Not sure about that Karl. I tested BG N, Nx & Nichicon ES results are positive. Very pronounce improvement when positive leg of cap is soldered to ground.
Oh I forgot I also tested out a Solen 15uf cap, flipping the cap over I didn’t hear that much difference so this means it only applies to Bipolar Electrolytic caps.
Oh I forgot I also tested out a Solen 15uf cap, flipping the cap over I didn’t hear that much difference so this means it only applies to Bipolar Electrolytic caps.
Thanks for publishing these results. But since the distortion figures are so low, I have a hard time interpreting them. It's even harder for me to relate them to other sonic aspects as described in other posts in this thread.Can somebody enlighten me, please?
Last edited by a moderator:
Hi johnego,
-Chris
Nope. For a DVM, both the accuracy and range suffers. Oscilloscopes also suffer from problems that an analogue scope doesn't.You can't beat a really good meter from Keysight, or Fluke (handheld). Same goes for the oscilloscope. There are several good brands of oscilloscopes out there. Keysight makes top stuff and even their inexpensive products typically outperform the Rigols of the world.Of course, Rigol is a much better 'scope than most others and certainly the small uP based models. I used to work in a calibration lab and have direct experience with the devices you are talking about. I also check out the new equipment every year, so I'm kept current.
-Chris
Last edited by a moderator:
Since you mentioned tube rectifiers AE50, there's something freakish that's been lurking at the back of my head. I will try to build a low voltage tube rect supply to power the analog side of my dac sometime. When I started to learn diy, tube was my first exposure & to this day to me, there's no better sound to be had then tube rectifier with clclc or lclc
Hi Karl,
-Chris
It's a simple bridge or full wave rectifier circuit. The resistance depends on your current draw and allowable voltage drop. Of course a high current supply with high dynamics will demand a lot from the design. There isn't a single value that will work. Mostly you want to control the charging current spikes.
-Chris
VX has well over 20dB more 3rd harmonic (-146dBV) than Silmic (below the -165dBV test set noise floor), so I'm not sure you reached the right conclusion. Still, for a circuit with adequate PSRR, it might not matter.
Thank you very much, Anatech... 
And thank you Monte for the very interesting posts! Monte, I have quite a few power supply big caps, they are still in specs. I would send them to you for proper lab tests, lol! Almost sure they still would be usable.
Anatech: there is a PDF doc online that shows about rectifiers spikes and a few combination of resistor+capacitor across a common 1N4004/7 that works brilliant, without the need of very fast diodes. I can't find the doc, but will try later...
And thank you Monte for the very interesting posts!
Monte, I have quite a few power supply big caps, they are still in specs. I would send them to you for proper lab tests, lol! Almost sure they still would be usable.Anatech: there is a PDF doc online that shows about rectifiers spikes and a few combination of resistor+capacitor across a common 1N4004/7 that works brilliant, without the need of very fast diodes. I can't find the doc, but will try later...
Last edited by a moderator:
In general, I find the large can capacitors tend to last a lot longer than smaller can caps. When re-capping an old piece of equipment, I usually don't have to replace the very large caps, which is good because sometimes these larger caps have no modern replacement, or some strange physical size or terminal arrangement.
Thanks for the offer, but I'll pass on the extra tests!
I performed these tests as part of a design for new piece of audio equipment, which calls for a low impedance, long life power supply capacitor. The distortion tests were done just to see if there is any significant difference among various otherwise suitable caps at high signal induced ripple currents, and apparently there is. The tests were by no means exhaustive, as I focused on the modern 105 degree, long life parts, with only a few other random parts thrown in for comparison. For my application, distortion at these low levels will not be a problem, But, I had no idea what order of magnitude to expect, so now I know.
Last edited by a moderator:
No, it's the opposite. In polarised caps, NEG is connected to the can (directly as in many axial type, or through the liquid as in many radial type). When the NEG is grounded, the can act as RFI shield. In case of axial MKP without can, the outer foil represents the can to act as RFI shield. And this outer foil is the NEG lead (as in polarised caps), or in case of BP/NP is the shorter lead (This outer foil or shorter lead is represented by the marking of 'BP/NP/NON-POLAR' in Rubycon and Nichicon). So, connecting the outer foil to POS defeats the above shielding logic, UNLESS the liquid or can (which shield the outer foil==short lead) is not connected to the short lead but to the long lead. Let's find out.
Wow. Never tried that stupid idea
In what position precisely?
No polarity. Only the outer foil == short lead == BP/NP marking in Nichicon and Rubycon.
Last edited by a moderator:
Your the smart one you should know far more then me Johnego. In any location you’ll hear a difference. Key to ones progress, be open minded & think out of the box.
I was just surprised that I have missed the possibility. I heard this from you, you heard this from someone else (probably Peter). I wondered what he had in mind when trying out this the first time. And the result is night and day you say? Wow.Peter mentioned that his experienced was with BGN. I don't think I will ever purchased such expensive cap anymore. But you said nichicon BP too? That's not too expensive and I already have a few, especially MUSE BP.Bipolar caps are usually like two capacitors in series, which means that many parameters are worsened. Such is usually good in feedback and signal coupling only. But I'm very curious to find out. So, I will use Nichicon BP for local decoupling, probably on the output transistor leg of an amplifier. And also for op-amp local bypass where I have used cheap low ESR such as Rubycon ZL. I will update the result here if I have the time and motivation to do it.
I just compared it to the other caps that you measure (it's second to the Silmic, isn't it?). It's only an indication of good quality, not necessarily good sound.
With little amount of electrolyte, it will dry out quickly. I like the idea of small footprint of 2u2 MKT caps (such as from Vishay).
I don't think we can relate such distortion with many sonic aspects. As you have probably guessed.
Of course, anatech
I will guess that this case will be similar to when you said we have different taste regarding the red Cerafine. I think I know the sounds you are referring to.
Last edited by a moderator:
Maybe we should ask Monte to help measure. Cannot explain why but you really hear the difference. If we connect base on longer leads to positive , the sound is thicker & resolution though is there is not so open but the minute you flip the cap its like window glass has been cleaned with Windex. Now whether this is to yours or others taste is subjective so one just have to try. Forgot dynamics increases as well. Yes Peter Daniels was the first who reported this effect but over the yrs I just forgot all about it till Eldam reminded me of his findings.
Nope Im a greedy guy Johnego. I want everything including warmth realism. Transparency, details etc is just part of the package all there but even sounding from top to bottom. That’s the reason why I’ve been tweaking my system for almost 3.5 yrs. but its already almost done now. Next thing to do is to package it up.
Nope Im a greedy guy Johnego. I want everything including warmth realism. Transparency, details etc is just part of the package all there but even sounding from top to bottom. That’s the reason why I’ve been tweaking my system for almost 3.5 yrs. but its already almost done now. Next thing to do is to package it up.
@Monte McGuire: Hi Monte ... & thank you very much for outlining the setup you use. It actually is quite similar to what I did use although I have chosen to focus a bit more on HF distortion phenomena (and economy ;-)) and therefore use an AD7760 evaluation board (2.5 MHz Fs, 24 bits resolution) in my setup. I also use one of Victor's oscillators (12 kHz version). I wonder though if there is some kind of systematic inconsistency here ... ? This morning I re-read some of Cyril Bateman's capacitor articles and - besides learning something new - I still notice that his distortion figures appear to be way higher than yours. One caveat to this could be that I do not completely know how to work with the dBV figure - so just to make sure .. Your:
@sumotan: Your observation about BP capacitor "polarity" is really interesting
.. Any chance I can ask you to say whether it is the leg on the Nichicon ES BP with the "BP" writing on the casing that you connect to ground - or the other one?
Cheers,
Jesper
corresponds to a distortion level of 0.0000025% 2nd harmonic (ref. 1 Vp)? In any case CB in his sixth article on capacitors measured both an Elna Silmic (I think it was Silmic I) and a Black Gate 100uF and with a 12 VDC bias 2H for the Silmic was - 94.4 dB, (0.002%), and for the BG -95.9 dB (0.0016%). Both were 25 VDC types so reasonably close voltage-wise to the one you have measured. Test signal levels were 0.1 V AC. This is quite a difference - about 800 times - so I just wonder how this happens ... I would have liked to attach a pdf of CBs paper but the file is 13 kB too big for forum limits. Should you be interested send me a PM with an email & I can email it to you.
@sumotan: Your observation about BP capacitor "polarity" is really interesting
.. Any chance I can ask you to say whether it is the leg on the Nichicon ES BP with the "BP" writing on the casing that you connect to ground - or the other one?Cheers,
Jesper
Last edited by a moderator:
Hi Jesper
The BP’s comes with one lead which is longer which is suppose to be the positive. In case that you already cut the leads, through varification on all my collection of
BPs, the none printed side of the Green sleeve is the positve leg. The last couple of days I’d been mucking around with this on PS of my SdTrans384 & my findings are consistent be it BG N, Nx or BP. Just flip the BP on my amp module & again I hear the difference as well. Now I have a new headache cause there’s some BG Ns on this module it means I have to individually flip & sound test again. Lol
The BP’s comes with one lead which is longer which is suppose to be the positive. In case that you already cut the leads, through varification on all my collection of
BPs, the none printed side of the Green sleeve is the positve leg. The last couple of days I’d been mucking around with this on PS of my SdTrans384 & my findings are consistent be it BG N, Nx or BP. Just flip the BP on my amp module & again I hear the difference as well. Now I have a new headache cause there’s some BG Ns on this module it means I have to individually flip & sound test again. Lol
Last edited by a moderator:
@sumotan: Many thanks ;-)
Wrt. diodes I reckon there's (at least) one more topic to consider: Snubbing of the transformer so at to avoid peaking. There's an article about this in Linear Audio volume 5, however, the short answer is to series connect a 1k resistor with a 1 nF capacitor on the leads of the transformer secondary wires.
Cheers,
Jesper
Wrt. diodes I reckon there's (at least) one more topic to consider: Snubbing of the transformer so at to avoid peaking. There's an article about this in Linear Audio volume 5, however, the short answer is to series connect a 1k resistor with a 1 nF capacitor on the leads of the transformer secondary wires.
Cheers,
Jesper
Last edited by a moderator:
Welcome Jesper.
A series resistor after the bridge is a better solution but the problem is if its a power amp then it becomes a headache. Besides I dont have any equipment to measure spikes under load. Theoritically a higher resistance is best to kill spikes but lots of heat, big watt resistors voltage drop etc. Perhaps the easier solution is high current inductors but ????
A series resistor after the bridge is a better solution but the problem is if its a power amp then it becomes a headache. Besides I dont have any equipment to measure spikes under load. Theoritically a higher resistance is best to kill spikes but lots of heat, big watt resistors voltage drop etc. Perhaps the easier solution is high current inductors but ????
Maybe the noise problem mentioned here about fast diodes is less so in fast diodes with soft recovery.
There was an article by Rick Miller in a '94 The Audio Amateur issue where the RFI peak that "regular" diodes, like those from 1N400X family produced, was shown.
Ultra-fast diodes that do not have this soft recovery might produce the noise mentioned here.
There was an article by Rick Miller in a '94 The Audio Amateur issue where the RFI peak that "regular" diodes, like those from 1N400X family produced, was shown.
Ultra-fast diodes that do not have this soft recovery might produce the noise mentioned here.
I might have to try flipping (if needed) my large sized Muse bp the next time I’m in there...Regarding the good/bad sound of the silmic vs VX, entirely subjective, but I prefer the clear, layered sound of the nichicon, along with the tube/transformer sound I guess. Silmic II is ok in a subwoofer, (if I had one maybe) but it doesn’t help any realism imo. The tendency to have a lower imd with that is another bonus, if you want more info, will have to read up.
Last edited by a moderator: