advertisement


Supercapacitor Power Supply tour

I brought this question over from the supercap DAC thread

Hi John

Not sure if you are willing to go into technical detail, just wondering how you maintain charge in the supercaps. I don't imagine you've taken the overly elaborate path taken by Uptone over swapping banks of caps as they deplete and charging the offline bank. Are you floating a regulated voltage across the caps?

Thanks
Mark
Sure, Mark
but I'll bring this question & answer over to the supercapacitor PS thread, as I wanted to not confuse the two its already seen in audiomisc & following posts.

I not going to give away all the techniques I'm using but IMO opinion people focus on the wrong thing when they are trying to build better power supplies for audio. From what I've seen there is a focus on breaking the connection with mains AC & ground as the source of all the ills of power supplies. I also thought this when I started down the road of using battery power many years ago. My experience taught me that this is not the magic' of battery/supercap power & this isolation is only of importance if you have a ground loop which should be identified & addressed system wide, anyway. Floating PS supplies bring their own issues when used in mixed systems, anyway.

So people put a lot of time & effort & money into making batteries or supercapacitors into floating supplies (usually with a switch to defeat this isolation ) :) Inevitably this requires putting a voltage regulator on the output of the battery or supercap & this kills all the magic. Why? I'm not exactly sure but I suspect that it's due to the feedback used in most regulators & it's generation of dynamic instability/noise when in use?

So yes I avoid this & supply output power directly from supercaps exactly as I did when using batteries. The power which energises/charges the batteries/supercaps has zero influence on the sound when of a reasonable quality. This has been shown many times by people unplugging the PS from my battery powered devices while playing & hearing absolutely no difference. The same applies to supercaps although it powers down in a short time.

So, yes I use linear low noise regulators to charge the supercaps & balance the voltage across them - this regulates their output voltage & can deliver voltage directly from the supercaps. There are other parts to this picture that I'll keep to myself but the above is the crucial information as far as sound quality (power quality) is concerned.
 
I've been a bit confused as there seem to have been two parallel similar threads. The other being here

https://www.pinkfishmedia.net/forum/threads/tour-of-supercapacitor-powered-audio-devices.223682/

I'm curious to know the output impedance of the 'supercapacitors' at higher frequencies than the few kHz I've seen in basic spec sheets. e.g. in the 100kHz to MHz region. Anyone know?
Yea, sorry about the confusion, Jim - this is the offer of a tour of a supercap power supply. The othe rthread is an offer of a tour of DAC, SPDIF converter, USB HUB powered internally by supercaps
Hope this explains it?

I can't answer your question about supercap impedance in the MHz region. Yes, I've seen it stated that some voltage regulators measure at lower impedance than batteries but my & others experience is that all voltage regulators tried (yes including the lowest noise LT3042/5) degrade the sound from the raw battery/supercap power. Now there may well be a discrete regulator that betters the supercaps but I've tried some Paul Hynes ones & they don't (& I would consider his discrete designs are one of the best).

So maybe it's not just about impedance or maybe we are measuring it incorrectly?
 
I brought this question over from the supercap DAC thread


Sure, Mark
but I'll bring this question & answer over to the supercapacitor PS thread, as I wanted to not confuse the two its already seen in audiomisc & following posts.

I not going to give away all the techniques I'm using but IMO opinion people focus on the wrong thing when they are trying to build better power supplies for audio. From what I've seen there is a focus on breaking the connection with mains AC & ground as the source of all the ills of power supplies. I also thought this when I started down the road of using battery power many years ago. My experience taught me that this is not the magic' of battery/supercap power & this isolation is only of importance if you have a ground loop which should be identified & addressed system wide, anyway. Floating PS supplies bring their own issues when used in mixed systems, anyway.

So people put a lot of time & effort & money into making batteries or supercapacitors into floating supplies (usually with a switch to defeat this isolation ) :) Inevitably this requires putting a voltage regulator on the output of the battery or supercap & this kills all the magic. Why? I'm not exactly sure but I suspect that it's due to the feedback used in most regulators & it's generation of dynamic instability/noise when in use?

So yes I avoid this & supply output power directly from supercaps exactly as I did when using batteries. The power which energises/charges the batteries/supercaps has zero influence on the sound when of a reasonable quality. This has been shown many times by people unplugging the PS from my battery powered devices while playing & hearing absolutely no difference. The same applies to supercaps although it powers down in a short time.

So, yes I use linear low noise regulators to charge the supercaps & balance the voltage across them - this regulates their output voltage & can deliver voltage directly from the supercaps. There are other parts to this picture that I'll keep to myself but the above is the crucial information as far as sound quality (power quality) is concerned.

That's a detailed reply, John, appreciated. I think many agree there is little logic or benefit in putting regs after supercaps - I was quite disappointed with the popular Californian psu that takes that approach. Maybe if you have really cruddy earthing or mains it has a benefit but my own crude experiments with supercaps directly feeding the audio circuitry have shown whorthwhile gains in detail and dynamic resolution (hate that word, but hey...).
 
That's a detailed reply, John, appreciated. I think many agree there is little logic or benefit in putting regs after supercaps - I was quite disappointed with the popular Californian psu that takes that approach. Maybe if you have really cruddy earthing or mains it has a benefit but my own crude experiments with supercaps directly feeding the audio circuitry have shown whorthwhile gains in detail and dynamic resolution (hate that word, but hey...).
Just shooting the breeze here, Mark but I've added my experience with battery/supercap power supplies together with my experience with USB signal isolation & come to a premise that what is at play in these scenarios is low level noise modulation & it's holding back our audio systems. Of course I'm aware that I may be fitting my observations into a theory rather than the other way around & I may well change my premise sometime if I get new information. But, atm, this is my best theoretical fit to my observations - I believe what is going on in feedback based voltage regulators is causing fluctuating noise - the USB signal wires are also carrying some common mode noise which when isolated improves the sound but also this isolation introduces jitter on the signal. But why should jitter on a USB signal be of concern - well if I remove it by reclocking, it improves the sound. My explanation is that this jitter is introducing some fluctuating noise somewhere within the USB handling/processing stage. In all these examples it's a secondary mechanism at play - noise modulation.

Now the other main plank to understanding this is to understand how this could possibly be audible & we need to turn to auditory processing for possible explanations for this. I can't say I've cracked this & have a watertight, well researched explanation but I have inklings that fluctuating noise riding on the audio signal which in itself isn't audible as noise, can change the perception of what we hear. For instance, again I'm drawing on my experience with all my efforts in this area, I hear a far more solid & defined soundstage with individual elements also being more solidly portrayed. There are other aspects but they are all perceptual elements that effect the overall sound but one aspect that first struck me & people always comment on is the bass - it becomes much better defined & textured - often appearing as more pronounced but again this is perceptual.

Now a lot of this points towards it being timing that is improved (we use ITD more so at lower frequencies for sound location) but we may have to look into it in a more nuanced way & not just look for a measured improvement in timing. If there is fluctuating noise riding on the audio signal, this can effect our perception of exactly when a sound starts - so this becomes a bit fuzzier than if there was no fluctuating noise. This perceptually effects the timing of the start of sounds & removing it can solidify the ITDs & hence the sound stage.

One way of trying to tease this out would be to try to develop a way of injecting noise modulation ate various levels & spectrum/timing into music? signal to test for perceptual effects & try to get a handle on what characteristics, if any, are responsible for the perceptual effects I'm suggesting.

The final piece of this puzzle for me, is why has this not shown up in measurements? My guess is that one has to setup tests to look for this condition as it doesn't just fall out from the standard tests. I also strongly suspect that it's not an easy test setup or run.

Hmm, didn't know I would write this when I started but that's what I ended up doing :)
 
Yea, sorry about the confusion, Jim - this is the offer of a tour of a supercap power supply. The othe rthread is an offer of a tour of DAC, SPDIF converter, USB HUB powered internally by supercaps
Hope this explains it?

I can't answer your question about supercap impedance in the MHz region. Yes, I've seen it stated that some voltage regulators measure at lower impedance than batteries but my & others experience is that all voltage regulators tried (yes including the lowest noise LT3042/5) degrade the sound from the raw battery/supercap power. Now there may well be a discrete regulator that betters the supercaps but I've tried some Paul Hynes ones & they don't (& I would consider his discrete designs are one of the best).

So maybe it's not just about impedance or maybe we are measuring it incorrectly?

I suspect you're correct that impedance isn't the only factor, or may not be the most relevant one.

FWIW I gave up on standard active regulators decades ago. They tend to be obsessed with clamping to a specific dc voltage. Whereas low noise, etc, may simply not need that but a 'smooth and stable' voltage even if it isn't *exactly* 5.0000 Volts, or whatever it says on the pack.

Having tried them and decided they were all a bit of a pest I DIY'd a 'smoothing' supply based on combining a zener reference as a 'guide' but then slugged with an active 'supercapacitor' as per the 730 diagram shown in the link from

http://ukhhsoc.torrens.org/makers/Armstrong/700_Range/index.html

That gave lower noise and better behaviour for the preamp than any active IC I could find.

I'm curious about the supercaps, though, because I may want to experiment with one in a different context of slugging LF noise, etc on a powerline. However I'm wondering about their HF impedance and what noise they may add when current flows in/out of them. Again, much earlier experience with battery power made me realise that batteries generate noise, some types being quieter than others. But I've never tried a supercap as they are 'new toys' from my POV. :)
 
Fundamentally supercaps, being ionic devices, tend to a lower impedance as frequency rises. Capacity also decreases as frequency rises but they both tend to stabilise at quite low (khz) frequencies. Probably similar to lithium batteries.

My own experience (I can't be arsed to put together the necessary test rig to actually investigate it) would suggest they are low noise at least within audio frequency range based on crude testing. Using a supercaps to run a crystal oscillator driving a dac "seems" to be an improvement but way down in the margins compared to batteries or a good reg - so could just be my subconscious blowing smoke.
 
I suspect you're correct that impedance isn't the only factor, or may not be the most relevant one.

FWIW I gave up on standard active regulators decades ago. They tend to be obsessed with clamping to a specific dc voltage. Whereas low noise, etc, may simply not need that but a 'smooth and stable' voltage even if it isn't *exactly* 5.0000 Volts, or whatever it says on the pack.

Having tried them and decided they were all a bit of a pest I DIY'd a 'smoothing' supply based on combining a zener reference as a 'guide' but then slugged with an active 'supercapacitor' as per the 730 diagram shown in the link from

http://ukhhsoc.torrens.org/makers/Armstrong/700_Range/index.html

That gave lower noise and better behaviour for the preamp than any active IC I could find.
Yes, couldn't agree more & it was something I felt like adding to my diatribe post but left it off as I thought it was too long anyway. I too believe that it may well be the characteristics of the noise fluctuations that are perceptually important rather than the actual noise after all we can hear through steady background noise easier than noise which fluctuates perhaps in semi-random patterns - we are creatures of habit or is that patterns or is it the same thing? Our auditory perception is certainly a pattern processing engine.

I'll have a look at that schematic, thanks. Is it using a capacitor multiplier as active "supercapacitor"?

I'm curious about the supercaps, though, because I may want to experiment with one in a different context of slugging LF noise, etc on a powerline. However I'm wondering about their HF impedance and what noise they may add when current flows in/out of them. Again, much earlier experience with battery power made me realise that batteries generate noise, some types being quieter than others. But I've never tried a supercap as they are 'new toys' from my POV. :)
What batteries did you find generated noise? I presume as it was "much earlier" you didn't have LiFePO4 available?

I'm happy to send you a few supercapacitors for experimenting with/measuring or if you want to test/measure my supercap power supply, I'm also happy to oblige
 
Fundamentally supercaps, being ionic devices, tend to a lower impedance as frequency rises. Capacity also decreases as frequency rises but they both tend to stabilise at quite low (khz) frequencies. Probably similar to lithium batteries.
Not quite - the equivalent circuit for modelling supercapacitors is a perfect capacitor paralleled by a leakage resistance both in series with ESR so if you look at this paper on testing supercapacitors (They used a 3F Yuden supercap) it says:
Above 10 Hz the magnitude and phase approach 100 mΩ and 0°. ESR dominates this region. • In the region between 100 µHz and 10 Hz, capacitance controls the impedance. Magnitude-versus-frequency is linear (on the log-log Bode plot) with a slope of –1 and the phase approaches –90°. • Below 10 µHz, the impedance begins a transition back towards resistive behavior as leakage resistance becomes dominant. This transition is incomplete, even at 1 µHz. EIS spectra of real devices rarely give much information about leakage resistance, because its effects are seen at impractically low frequencies.
My own experience (I can't be arsed to put together the necessary test rig to actually investigate it) would suggest they are low noise at least within audio frequency range based on crude testing. Using a supercaps to run a crystal oscillator driving a dac "seems" to be an improvement but way down in the margins compared to batteries or a good reg - so could just be my subconscious blowing smoke.
I agree that oscillator power is crucial but once stable & low noise, there probably isn't much improvement left apart from changing the oscillator itself - there's lower hanging fruit.
 
Having PMed with Dan, it seems like a 25V version is feasible for Naim HiCap users.
I propose to put something together along these lines for some user(s) to evaluate to see if it's worth proceeding with looking at the ergonomics of half case size.
Having looked into the HiCap a bit it looks like it has 3 input sockets on the back for signal input (CH1 & CH2 & signal common) but looking at the inside shots of the HiCap it appears that CH1 has a wire looped to each of the 3 input sockets (same for CH2 & signal common) - which means all three sockets are the same i.e only 1 input can be used? This couldn't be right - what am I missing?

The CH1, CH2 & signal common then loop into a 5 pin DIN? socket which also outputs two 24V power lines
Is this nearly correct?
 
The ch1 and ch2 signal travel back down the same single Naim interconnect (5 pin SNAIC) into the 4th socket - that's the same one that supplies the 2 x +24V to the pre/ source. A second 4 pin SNAIC is needed to then take the ch1 and ch2 signal to a stereo power amp.
 
The ch1 and ch2 signal travel back down the same single Naim interconnect (5 pin SNAIC) into the 4th socket - that's the same one that supplies the 2 x +24V to the pre/ source. A second 4 pin SNAIC is needed to then take the ch1 and ch2 signal to a stereo power amp.
OK, thanks Dan - my mistake was thinking socket 1, 2 & 3 were input signal sockets - they're actually for output signals - I knew I was a bit of a dummy about Naim.
So I only need 5 pin socket & one 4 pin socket on my prototype, I presume?
 
That's right, unless someone uses a pair of mono power amps (NAP 135) then you'll need a pair of 4 pin signal output (and earth return) sockets. I think the earth of the power amp input section is also referenced back to the HiCap as the single zero volt point for the whole system - I might be wrong, there are plenty on here who will know for certain.
 
I'll have a look at that schematic, thanks. Is it using a capacitor multiplier as active "supercapacitor"?


What batteries did you find generated noise? I presume as it was "much earlier" you didn't have LiFePO4 available?

I'm happy to send you a few supercapacitors for experimenting with/measuring or if you want to test/measure my supercap power supply, I'm also happy to oblige

Yes, a combination of capacitance multiplier and using a zener. I've snipped the relevant part of the diagram and put it here:

http://jcgl.orpheusweb.co.uk/temp/cmult.png

For the application I wanted the order of of presented capacitance and series resistance this gave. It also suppresses hf noise from the zener.

My experience with batteries was decades ago. Using them to bias detector elements in very sensitive far-infrared detectors that got cooled to liquid helium temperature. So ultra low noise was vital. IIRC in the end we settled for 'mercury cells' at the time as having the lowest current noise levels. But it is so long ago I can't recall details. However, yes, batteries tend to generate a form of noise due to their chemical reactions. And react in a 'clumpy' way, depending on how they were made, etc. In normal terms, the noise level is low, but present.

I'd be interested in a supercap. But the interest I have - prompted by this thread - is perhaps very different to the conditions of use you have.

In my case I'm wondering if a suitable supercap would give a lower output impedance if put in parallel with a stack of NiCads I'm using with a radio set. However this is circa 12V and I probably only want the order of a Farad. Not the very large capacitances I've seen when I did a quick search. Must admit I'm wary of high voltage 'capacitors' of 100F or more with low impedances. An "ooops!" with one of them might have quite dramatic consequences! 8-]
 
Yes, a combination of capacitance multiplier and using a zener. I've snipped the relevant part of the diagram and put it here:

http://jcgl.orpheusweb.co.uk/temp/cmult.png

For the application I wanted the order of of presented capacitance and series resistance this gave. It also suppresses hf noise from the zener.

My experience with batteries was decades ago. Using them to bias detector elements in very sensitive far-infrared detectors that got cooled to liquid helium temperature. So ultra low noise was vital. IIRC in the end we settled for 'mercury cells' at the time as having the lowest current noise levels. But it is so long ago I can't recall details. However, yes, batteries tend to generate a form of noise due to their chemical reactions. And react in a 'clumpy' way, depending on how they were made, etc. In normal terms, the noise level is low, but present.

I'd be interested in a supercap. But the interest I have - prompted by this thread - is perhaps very different to the conditions of use you have.

In my case I'm wondering if a suitable supercap would give a lower output impedance if put in parallel with a stack of NiCads I'm using with a radio set. However this is circa 12V and I probably only want the order of a Farad. Not the very large capacitances I've seen when I did a quick search. Must admit I'm wary of high voltage 'capacitors' of 100F or more with low impedances. An "ooops!" with one of them might have quite dramatic consequences! 8-]
I strongly suggest that you should try Lithium Ferrous nanoPhosphate (LiFePO4) batteries for your application - the best manufacturer is A123. They are every low noise & impedance & run at a nominal 3.2V. I'm not sure what you want the supercap in parallel with the Nicads but I reckon these batteries will surprise you?
 
I strongly suggest that you should try Lithium Ferrous nanoPhosphate (LiFePO4) batteries for your application - the best manufacturer is A123. They are every low noise & impedance & run at a nominal 3.2V. I'm not sure what you want the supercap in parallel with the Nicads but I reckon these batteries will surprise you?

They may do. I'll have a look, thanks. :)
 


advertisement


Back
Top