Supercapacitor b*ll*cks
Posted 16th February 2015 at 12:50 AM by abraxalito
Updated 6th September 2015 at 05:28 AM by abraxalito (Red Wine it isn't, just the same designer)
Updated 6th September 2015 at 05:28 AM by abraxalito (Red Wine it isn't, just the same designer)
Here's something very misleading in the 6moons preview of Vinni Rossi's (of Red Wine Audio fame) latest supercapacitor powered kit. They have 18 * 350F supercaps in the box and they say '...the total rating becomes a whopping 6300F'. However each capacitor is only 2.7V so presumably to get any audio signal at all out, they need to operate these caps in series. Caps in series don't multiply, they divide so the total capacitance drops by a factor of 18. Hence only 19.4F. Not a small difference from 6300F or did I miss something?
https://www.6moons.com/audioreviews2/vinnierossi/3.html (the text at this link has since been updated to remove the 'total rating...whopping' remark).
I think I may have missed something - they aren't using all the supercaps all the time, rather there are 9 discharging and 9 charging. So the total capacitance in the audio circuit at any time is 350/9 = 39F. Still an impressive amount of capacitance but the problem is that the ESRs add too. Here's the datasheet for the Maxwell 350F ultracap they appear to be using - https://www.tecategroup.com/capacitor.../BC_series.pdf
Note the ESR is 3.2mohm so for 9 in series that gives 29mohm. That's the initial ESR - over the 10 year projected life of the cap, its spec'd to rise to double the initial value. So if you were looking for a low impedance supply, supercaps don't look to be the best choice but here its not just the impedance that's crucial, rather their ability to be isolated from external power. However if isolation is the primary aim a ferrite-cored transformer can offer impressive isolation, probably lower capacitance than across the relay contacts Vinni Rossi's using to disconnect his ultracaps from the mains.
So overall this product looks to be fairly typical for audiophile fare - solving problems that really don't exist in practice in order to offer some kind of market differentiation. What d'ya all think?
Update - now that the completed review is up on 6moons, the hype has been removed and the series combination of caps is confirmed as 33.89F - https://www.6moons.com/audioreviews2/vinnierossi/3.html.
That figure though looks to be due to a typo - putting 350 into the calculator as 305.
Update2 - there's now a Stereophile review up and in the comments Mr Rossi confirms that the series array of supercap's output voltage drops around 4V before the other bank is switched in to replace them. And he also mentions that there are downstream regulators - hence the low ESR of the supercaps doesn't come into play. https://www.stereophile.com/content/vinnie-rossi-lio-modular-integrated-amplifier-measurements#WdsROLWXZXSGj8rI.97
Thus its looking that achieving the best isolation is the reason for the choice of cap bank switching in this product. I'd be interested to know how good isolation could be gotten from a ferrite trafo without all that cap bank-switching work-up.
https://www.6moons.com/audioreviews2/vinnierossi/3.html (the text at this link has since been updated to remove the 'total rating...whopping' remark).
I think I may have missed something - they aren't using all the supercaps all the time, rather there are 9 discharging and 9 charging. So the total capacitance in the audio circuit at any time is 350/9 = 39F. Still an impressive amount of capacitance but the problem is that the ESRs add too. Here's the datasheet for the Maxwell 350F ultracap they appear to be using - https://www.tecategroup.com/capacitor.../BC_series.pdf
Note the ESR is 3.2mohm so for 9 in series that gives 29mohm. That's the initial ESR - over the 10 year projected life of the cap, its spec'd to rise to double the initial value. So if you were looking for a low impedance supply, supercaps don't look to be the best choice but here its not just the impedance that's crucial, rather their ability to be isolated from external power. However if isolation is the primary aim a ferrite-cored transformer can offer impressive isolation, probably lower capacitance than across the relay contacts Vinni Rossi's using to disconnect his ultracaps from the mains.
So overall this product looks to be fairly typical for audiophile fare - solving problems that really don't exist in practice in order to offer some kind of market differentiation. What d'ya all think?
Update - now that the completed review is up on 6moons, the hype has been removed and the series combination of caps is confirmed as 33.89F - https://www.6moons.com/audioreviews2/vinnierossi/3.html.
That figure though looks to be due to a typo - putting 350 into the calculator as 305.
Update2 - there's now a Stereophile review up and in the comments Mr Rossi confirms that the series array of supercap's output voltage drops around 4V before the other bank is switched in to replace them. And he also mentions that there are downstream regulators - hence the low ESR of the supercaps doesn't come into play. https://www.stereophile.com/content/vinnie-rossi-lio-modular-integrated-amplifier-measurements#WdsROLWXZXSGj8rI.97
Thus its looking that achieving the best isolation is the reason for the choice of cap bank switching in this product. I'd be interested to know how good isolation could be gotten from a ferrite trafo without all that cap bank-switching work-up.
Total Comments 9
Comments
-
Maybe the best way to describe it is to say the supercapacitors store a total of seventeen thousand Coulombs of charge.
Since 1 ampere = 1 Coulomb per second, when you've got 17,000 Coulombs you can supply ten amperes DC for 1700 seconds. Let us send 5 amps DC to the Left channel and 5 amps DC to the Right channel. A little calculator exercise will show you that 5 amps DC will drive a 4 ohm loudspeaker to 50 watts RMS (and twice that for an 8 ohm speaker).
The announced LIO amplifier stages are rated for only 25 watts, so these seem quite manageable.
HOWEVER it would be dangerous/silly to discharge the supercapacitors all the way from 100% to 0%. Instead assume they discharge from 100% to 90% (giving up 1/10th of their total charge). Now they can supply ten amps DC for "only" 170 seconds before recharging commences.Posted 16th February 2015 at 03:00 PM by Mark Johnson
-
Yes I agree it would be a much fairer way to describe the array of supercaps to talk about charge. Or even to talk about energy storage. However it is rather common to flaunt uF in power supplies, meaningless as it is without knowing the working voltage.
I can't imagine how an amp would be able to use the charge all the way down to 0V so I agree that the caps would only be useful over a relatively small proportion of their working voltage.Posted 16th February 2015 at 04:00 PM by abraxalito
-
Hmmm, I've thought of trying something like this concept - the trick is to completely isolate the charging circuitry from the playback area - unless there is zero, audible cross interference there [I]is [/I]no point ...
The proof of the pudding is in the ... . Experiments should determine whether there is significant gain in SQ performance using this more complex approach, and whether that gain is worth the money. In my own high energy gainclone, there was no noticeable variation in the tonality while the amplifier was playing off "battery" power, having pulled the mains lead from the wall socket.
So, solving problems? IME, there are numerous ways of doing "it" -whatever works, and/or floats your boat ...Posted 17th February 2015 at 12:00 AM by fas42
-
Right - so what problem is being solved here by having the two banks of supercaps and switching between them? I rather suspect its all down to marketing narrative - once a company has established its story (which for Red Wine was in a major part the isolation from mains power) then it can't go back on that. Even though sufficient isolation from mains power is possible without resorting to the complexities of supercap charging/switching.Posted 17th February 2015 at 12:27 AM by abraxalito
-
I sense some new math in the first comment here by Mr Johnson. He is correct in that 1A = 1C/s but he forgets that any old current can not be delivered into a load unless there is sufficient potential to do so. To state that 5A can drive a 4 ohm load to 50W is a bit misleading when not also stating that the potential must be maintained at or above a voltage equal to V = P/R. Assuming he is talking about clean peak power, then V = 50/4 or 12.5 Vpk. Now think about how the charge reaches the load - it starts in a capacitor bank and then goes through an amplifier with some efficiency before reaching the load. A capacitor discharges according to an exponential law - that is the voltage across the terminals falls exponentially as charge is removed.
Because of the need to maintain a certain voltage to power the load, and because of the exponential decay of the capacitor's potential, it would seem that the duration that the isolated cap bank can support the load is much less than what one might think using Mr Johnson's argument. I would guess that the limiting factor is actually the voltage that the cap bank can support at any given instant that determined when the re-charging cycle must begin. Even when the average playback level is low, you need voltage headroom for short duration music transients if you want to claim "high fidelity" reproduction that is true to the source. Thus the caps must remain highly charged to retain a high potential that can serve the amplifier as needed by program demand.Posted 23rd February 2015 at 04:53 PM by CharlieLaub
-
Very good points. I was picturing a ripple waveform as I was reading your writing - this supercapacitor powered unit is just like a vastly slowed down version of a direct mains powered unit. The ripple voltage is still there - I wonder how much ripple they accept on their 24V rails? That's obviously rail noise and with 9 supercaps in series the impedance isn't particularly low.Posted 24th February 2015 at 04:24 AM by abraxalito
-
If the capacitors are in series then the capacitance is no longer the algebraic sum of the values printed on the can. Just as the "total resistance" of two 100 kohm resistors in parallel isn't 200 kohms.
I suppose they can wiggle out of that by saying the capacitance reflects the physical parts count, and I doubt the intended market for those gadgets is savvy enough to appreciate the distinction anyway.Posted 25th February 2015 at 05:26 AM by rjm
-
Right. Even if the ripple voltage is still high, if it is slowed down to be discharging and charging over several seconds it is essentially constant when compared to music signals, e.g. Fripple << Faudio or Fripple ~ 0.1Hz while 20Hz < Faudio <20kHz. This could be considered advantageous in an of itself. But as other have pointed out, how does the cap bank really compare to a good quality PS across the board, and in the balance is it really any improvement or just a new approach that doesn't really improve on the tried and true transformer>CRC>linear regulator arrangement.Quote:Originally Posted by abraxalito
Very good points. I was picturing a ripple waveform as I was reading your writing - this supercapacitor powered unit is just like a vastly slowed down version of a direct mains powered unit. The ripple voltage is still there - I wonder how much ripple they accept on their 24V rails? That's obviously rail noise and with 9 supercaps in series the impedance isn't particularly low.Posted 27th February 2015 at 06:06 PM by CharlieLaub
-
Yes its constant on the timescale of music signals, except for the effect of the ESR. Since the ESR of the total bank is about 30mohm this is going to result in some audio-induced ripple voltage, assuming that the powered circuits aren't pure classA. If we assume that ripple below the audio band doesn't matter then a 30mohm ESR gives the same ripple at 20Hz as roughly 0.25F. So I reckon a cap bank (using normal caps) bigger than 250,000uF will out-perform this supercap bank in practice.Posted 28th February 2015 at 01:09 AM by abraxalito
