Best electrolytic capacitors

Roederstein EK elcos

Joachim, since you know a lot about components and this is a cap thread:

I happen to have old Roederstein EK type elcos, the ones in red plastic cases, and I see that at least one well kown manufacturer is still using them today. Can you tell me anything about those caps and whether they have any "sprecial" quality due to the plastic case? I was not able to find any specific info about them.
 
You don't know how funny this sounds. I guess my age is showing. C-L-C pi-filters, or C-R-C were the foundation of power supplies before solid state regulation. In tube circuits, you still see both. The high cost of inductors has moved designers to regulators. Not in tubes as much because 25 cent regulators don't like 380 volts.

To the OP's observation, you can throw all the esoteric requirements for magic caps out the window for the main filter bank IF you provide three-tier, properly executed, local bypass where the current dynamics are required. Spend you money as close to the transistors as you can. Pay attention to the return impedance. ( I don't say ground, as that is some abstract never achieved). It is the execution more than the parts.
 
You don't know how funny this sounds. I guess my age is showing. C-L-C pi-filters, or C-R-C were the foundation of power supplies before solid state regulation. In tube circuits, you still see both.

Then you may find it even more funny that I use a choke input filter PS for my transistor amp

Spend you money as close to the transistors as you can. Pay attention to the return impedance.

That may explain why shunt regulators can have a sonic impact as they create a local return for signal induced currents on the supply. Also, for class AB amps, I saw at least one manufacturer put his PS caps right next to the output transistors (Spectral), some others seem to use at least local film bypass caps.
 
Last edited:
One caveat is that coils have interwinding capacitances and at high enough frequencies will no longer filter HF, quite analogous to ESL in caps.

Very true, which is why I specified air coils though probably should have differentiated from those used in crossovers. However it's also very likely that at those frequencies the cap's internal inductance is being shunted by case parasitic capacitance. Proper component Spice models for way up there are lacking, I 'best guess' up to about the AM band.
 
That would be very interesting indeed, these values are not published very often and most diyers including me do not have the tools to measure those parasitics.

For power amps, coils for crossover networks are suitable inductors. As a speaker manufacturer maybe you happen to know typical values? I am led to believe that foil inductors have lower interwinding capacities and may have higher bandwidth, but again that is just a statement on a manufacturer page.

I think it is just the reverse in fact, think about all of that surface area exposed to the adjoining winding(s). I think in a lot of cases the use of an air cored foil inductor is more of an audio fashion statement - it costs a lot more, therefore it must be a lot better, right? I have used them where power handling, (sub-woofer for example) and low DCR were critical concerns, otherwise I don't. FWIW in the few cases where I substituted foils for conventional and properly sized air core types I was not able to hear any difference at all. (Heresy?! :cheeky: ) 😀

IMO They seem in many instances to be nothing more than an excellent way to liberate the unwary of some extra cash - in some cases a lot of extra cash.
 
I think it is just the reverse in fact, think about all of that surface area exposed to the adjoining winding(s)

Which in turns explains why foil inductors has no place in a CLC PSU

Na, I don't think so. In the absence of any real data on interwinding capacitances for comparable wire vs. foil coils we are all confined to guesses. However, note that each winding in a foil cap only has a capacitance to the adjoining winding whereas in a conventional coil there is capacitive coupling between the layers, i.e. quite a number of turns ahead. This is also the reason why you have sometimes dual or multiple chambers (bobbins ?) in transformers/inductors when low parasitic capacitance is imperative.

In the meantime I prefer to believe alphacore who state that their foil inductors have LOWER parasitic capacitance and WIDER bandwidth than conventional coils, exactly what I want for my PS.

The best argument I have is off course that I scavenged my coils from a previous crossover project I had done and did not have to buy new coils. Now, if you send me a pair of your favorite wire coils I will immediately try these (something in the range of 2.2 to 3.9 mH pretty please) 😀

P.S. As an aside to crossovers, I have seen measurements of structural resonances of different types of coils, standard wire, same vacuum impregnated, and foils - I think I still have a picture somewhere that was posted in a German speaker magazine. The foil coils where the only ones that had negligeable resonances. Obviously I cannot comment on whether or not that is audible as I never made any comparison.
 
Last edited:
Na, I don't think so. In the absence of any real data on interwinding capacitances for comparable wire vs. foil coils we are all confined to guesses. However, note that each winding in a foil cap only has a capacitance to the adjoining winding whereas in a conventional coil there is capacitive coupling between the layers, i.e. quite a number of turns ahead. This is also the reason why you have sometimes dual or multiple chambers (bobbins ?) in transformers/inductors when low parasitic capacitance is imperative.

In the meantime I prefer to believe alphacore who state that their foil inductors have LOWER parasitic capacitance and WIDER bandwidth than conventional coils, exactly what I want for my PS.

The best argument I have is off course that I scavenged my coils from a previous crossover project I had done and did not have to buy new coils. Now, if you send me a pair of your favorite wire coils I will immediately try these (something in the range of 2.2 to 3.9 mH pretty please) 😀

P.S. As an aside to crossovers, I have seen measurements of structural resonances of different types of coils, standard wire, same vacuum impregnated, and foils - I think I still have a picture somewhere that was posted in a German speaker magazine. The foil coils where the only ones that had negligeable resonances. Obviously I cannot comment on whether or not that is audible as I never made any comparison.

I think it depends heavily on whose foil inductors you are purchasing, and the quality of the dielectric used because the foil exposes a lot more surface area between windings than does any individual wire in a conventional inductor - and it is no coincidence that you will not find any foil types used in low frequency power and RF applications. (You will however find plenty of conventional magnet wire and litz)

Just because Alpha-Core makes the claim does not mean it is demonstrably true, unless they cite the exact test set-up used and their measurement methodology along with the published data for theirs and a few similar items from the competition.

Capacitance is a function of surface area, dielectric and distance - so given that their dielectric must be pretty special stuff. 😛

See here for why I am sceptical of their claims about capacitance:

Capacitance - Wikipedia, the free encyclopedia - specifically the equation relating to a parallel plate capacitor..

Some foil inductors I worked with had significant resonances at a few hundred kHz which IMO is not bad for their intended purpose, but I had heard of (not witnessed) instances where much worse performance was reported.

As an aside I would be very interested in learning more about the article you cite, possibly if you locate it you could translate what you thought was relevant as a comment or two?
 
Last edited:
I had small RF coils in mind. The two LTspice plots below illustrate this. The teal trace shows the filter formed by 50 ohms in series with a 100 uF ASC poly/oil motor start cap. The cap model was derived from measurements of self-resonant frequency and terminal-to-case capacitance. The downward slope to ~60 kHz is self-explanatory. The cap's internal inductance then takes over. (I don't recall such a low self-resonant frequency but can't find the original measurements.) The capacitance between one of the isolated terminals and the metal body was measured at ~300 pF, the ~30 MHz resonance is caused by the interaction between that parasitic capacitance and the cap's self-inductance.
The green trace shows what happens when a Wruth 100 uH inductor, selected for a value greater than the ASC's self-inductance but otherwise at random from the LTspice library, is inserted after the 50 ohm resistor.
 

Attachments

  • ASC_Wruth.gif
    ASC_Wruth.gif
    8.1 KB · Views: 1,791
I know it sounds counterintuitive. There is only one way to find out and that is by measuring different coils or determining their resonance frequency - and btw., resonances at a few hundred KHz would be pretty good for PS filtering IMHO.

Re. structural resonances, Mundorf show a diagram from the original artcile and use is as a sales blurb, and there they also make the same claim as aplhacore though they may have just copied from them, who knows: http://www.mundorf.com/english 1.1/spulen-folien.htm

The original article is from "Hobby HiFi", one of two speaker builder magazines in this country, and I remember having read the original article though the details escape me (4/2000). Maybe if you ask them they can provide the article, though it looks like they want to sell their back issues: kontakt@hobbyhifi.de
 
I don't recall such a low self-resonant frequency but can't find the original measurements

Typical inductances for this type of foil cap can easily be 50 - 100 nH (I looked up datasheets for Epcos MKV caps some time ago) and that explains the low resonant frequency that you get.

Re. coils for PS supply filtering: There are two types of noise that you might like to get rid of. One is PS ripple, and here you need inductors with several mH, but they have limited bandwidth. The other is RF noise from various sources, and here smaller values but with much more bandwidth are better. So maybe you combine both.
 
Typical inductances for this type of foil cap can easily be 50 - 100 nH (I looked up datasheets for Epcos MKV caps some time ago) and that explains the low resonant frequency that you get.

Re. coils for PS supply filtering: There are two types of noise that you might like to get rid of. One is PS ripple, and here you need inductors with several mH, but they have limited bandwidth. The other is RF noise from various sources, and here smaller values but with much more bandwidth are better. So maybe you combine both.

I can pretty much confirm this from direct experience as well, and with some ASC motor run PPIO I used in the power supply of a 2A3 amplifier - I was just curious.. (The resonance was in the region of 60kHz IIRC.) Note that a lot electrolytic caps are not too good in regards to this, I've seen some inexpensive Chinese electrolytic caps where the self resonant frequency was not much above 20kHz..

I think the idea of combining several inductors of different types to get optimum filtering performance is an excellent suggestion. In addition to this sometimes adding lossy ferrite beads can be helpful. (Make sure upon inspecting their data sheets that they do not have a high Q self resonance which should be a good indication that they are what I am talking about.)
 
Can the Spice programs be used to simulate the actual current charging behaviour of power amp supplies - it's not the usual 100Hz (or 120) half wave voltage waveform but a short current burst complete with nasty spikes at the start and end of it every 10mSec. Plus the usual excitation on the secondary winding and other inductive/capacitive sources.

I don't know if a Spice/Other program will actually sim this way a power supply cap works in practice - perhaps someone can add some info about this?
 
Can the Spice programs be used to simulate the actual current charging behaviour of power amp supplies - it's not the usual 100Hz (or 120) half wave voltage waveform but a short current burst complete with nasty spikes at the start and end of it every 10mSec

I think it can be done quite well, provided you have accurate models for all components and all the relevant data to put into your models. One example is in this article, certainly a good start: http://www.hagtech.com/pdf/snubber.pdf

Also, I think you cannot look at the caps in isolation, it is the whole PS circuit that you need to analyse.
 
There are a wide range of hifi magazines where are tests between different amplifier brands (and different amplifier brands would be compared).

Are there also magazines for the electronic industry, where are tests between different capacitor brands (e. g. 10.000uF/63V from different series and different brands)?

This would be very interesting to know.

In the moment we read the data (like ESR, ESL, etc.) from the data sheets of capacitor manufacturers and trusts that the parasitic resistances with the lowest values also the best available versions. That's not enough in my opinion.
 
Yeah Martin,
I keep coming back to Jim Haggermans work - he makes much better sense than most ...
Pardon this digression - I got some of those fancy Cree SilCarbide diodes to see if their approach to spike control,etc gave any significant sound advantage, but i'm darned if I can get them to sound any better than basic MUR type diodes - bare, snubbed, filtered, trapped (tuned filter cct), anything - simple C supply, R-C, R-C-R-C, C-L-C, added Cmultipliers, just not right for an F3, F5 classA amplifier IMO, but might be okay for others.

Tie,
Many of the "better" cap manufacturers don't specifically make audio caps but industrial componennts and as the requirements of the various caps are quite different, the testing standards are also quite different and this is why the specs are often impossible to compare - ie Epcos Sikorels B41550 and Rifa PEH200 are intended for different use and neither aimed at anything remotely to do with audio, but just happen to be very good for power supply filters.

On the other hand, the F&T factory make quite a few different Audio branded caps (F&T, Mundorf, Jensen, etc) and are intended for audio and are marketed in the smaller quantities but are tested & specified in a totally different way (or should be!) to the industrial caps.

That's why reading cap specs are not really a good indication bout what a particular cap SHOULD do in Audio applications - a lot of guess work, unfortunately.

Incidently, this is my opinion - a users view - many other people know quite a bit about this subject - have a troll thru this site - lot's of good info, but need a bit of patience.
 
Yeah!

Caps are specified according to the purpose they are made for.
Looking at audio caps, it is completely different and often based on sonic impact in certain contexts.
I.e. Duelund is a cap like that, it is not at all based on science, but instead on listening results in a certain context. Duelund has always been against plastic anywhere in audio gear and especiallly in components as i.e. caps.
This Doctrine has lead to the paper in oil caps with both alu, copper and silver foil etc. Whether you find this a good or bad way of makingcaps is up to upu, but it shure is very expensive.
I´ve talked with an old friend, who is actually building speakers for his living, he explained to me, that they were actually pretty good, but the values lacked precission, which was a point of high importance for his X-overs.
Therefor Solen and also Bennic were the ones doing the job for him.

Industrial components or general purpose components very often are the best available. Audio components are often downgrades of the industrial ones. Often science is to some extend replaced by sonic impressions in more or less controlled contexts.
At least you´ll never find audio components used for industrial purposes, both quality and also price is often outside any reason.

Elna is probably the make, where they did the most thorough and seroius examination of various ingredients in caps, have a look here:

Aluminum electrolytic capacitors for audio applications | ELNA
 
JH, to disgress even further: I use IXYS DSEI 30-06 fast/soft recovery diodes for power amps and 11DQ10 & 31DQ10 Schottkys for preamps, but I never made comparisons to other diodes as I have barely enough time to get my projects finished. This is why I try to "research" the optimal solution beforehand, like outboard PS, multiple filter stages (LCLC for my power amp), additional RF filtering, etc.. Also, I believe that if you filter the PS sufficiently and use decent soft recovery/low noise diodes they "should" not be audible.

FWIW, the following link shows measurements of RF spikes, for schottky diodes !!!, and most importantly it shows what happens if your snubber resistor is too small or too big: http://www.irf.com/technical-info/appnotes/an-968p8.pdf

Industrial components or general purpose components very often are the best available. Audio components are often downgrades of the industrial ones. Often science is to some extend replaced by sonic impressions in more or less controlled contexts.

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.

Also, the original post was referring to coupling caps, which off course have a different effect than power supply caps. For PS caps I would think that high bandwidth (low ESL) is a major criterion. For coupling caps, I belive that structural resonances can have an audible effect, and ELNA Silmics claim to reduce those resonances (if I understood them in the first place 😀).

Speaking of resonances: WIMA produce a special "low" noise MKS cap - MKS-4-LN - which are literally quieter and produce less audible noise that their standard range: WIMA. Not that these caps should be recommended as coupling caps but it shows that caps have mechanical resonances that can be measured - and heard. Even the most die-hard objectivist will not claim that Wima is an "audiophool" company producing marketing blurb.

Just my 2 cents
 
Thanks Martin,

please excuse the continuing the offtopic ... it's sorta relevant!

I also thought the Shottkys would have very small current spikes, on and off - very useful article that "irf" one - required reading for everyone on this thread.

http://www.irf.com/technical-info/appnotes/an-968p8.pdf

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!

In the case of the Single ended amplifier, the DC "blocking" cap operates in a very different way to the power supply caps even tho the uF size is often similar.
I currently use the Epcos Sikorels for full range, but for midrage upwards, would use the BHC T Network Caps (cleaner mids) - I have been meaning to see if a combination of the two would give better results, but ....

Good to see you continuing your "projects"