Can Ultracaps be used in *amplifier* PSUs?

[absolutk]

The last 10 years have been fascinating for power supply technologies in audio and non-audio applications. Batteries and Ultra capacitors are widely used in DAC designs and GaN FETs are gaining popularity in SMPS and amplifier designs. Until recently, Ultracaps were limited to fairly low voltage ratings (<5.5V), making them impractical for anything but source-related power supplies.

However, I came across the following two New Arrivals at Mouser:

XTM-18R0626-R by PowerStor/Eaton
61F storage
18V rated
22 mOhm ESR
$137 ea

BMOD0009 P024 B02 By Maxwell/Nesscap
9F storage
24V rated
139 mOhm ESR
$213 ea

Is there any drawback to using one of these to Amp PSUs? Even if used in series, it would offer orders of magnitude more capacitance than the mF of capacitance used in PSUs.

For example,

I use Folsom’s Antipole PSU with a 16V transformer (23 VDC rectified). So the Nesscap could be drop-in, or if headroom is recommended, then two in series.

Same for my Neurochrome Power-686 which is powered by a 25V transformer (35 VDC).

Thanks!

 

[Douglas Blake]

While they do appear to improve reserves of energy by leaps and bounds, you also have to consider the initial inrush of current to get them charged up. I'm betting that would kill your average bridge rectifier pretty quick.

 

[eduard]

Hello,
I remember the French l'Audiophile magazine started using them in an mc preamplifier in the eighties. But surely at that time they had some disadvantages too. They used a mix of big batteries , big capacitors and a one farad cap on each side. If they would have only advantages they would have been used in many more applications for sure. Greetings, Eduard

 

[analog_sa]

Is there any drawback to using one of these to Amp PSUs?

A complex charge/discharge solution would be needed in order to implement these at high voltage supply rails. Even a mock-up that lets you hear the effect these may have on the sound of an amplifier is a serious and dangerous project.

 

[jan.didden]

The nice thing with using these for a power amp, is that they can provide the burst high currents required by the amp now and then, while charging them with a continuous charging current of less than say 1A. Possibly even only a few 100mA.

So you get the high dynamic power reserve to run the power amp, while needing only a puny charging circuit. This is because even with high dynamic current load bursts, the average power an amp needs is only a handful of Watts.

It is a very interesting possibility.

Jan

 

[absolutk]

The nice thing with using these for a power amp, is that they can provide the burst high currents required by the amp now and then, while charging them with a continuous charging current of less than say 1A. Possibly even only a few 100mA.

So you get the high dynamic power reserve to run the power amp, while needing only a puny charging circuit. This is because even with high dynamic current load bursts, the average power an amp needs is only a handful of Watts.

It is a very interesting possibility.

Jan

This is precisely the intention. Isn’t that also the goal of putting massive amounts of capacitance on PSUs? Should the reserves run out, one would be no worse off than the PSU’s capability as designed.

@Douglas Blake
You are right. There might need to be a current limiting mechanism for a set amount of time. Or some serious heat sinking on the rectifier.

@analog_sa
I’m not sure I follow. I’m not referring to a complex dual bank supply like Vinnie Rossi, which provides a floating supply, completely off the grid. I’m referring to the replacement of standard caps in a regular PS design, just so a big power reserve is available to the amplifier. Why is using 9F at 24V any more dangerous than using, say, 0.1 F at 24V in the same exact design?

 

[analog_sa]

Why is using 9F at 24V any more dangerous than using, say, 0.1 F at 24V in the same exact design?

Because of the maximum short circuit current they can pass when fully charged.

Same with charging: once charged up to a nominal voltage a normal rectifier/transformer will have no problem keeping them charged up, but not in the initial charging stage when the current will need to be limited. I don't believe these can be dropped into an existing design without something melting down.

It is not so much that the ESR is low, it is comparable to standard electrolytics, but the time it takes for a full charge.

 

[CBS240]

Maybe slow start the transfo to limit inrush current and ramp up the charge.

61F at 18V, 22mR is an awfully lot of stored energy with a significant SS current.. Could be dangerous if shorted, like accidentally shorting a car battery.

 

[jan.didden]

Because of the maximum short circuit current they can pass when fully charged.

Same with charging: once charged up to a nominal voltage a normal rectifier/transformer will have no problem keeping them charged up, but not in the initial charging stage when the current will need to be limited. I don't believe these can be dropped into an existing design without something melting down.

It is not so much that the ESR is low, it is comparable to standard electrolytics, but the time it takes for a full charge.

With supercaps you would not use a traditional xformer/rectifier, you would use a current source, or current limited voltage source, to charge the supercaps.
You may need to wait say a minute after switch-on until the supercap is charged, then start playing and keep the supercap charged.

This needs a different mind set.

Example: it takes 48 seconds to charge 10F to 24V with a charge current of 5A.

After that, assume that you supply a 2 x 50W stereo amp, with the typical music duty cycle of 10%, and 50% efficiency. Just some ballpark figures to get a very rough idea.
So you need to supply long-term 20W at 24V = about 800mA.

Jan

Jan

 

[Mark Tillotson]

The nice thing with using these for a power amp, is that they can provide the burst high currents required by the amp now and then, while charging them with a continuous charging current of less than say 1A. Possibly even only a few 100mA.

So you get the high dynamic power reserve to run the power amp, while needing only a puny charging circuit. This is because even with high dynamic current load bursts, the average power an amp needs is only a handful of Watts.

It is a very interesting possibility.

Jan

This whole argument fails if you play dub reggae (or organ music or bagpipe music or any one of many other genres with strong LF content) at high volume. Then you need a conventional full power supply anyway since the peak to average ratio is small, and the supercap represents only a marginal (fractional dB) improvement in headroom for excessive cost.

I think you are better off catering for headroom for short percussive transients which otherwise can intermodulate nastily. That takes supply voltage headroom rather than extra energy storage since a mains supply must always have storage good for several milliseconds of full load.

For supercaps to make sense the occasional loud transient must last seconds, but not minutes. This is probably a better match for home cinema more than music, where the odd explosion and such will demand high power for the subwoofer, but not for extended periods.

Its also a good match for motor supplies where the thermal limits of the motor preclude extended operation over 100% of the continuous power limit, but most motors can go to ~300% of continuous rating for short periods.

 

[jan.didden]

Yes good points. Continuous full power 6dB dynamic range music sounds like white noise to me, so I don't cater to it

But yes, if you do, things are different.

Jan

 

[cbdb]

If the only advantage is stiffer rails, is the extra cost and complexity worth it? Guess it depends on the amp, but hasnt Power supply rejection been solved.

 

[billshurv]

This whole argument fails if you play dub reggae (or organ music or bagpipe music or any one of many other genres with strong LF content) at high volume. Then you need a conventional full power supply anyway since the peak to average ratio is small, and the supercap represents only a marginal (fractional dB) improvement in headroom for excessive cost.

Any measurements of crest factor for organ music? whilst much popular music is compressed to death I can't see it being a problem for well recorded Organ in an acoustic setting (aka church/cathederal).

 

[indianajo]

Any measurements of crest factor for organ music? whilst much popular music is compressed to death I can't see it being a problem for well recorded Organ in an acoustic setting (aka church/cathederal).

The heat sink for a 100 W Allen organ amp, 2 MJ802's, is 6" x 6" x 6" with a lot of fins. They expect to put out some serious power for a long time. Organ installers are warned not to use woofers designed for rock or disco PA; they don't have the heat rejection necessary even if the enclosure is ported to go down to 16 hz.

 

[billshurv]

I am well aware that an Allen organ can fill a cathederal (Chichester used one in the 80s when I could still sing). However this has zip to do with domestic reproduction in a room not much bigger than an organ loft!


DR tool output is a poor substitute for accurate crest factor calculation, but I'll pull a file off my server tomorrow and see what it gives.

 

[billshurv]

Most of my Organ music is on vinyl for the delicious irony. But I pulled off an album from the server, picked the loudest track and ran through the Foobar DR plugin. Sample of one so usual bag of salt, but this looks a lot less challenging than a large number of recordings from an average power perspective.

 

[MarcelvdG]

Supercapacitors have a rather poor lifetime compared to long-life aluminium electrolytics: 1500 hours at 65 degrees C instead of 5000 to 10000 hours at 85 or even 105 degrees C. Both improve a lot at lower temperatures. When you can keep them at 20 degrees including self-heating, you can get supercapacitor lifetimes of the order of 10 years.

 

[martin clark]

Why not just design (or select) an amplifier with reasonable-to-great PSSR, and avoid messing-about with such band-aids?

 

[PRR]

Any measurements of crest factor for organ music? whilst much popular music is compressed to death I can't see it being a problem for well recorded Organ...

Depends on a lot of factors. I know an organ work which, if played in disco woofers, would probably call the fire department. OTOH much organ is limited by blower power: you can get peaks from the regulator but not long-term.

I can't find that heavy-pedal new-organ track. So I turned to Virgil Fox' Filmore recordings. Peak, 50mS RMS, and track-average. For contrast I held my nose (and muted my speakers) and analyzed Tarzan Boy by Baltimora, which was a fair Pop hit in many lands.

Even allowing for 2dB-6dB unused headroom on the 1970s Fox tracks, it seems clear that the Pop track has much higher average power.

 

[analog_sa]

Why not just design (or select) an amplifier with reasonable-to-great PSSR, and avoid messing-about with such band-aids?

Because we live in the real world? Changing the capacitor type is audible on any amp. Or any opamp, no matter what the PSRR. Just an inexplicable empiric observation, like everything else that matters