I like the Evox-Rifa PEH 169 series. Check the specs, they speak for themselves:
http://www.evoxrifa.com/electrolytic_cat/peh169_105deg.pdf
Of course they would be used in the power supply rather than signal circuitry. Maybe if I was forced to use a cap on the output of a single ended amp then these would be the boys. The older ones had liquid electrolyte that sloshes about. Shake one, nice. This does mean that you can only mount them upright or on their side with the vent at the top.
As far as I know they are the highest quality electrolytics you can get.
Cheers Matt.
http://www.evoxrifa.com/electrolytic_cat/peh169_105deg.pdf
Of course they would be used in the power supply rather than signal circuitry. Maybe if I was forced to use a cap on the output of a single ended amp then these would be the boys. The older ones had liquid electrolyte that sloshes about. Shake one, nice. This does mean that you can only mount them upright or on their side with the vent at the top.
As far as I know they are the highest quality electrolytics you can get.
Cheers Matt.
Some years ago Panasonic made a bad series of low ESR caps broadly used in PC Motherboards that leaked out and since we speak about bad caps...
Every brand could do a mistake like Panasonic did but today it's still regarded as one of the best elcos producer...
I cant' speak for Rubycon's Black Gates because I've never tried it but regarding Elna's Silmic and Silmic II in my 8 years old Sony SACD player there's plenty and not a single one leaked out.
And I can assure you that the player is pretty hot when powered on and as you know heat is an enemy for elcos life...
For sure there's a lot of snake oil in a number of 'audiophile' and 'for audio' components but this is not the case.
Rubycon, Elna, Nichicon and Panasonic are premium brands that made serious products.
DNM and it's Slit foils caps could be a different case but please don't generalize.
Gotta say I agree in general with these comments about brands and designs intended for audio applications.. I generally choose components that are specified for use in audio designs, and you will see a lot of other designers do too.
There were plenty of counterfeit and/or off brand caps showing up in pc motherboards and failing after some period of service. D*ll, H*, and other major US computer vendors had to replace thousands of motherboards with these particular capacitors. (These failing caps were not made by Panasonic or any other reputable Japanese manufacturer, but by a Taiwanese company and its subsidiaries that stole the design information, [industrial espionage is common everywhere] but somehow missed one key ingredient in the chemistry that guaranteed longevity.)
Last edited:
IMHO the best electrolytics for audio are not very often specified for audio at all.
Large cans for powersupplies are crowned by the Sikorels and the Rifa PEH169 and PEH200.
Theese caps can cary enormous ripple currents @ burning hot temps year after year.
Nothing really comes close to theese caps in that sence.
The sonic impact though is somewhat different from each other, so one has to check it out, if they are suitable. The PEH 169 is a very large cap, so have a good look at the dimensions before choosing. The special thing about the Rifa´s is that the winding do not fill up the can completely. It has air around the winding, which should optimize heat transfer, hence the large ripple current capabillity. Also the equivalent series induction is very low and ESR is state of the art. This I think is accomplished by connecting the terminal to the foil 4-8 different places or even more.
Surely a lot of other parameters are optimized, but these are at least visible.
For small signal Elna Silmic has been mentioned, besides Panasonic FM/FC which are very fine caps, but also Sanyo has some interesting ones, i.e. the MV AX series.
Large cans for powersupplies are crowned by the Sikorels and the Rifa PEH169 and PEH200.
Theese caps can cary enormous ripple currents @ burning hot temps year after year.
Nothing really comes close to theese caps in that sence.
The sonic impact though is somewhat different from each other, so one has to check it out, if they are suitable. The PEH 169 is a very large cap, so have a good look at the dimensions before choosing. The special thing about the Rifa´s is that the winding do not fill up the can completely. It has air around the winding, which should optimize heat transfer, hence the large ripple current capabillity. Also the equivalent series induction is very low and ESR is state of the art. This I think is accomplished by connecting the terminal to the foil 4-8 different places or even more.
Surely a lot of other parameters are optimized, but these are at least visible.
For small signal Elna Silmic has been mentioned, besides Panasonic FM/FC which are very fine caps, but also Sanyo has some interesting ones, i.e. the MV AX series.
Last edited:
I do not think Epcos is a part of Evox-Rifa, but BHC and Dectron are.
Also now most Rifa´s are made in the UK instead of Sweden.
Epcos bought a license to manufacture Sikorel from Siemens some years ago, so Sikorel can be found as both Siemens and Epcos products.
Looking at the link above on page 109, you´ll find a graph on impedance versus frequency in theese Epcos Sikorel electrolytics.
I´m quite sure that this is @ or close to state of the art, but anyway this is a very interesting characteristica, as I find this to look very much the way electrolytic caps actual sonic impact is.
I´ve tried the experience of both using electrolytics and avoiding them.
Of course only in decoupling, as smoothing caps are unavoidable.
When no caps are present, electronic circuits suddenly begins to behave as they are supposed to. It is difficult to explain, but to me it simply seems as if the non linearity of impedance shows it self in frequency dependant q values or so. Maybe some would say "not that much", but IMHO this is a source of almost endless worries in audio design.
Exact I think the same.
About
http://www.evoxrifa.com/electrolytic_cat/peh169_105deg.pdf
and the most other brands there is an other interest detail:
If you compare the impedances by the voltage, you will note, that the lowest impedance there is by the 63V versions (check it e. g. by 10.000uF)
This means, not in all cases the highest value of capacitance (according the highest possible mounting space) is the best choice. In some cases, where I normally use 40V version, the 63V Version by a lower capacitance could be better (10.000/63V is the better choice than 22.000uF/40V). I think, by the 63V versions all parasitic effects are at lowest.
By the follow datasheets this mentioned behavior is not so:
http://www.ftcap.de/downloads/elektrolyt/datenblaetter_2009/G2009.pdf
one of my favorite caps for amp power supplies are the follow:
http://www.ftcap.de/downloads/elektrolyt/datenblaetter_2009/GW2009.PDF
large sizes and expensive, but one of the best caps (47.000uF/63V).
Only the mentioned types (post #2) about this weblink could be better, but unfortunately much more expensive:
http://www.diyaudio.com/forums/soli...ster-reference-amplifier-schematic-wanted.htm
Dou you have heard about such high capacitances in non polar MKP versions?
Also WIMA had introduce some new types like this:
http://www.wima.de/DE/supercap_mc_1.htm
not for power amplifiers but interesting for RIAA head amp power supplies
Last edited:
FT caps look nice, but you can still cut the ESR to ½ with PEH169, and ripple capabillity of the Rifa´s are just in a class their own.
So I think availabillity would make the choise for me.
Anyway!
Have a look at this picture.
It is a preamplifier with a separate powersupply.
This is IMHO the way NOT to use elctrolytics, or even design powersupplies.
If you look here @ page 109 http://www.epcos.com/inf/20/30/db/alu_03/01010109.pdf you can see the impedance characteristics of Sikorel state of the art electrolytics.
If you use such massive amounts of capacitors as done on the picture, you´ll normally bring down ESR and impedance, and you even might be lucky not to highten inductance, but the characteristics of R vs Freq will still have the same relation, maybe even worse.
Also by lowering the impedance in the capacitorbank, you could say you move "closer" to the amplifying circuits. This makes the sound even more coloured by the electrolytics.
Smoothing caps should be used for smoothing the energy jerks from the rectifiers, when done with low or hardly any ripple left, regulation or multistage regulation can suppress it further with no or just sufficient decoupling.
Both sonically and electrically this IMHO would be a much more elegant solution, with almost completely linear impedance characteristics.
Have a look at this picture.
It is a preamplifier with a separate powersupply.
This is IMHO the way NOT to use elctrolytics, or even design powersupplies.
If you look here @ page 109 http://www.epcos.com/inf/20/30/db/alu_03/01010109.pdf you can see the impedance characteristics of Sikorel state of the art electrolytics.
If you use such massive amounts of capacitors as done on the picture, you´ll normally bring down ESR and impedance, and you even might be lucky not to highten inductance, but the characteristics of R vs Freq will still have the same relation, maybe even worse.
Also by lowering the impedance in the capacitorbank, you could say you move "closer" to the amplifying circuits. This makes the sound even more coloured by the electrolytics.
Smoothing caps should be used for smoothing the energy jerks from the rectifiers, when done with low or hardly any ripple left, regulation or multistage regulation can suppress it further with no or just sufficient decoupling.
Both sonically and electrically this IMHO would be a much more elegant solution, with almost completely linear impedance characteristics.
Attachments
I dont understand this claim. If I compare the parameters by the 47000uF/63V version about
http://www.evoxrifa.com/electrolytic_cat/peh169_105deg.pdf and
http://www.ftcap.de/downloads/elektrolyt/datenblaetter_2009/GW2009.PDF
I read
1) EVOX/RIFA PEH169MV5470Q 47000uF/63V 75 x 145
ESR 100Hz: 5 milliohms ESR 10 KHz: 4 milliohms ESL: 17nH
and
2) FTcap 47306375069 47000uF/63V 75x69
ESR 100Hz: 5 milliohms ESR 10 KHz: not mentioned ESL: not mentioned
Where you see the half-ESR value by PEH169?
You say additional , that the special thing about the Rifa´s is that the winding do not fill up the can completely. It has air around the winding, which should optimize heat transfer, hence the large ripple current capabillity.
The result is to observe at the sizes (same diameter than FTcap, but more than twice of the hight).
FTcap optimizes their heat transfer through a possibility of heatsink montage.
Is it possible, that this detail causes bader parasitic results by the 10 KHz value of ESR and ESL and therefore this values FTcap don't mentioned?
You are right about the ESR, I think I looked at the other FT datasheet, but still Ripple is very superior on the Rifa´s.
The body shape of the caps, is a result of cooling amongst other matters.
The Rifas are also made for heatsink mounting, hence the larger can diameter, which provides a larger area for heat transfer through the buttom end. Also the winding is pressed hard onto the buttom of the can, which also electrically connects the negative pole to the can.
But I do not think the lower can for the FT will cause any trouble @ higher frequencies. Actually amongst other details, the lower can should theoretically make it more suitable to HF measures.
But a lot of things influences on this kind of stuff, terminal design, number of contact points between winding and terminal, separator, electrolyte electrodes and so on and on.
Be aware that the data provided by the manufacturer always is data for the electrolytic cap, an nothing else. You will probably never be able to implement the caps to theese standards, because PCB tracks, even short cables and this kind of stuff, will ruin the data.
When critical engineers have to optimise these parameters in their design, then the cap contributes with precisely what is stated.
You may find more details for the F&Ts on the Mundorf site.
I remember my surprise at the large sonic variations in the ClassA amps between Rifa, BHC, F&T, BGs and the Sikorel power supply caps, and then when trimming them up with a variety of bipass methods - good fun, actually, and lot's of "lively" discussions!
Curious thing about the big electros that are used in hifi power supplies is that one of the main jobs that they're required to do is provide large short term current transients to the gain stages (rarely sinusoidal) and (at the other end) to also absorb the current surges when the diodes are conducting (and spikes when turning on/off) and the caps are charging and to bury the incoming line noise - we don't seem to get any sort of measurements that relate to this typical behaviour of transient characteristics, etc apart from the few mentions of ESR at a few frequencies so capacitor behaviour under these everyday conditions remains mostly guesswork and anacdotal reporting.
Bipassing behaviour has been well researched and the mechanisms quite well known, but applying it to hifi power supplies is still not directly applicable if you see the power supply in terms of current and energy transfer rather than ripple reduction and rail voltage modulation.
... my 2 cents.
I remember my surprise at the large sonic variations in the ClassA amps between Rifa, BHC, F&T, BGs and the Sikorel power supply caps, and then when trimming them up with a variety of bipass methods - good fun, actually, and lot's of "lively" discussions!
Curious thing about the big electros that are used in hifi power supplies is that one of the main jobs that they're required to do is provide large short term current transients to the gain stages (rarely sinusoidal) and (at the other end) to also absorb the current surges when the diodes are conducting (and spikes when turning on/off) and the caps are charging and to bury the incoming line noise - we don't seem to get any sort of measurements that relate to this typical behaviour of transient characteristics, etc apart from the few mentions of ESR at a few frequencies so capacitor behaviour under these everyday conditions remains mostly guesswork and anacdotal reporting.
Bipassing behaviour has been well researched and the mechanisms quite well known, but applying it to hifi power supplies is still not directly applicable if you see the power supply in terms of current and energy transfer rather than ripple reduction and rail voltage modulation.
... my 2 cents.
Consider using coils too. They can do things a cap can not. I think it is a superiour way to combine caps and coils. You can use them inbetween two caps. A small resistor instead of the coil can help too.
It's amazing what Spice has to say about the HF attenuation benefits of small air coils between large caps.
You can off course tune these spikes to a lower frequency where they are filtered by the caps, and you can dampen them to an extend. Read the snubber article by Jim Hagerman for details.
On bandwidth, you can see those nice diagrams in datasheets and you can off course calculate the resonant frequency from the capacitance and ESL of your preferred caps. To extend the bandwidth it is a good idea to parallel a large elco with a small elco, especially since the smaller cans of the same type have usually about half the ESL of the big cans (e.g. a 22mF & a 2.2mF), but then off course the indutance of the bus bars come into play, and using a plus and minus plane (PCB) will minimise this inductance.
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.
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.