headphones...both pairs are 50 Ohms, and in parallel with the amp's 1K load resistor, the total load is 47.6R.
Here are my listening impressions:
Nichicon KW
value tested: 25V 1000uF
Reasonably pleasant, but the treble is quite dulled. Nicely engaging, slightly grainy mids. Bass extends low but is a little uncontrolled and surprisingly quiet for a 1000uF cap. Subpar detail and dynamics.
Nicicon FW
value tested: 25V 1000uF
Brighter than KW. More balanced mids sit on par with the rest of the spectrum. Clearer treble, with some sparkle, but still lacking air up top--stuffy. Good presence in mids, with just the right amount of bite. Bass quantity is better than KW but a bit distorted.
Nichicon Fine Gold FG
values tested:16V 470uF and 25V 330uF
A small step closer to KZ sound in resolution and overall engagement. Smoother than KW or FW, with a balanced frequency response. Deep, punchy and controlled bass. Still a bit "closed in" due to being a bit short on high treble. These can sound strained in louder passages. The 25V may be a hair smoother, with a little more treble, but not enough to avoid a similarly small, stuffy soundstage. Where is the high treble, Nichicon? In the KZ only, I guess. Sheesh.
Nichicon KA
values tested: 25V 470uF and 25V 1000uF
The KA actually stands out from the typical Nichicon "house sound" IMO. It has a slightly mid-centric signature with enough treble to sparkle some, though the treble is a tad hot. Midrange and vocals come forward in the mix nicely, but without grain--they are the smoother than even the FG. Bass has decent weight and control. Somehow I find the KA more natural in timbre than all the other Nichicons. The sound is clearer, though not particularly airy or exciting.
Panasonic FR
value tested: 25V 1000uF and 25V 1500uF*
FAR more neutral and revealing than all of these Nichicons. Maybe its the ultra low ESR…it just gets out of the way and lets the music come through. It gives the cleanest and most spacious presentation, with excellent detail and real air on top for a beautifully open soundstage. The bass is awesomely powerful with deeper extension than any of the Nichicons, and tight control. Blackest background. Fastest transients. Need I say more? If I had to nitpick, I'd wish for a touch more warmth and presence in the mids, but I suppose that might just be a coloration that suits my taste. These caps simply sound REALISTIC.
Panasonic OSCON
values tested: 25V 390uF and 16V 1000uF
I found these to be extremely well balanced thorough the frequency range, with excellent bass weight and particularly great bass control. The bass is just effortless. Overall these have a full, warm and creamy sound signature. But it has a soft/dull one-dimensionality to my ears, and fails to engage me. These fall short on detail compared to all other caps I have tried.
Elna Cerafine
values tested: 16V 330uF, 25V 220uF, 25V 470uF*
I can see why a few people like these more than Silmics. They have more midrange forwardness and bite to them, and at first blush seem to offer a more engaging sound. However, their treble can be a little harsh after a while, and the response is rather poorly extended on both extremes of the frequency spectrum, resulting in a less detailed, somewhat boomy and congested sound, with noticeably less bass control than the Silmic. Not my cup of tea.
Elna Silmic II
value tested: 16V 220uF
Bass has flawless texture, speed, and extension, even with a marginally undersized 220uF cap. A polite, soft midrange and treble makes you want to crank the volume, and the overall sound is nice and full. It hides the flaws of harsh recordings, but also seems to push vocals farther away and hide some harmonic texture, resulting in an odd hollowness in midrange timbre. I'd say these sound spacious due to decent extension and a somewhat exaggerated treble sparkle, but the high treble is missing. I particularly dislike the way these do metal percussion. Cymbals and hi-hats do not sound realistically metallic. Other than that, these sound great, if slightly veiled in parts of the spectrum.
Nichicon Muse KZ
value tested: 25V 330uF*
Nice. These are quite detailed, with a beautiful, jangly midrange and sparkly treble that gives an incisive "electric" sound to the upper ranges. Vocals really jump out and grab my attention so that song lyrics have excellent intelligibility. The overall presentation is ever so slightly on the hot/forward side of neutral, with a hint of upper mid harshness, but the overall timbre is natural, with enough treble to give an open soundstage. Metal strings and percussion sound right. This size gives impactful, fast, and detailed bass with my 50R cans. With these, I never feel like I'm missing any detail whatsoever, and I wouldn't say the soundstage is huge, but certainly bigger than either Elna. They're lightning fast, and darn revealing.
------------------
Conclusion
After all this, I now believe that NO electrolytic cap on the face of the planet is totally transparent in the signal path. My favorite in this application turned out to be the Panasonic FR, and I believe these to be the most transparent of the bunch, although I can absolutely understand how others might prefer the other "flavors"--particularly the KZ and Silmic, which each have their strong points and are justifiably loved by many.
Conspicuously missing from this lineup is the Nichicon Muse Bipolar BP and the Panasonic FC. I have heard good reports about both and hope to test them eventually.
Last edited by a moderator:
The capacitor should be at least 1600uF for 50R headphones. Otherwise the comparisons are unfair
because measurable distortion is generated in smaller values.
Last edited:
Are you referring to phase shift above the corner frequency?
It is my understanding that increasing voltage rating reduces distortion. In the Capacitor Sounds series by Cyril Bateman, this effect is shown on FFTs. I have never heard that increasing the capacitance has an effect on distortion. By your logic every tweeter with an RC highpass filter would suffer terrible distortion in the passband...
It is my understanding that increasing voltage rating reduces distortion. In the Capacitor Sounds series by Cyril Bateman, this effect is shown on FFTs. I have never heard that increasing the capacitance has an effect on distortion. By your logic every tweeter with an RC highpass filter would suffer terrible distortion in the passband...
The most basic point about using electrolytic capacitors for coupling is to minimize the voltage drop
across them, to minimize the signal generated distortion, by using a large enough value of uF.
Yes, a tweeter capacitor (with signal voltage across it) does generate distortion, if the capacitor
is electrolytic. This is all well known, and is why many prefer film capacitors for speaker crossovers.
Last edited:
Rayma,
You may well know something I don't. FWIW I did test a 25V 1500uF Panasonic FR in my amp. It sounded exactly the same to me as the 1000uF version. I believe that the major sonic differences that I heard between caps are due to different materials and construction of each capacitor brand and series.
Would every electrolytic cap type sound better if it was sized so the -3dB point was set at 2Hz instead of 3 or 4Hz? Theoretically yes, but based on my experience I'm not sure that I'd be able to hear the difference. Regardless, I had to choose something that would fit in an Altoids tin
Certainly other factors can enter as well, but the uF value should be enough, and the same
for all parts for comparisons (voltage rating too). The exact pole frequency is not as important
as setting it low enough, and usually a decade below 20Hz is chosen. The music chosen can be
more or less revealing, depending on its spectral content.
Last edited:
I agree that using different capacitance values is a weakness of my test. Yet I noticed that when I tested multiple values of a certain type, they all had the same overall "flavor."
Regarding burn-in, I rolled caps in this amp for a long time, because I already had my first build of the same amp up and running (it has Nichicon KA in it). I gave up on most of the caps in under 6 hours, but I think the Silmics and Muse KZ must have been played for probably at last 50 hours by the time I made my final choice and soldered in the Panasonics.
A blind test would be really fun. However I'm not sure if I could persuade any of my friends to actually do this.
Hi Sumotan,
I have always wanted to hear a Black Gate, but man they are insanely pricey!
I am very familiar with the distortion effects of different voltage ratings. As for different width to height ratios having different sound characterics, that's new to me, but I'm interested. Have you noticed a trend? Do the stubby ones sound "fatter?"
Last edited:
Thank you for posting the results of your trials. I'm pretty much in agreement with your conclusions, although we differ on FR vs KZ. I am sure there will be plenty of critics coming along no nitpick your testing protocol. Ignore them.
Are you aware of the Audio Note Kaisei series? I haven't tried them yet, but I intend to compare them to BG std in output coupling duty. From what I've read, which is limited because they're fairly new, they possess many of the qualities of BG (which was the developer's goal), and even surpass the BG's in a few areas.
Peace,
Tom E
Are you aware of the Audio Note Kaisei series? I haven't tried them yet, but I intend to compare them to BG std in output coupling duty. From what I've read, which is limited because they're fairly new, they possess many of the qualities of BG (which was the developer's goal), and even surpass the BG's in a few areas.
Peace,
Tom E
As I made progress in my listening & tweaking experience, I’ve discovered that BGs are really not that transparent nor good at hi frequencies but what they bring is a special magic of bringing instruments, voices etc to life.
As Asuslover stated, a more accurate test is to use caps of the same values & voltage, yes sonic signatures of each cap even if different values can be heard but are you hearing all there is? Example KZ regardless of values 50v versions sounds better than 25v.
Cheers
As Asuslover stated, a more accurate test is to use caps of the same values & voltage, yes sonic signatures of each cap even if different values can be heard but are you hearing all there is? Example KZ regardless of values 50v versions sounds better than 25v.
Cheers
This thread is a good example of two things:
1. if you use a component wrongly (e.g. by using the wrong value) then you may be able to hear it
2. if you do an experiment without properly understanding what you are doing then all you get is raw data; to get measurements you have to apply correct theory too
For 50 ohm headphones the minimum electrolytic value is about 1500uF, so you get a corner frequency no higher than 2Hz. This is not because of phase shift, but because of signal voltage across the capacitor. Any test done with a smaller value merely tells you that the value is too small, so all you are doing is comparing different distortions to see which one you prefer. Given the correct value, you are likely to find that they all sound very similar to each other.
1. if you use a component wrongly (e.g. by using the wrong value) then you may be able to hear it
2. if you do an experiment without properly understanding what you are doing then all you get is raw data; to get measurements you have to apply correct theory too
For 50 ohm headphones the minimum electrolytic value is about 1500uF, so you get a corner frequency no higher than 2Hz. This is not because of phase shift, but because of signal voltage across the capacitor. Any test done with a smaller value merely tells you that the value is too small, so all you are doing is comparing different distortions to see which one you prefer. Given the correct value, you are likely to find that they all sound very similar to each other.
DC bias was measured at 7.5V across the cap leads. That's 30% of rating for most of the caps tested (the 25V rated ones) and up to 47% for the rest.
Perhaps my cap experience is not nearly as extensive as yours or the topic starter's but for years i have been liking the FG in preference to many others. Not really in coupling positions, but as cathode resistor bypasses.
Getting the first Kaiseis made me realise how coloured the FGs are. Pleasantly coloured indeed, but certainly not uniform within the audible range and slightly dynamically subdued. By now i have replaced all cathode bypasses with Kaisei and cannot be happier. Interestingly, never warmed up to the sound of the standard BGs, to which the Kaisei are sometimes compared.
