It didn't explode....... YET (aka niges amp)

[jkeny]

Rick
Quality of diode has no bearing - use a low dropout one to prevent too much heat & size it according to how much current you expect to pass

[gstew]

I've been using one of Ian Canada's UCPi boards set to 3.3V to power a Raspberry Pi DAC with good success. He has 2 325F-350F 2.7V Ultracaps in series with a 1K resistor across each which is supposed to keep them fairly balanced. I charge them together with a single 3.3V LT3042-based regulator that has been set to current-limit at about 200mA... takes awhile, 1/2 hour or so, to charge up from dead.

I run them in the 'should-be-patented' float charge mode, using an Uptone Audio LPS-1 set to 5 volts as the voltage source. I need to check with an AC-connected supply, but with the LPS-1 the sonic difference between supply connected and not is almost un-noticeable.

I have not compared it to an LiFePO4 setup... I will soon, have some here. BUT with this supply this little DAC is seriously good and one of my best overall. The UCPi powers the DAC chip directly along with the 1.2V rail via another LT3042 regulator on the DAC board.

You can see more info in Ian's thread here, including pictures, the UCPi schematic, and pix of his prototype LiFePO4 supply board:

https://www.diyaudio.com/forums/power-s ... ply-4.html

Rick, the 2 Ultracap with a 1K balancing resistor across each and a single supply would be an easy one to implement for you... single regulator, should work still very well with an AC-connected supply.

John, I found your comments comparing Ultracaps to battery power interesting and that will prompt me to do that comparison soon. Ian did that and preferred the Ultracap. OTOH, he also found sufficient sonic penalty to float-charging either type that he's developing his supplies to run only from the batteries or Ultracaps.

Later!

Greg in Mississippi

[jkeny]

I've been using one of Ian Canada's UCPi boards set to 3.3V to power a Raspberry Pi DAC with good success. He has 2 325F-350F 2.7V Ultracaps in series with a 1K resistor across each which is supposed to keep them fairly balanced. I charge them together with a single 3.3V LT3042-based regulator that has been set to current-limit at about 200mA... takes awhile, 1/2 hour or so, to charge up from dead.

Yea, at a lowish charging current this will work but I tested charging at higher currents (2-3Amps for faster charge up)& it's on the borderline of balanced - the 1K resistor just doesn't pass enough current to maintain balance. I will be experimenting with a lower value/higher current resistor but the active balancers are probably more accurate & some can handle this.

I run them in the 'should-be-patented' float charge mode, using an Uptone Audio LPS-1 set to 5 volts as the voltage source. I need to check with an AC-connected supply, but with the LPS-1 the sonic difference between supply connected and not is almost un-noticeable.

Again, if you want to have a supercap PS which is able to deliver 2 Amps current, the 1K resistor may not be sufficient.
You say "almost un-noticeable" - what do you notice? Is your PS floating or grounded ?

I have not compared it to an LiFePO4 setup... I will soon, have some here. BUT with this supply this little DAC is seriously good and one of my best overall. The UCPi powers the DAC chip directly along with the 1.2V rail via another LT3042 regulator on the DAC board.

Yes, I discovered a long time ago that stable (low impedance), low noise PS improves all audio devices but especially digital audio devices which often require current in gulps, which is related to processing of digital signals

You can see more info in Ian's thread here, including pictures, the UCPi schematic, and pix of his prototype LiFePO4 supply board:

https://www.diyaudio.com/forums/power-s ... ply-4.html

Rick, the 2 Ultracap with a 1K balancing resistor across each and a single supply would be an easy one to implement for you... single regulator, should work still very well with an AC-connected supply.

John, I found your comments comparing Ultracaps to battery power interesting and that will prompt me to do that comparison soon. Ian did that and preferred the Ultracap. OTOH, he also found sufficient sonic penalty to float-charging either type that he's developing his supplies to run only from the batteries or Ultracaps./quote]Yes, I've seen Ian's thread & it's very interesting although I believe it is overkill unless someone can demonstrate the audible difference between float charging & running directly just from caps/LiFePo4 - I haven't heard any audible difference myself - having a ground loop or using a particularly noisy charging PS might change this? But I feel that a ground loop should be tracked down & sorted out anyway - using a battery or supercap is not the way to address this - as ground loops are system wide issues & best handled in a system wide fashion, not by patching up one noise channel

[gstew]

Yea, at a lowish charging current this will work but I tested charging at higher currents (2-3Amps for faster charge up)& it's on the borderline of balanced - the 1K resistor just doesn't pass enough current to maintain balance. I will be experimenting with a lower value/higher current resistor but the active balancers are probably more accurate & some can handle this.

Again, if you want to have a supercap PS which is able to deliver 2 Amps current, the 1K resistor may not be sufficient.

Thanks. Very useful. At least for now, I can live with a slow charge period before use, I generally power my systems on and leave them running without daily on-off cycles, only powering-down for extended absences or predictions of thunderstorms. My amps have a useful standby mode I use for non-listening periods (they come up to full throat in about an hour).

BUT for a product, that is not acceptable.

You say "almost un-noticeable" - what do you notice? Is your PS floating or grounded ?

The difference is VERY small... could even be my imagination. VERY slight lowering of dynamics and resolution.

The Uptone Audio LPS-1 (and 1.2) are floating dual-bank Ultracap-based supplies with output regulation. Prior to using Ian's UCPi, they were my best supplies for digital setups. I expected direct LiFePO4 and Ultracaps to better them and based on what I hear with this one setup, the Ultracaps do (and I fully expect the batteries to do so too).

Yes, I discovered a long time ago that stable (low impedance), low noise PS improves all audio devices but especially digital audio devices which often require current in gulps, which is related to processing of digital signals

Agree totally and have always appreciated yours and others thoughts and examples, here and on other outlets.

Yes, I've seen Ian's thread & it's very interesting although I believe it is overkill unless someone can demonstrate the audible difference between float charging & running directly just from caps/LiFePo4 - I haven't heard any audible difference myself - having a ground loop or using a particularly noisy charging PS might change this? But I feel that a ground loop should be tracked down & sorted out anyway - using a battery or supercap is not the way to address this - as ground loops are system wide issues & best handled in a system wide fashion, not by patching up one noise channel

I've wondered at length what is causing Ian to hear the difference that y'all don't. Some thoughts to try, but I won't speculate here. Again, I will be trying my Ultracap and LiFePO4 setups with both the floating Uptone Audio supplies and AC-connected supplies to try and duplicate y'alls & his results.

My preference is to use a float-charging setup as I don't like the idea of a non-constant voltage level. I REALLY like how y'all do that.

AND so far I have a predisposion towards the Ultracaps in that they aren't damaged if discharged too low, so no in-line relays or transistor/FET switches needed.

BUT I'm willing to be converted either way on either based on listening.

The other thing that I don't see with many who are following Ian's thread is grappling with what it takes to implement this safely while gaining the full benefit. Using these supplies well is not a plug and play thing, but requires a rethinking and redesign of how your gear is configured and how power is used and distributed, IMHO. I'm still getting my head around all of this!

Later!

Greg

[jkeny]

A useful description of the various techniques for balancing supercapacitors here - you can see that passive resistor balancing is dodgy & very slow (especially with 1K resistor) & leads to shortened supercap life - a 50ohm or lower is used in the paper

Our technique of active balancing using individual voltage regulators that we use on batteries is one of the preferred methods
https://hal.archives-ouvertes.fr/hal-00411482/document

[nige2000]

A useful description of the various techniques for balancing supercapacitors here - you can see that passive resistor balancing is dodgy & very slow (especially with 1K resistor) & leads to shortened supercap life - a 50ohm or lower is used in the paper

Our technique of active balancing using individual voltage regulators that we use on batteries is one of the preferred methods
https://hal.archives-ouvertes.fr/hal-00411482/document

Cant help thinking that in digital circuits where the current demands are small the resistor method will work ok
but in the likes of power amps or class a circuits the movement of current will be too vast for the balancing resistors to keep individual cap voltage control?

i suppose when u look at a tesla battery pack you can see the amount of control that needs to be in place when current demands are high
im hoping its just horses for courses

[jrling]

A useful description of the various techniques for balancing supercapacitors here - you can see that passive resistor balancing is dodgy & very slow (especially with 1K resistor) & leads to shortened supercap life - a 50ohm or lower is used in the paper

Our technique of active balancing using individual voltage regulators that we use on batteries is one of the preferred methods
https://hal.archives-ouvertes.fr/hal-00411482/document

Cant help thinking that in digital circuits where the current demands are small the resistor method will work ok
but in the likes of power amps or class a circuits the movement of current will be too vast for the balancing resistors to keep individual cap voltage control?

i suppose when u look at a tesla battery pack you can see the amount of control that needs to be in place when current demands are high
im hoping its just horses for courses

My view too. 1k balancing resistors on series Maxwell 350F Supercaps float charged by A123 LiFePO4 pack, works perfectly for me.

[jkeny]

Yes, this seems reasonable - digital circuits draw little current but what if you are not constantly charging the supercaps (when you go on holidays, for instance or when you just don;t want to be permanently charging the supercaps)? These are not like batteries that hold their charge with little to no self-discharge - caps have quite high self-discharge & will discharge to zero in about 2 or 3 days without resistors across the cap terminals - even faster with resistors across terminals.

From full discharge, recharging can be done slowly to avoid too much recharge current that the balancing resistor can't handle - don't know how long this will take to recharge fully if you keep within the balancing current capability of the balancing resistor. A bit of a long wait, I suspect?

If you don't stay within this current handling, you run the risk of one cap going over its 2.7V max & being damaged

Using a battery to keep the caps charged will work but you do have the added complexity of battery charging, relays, etc

Just playing devil's advocate & seeing possible issues - how it all works out in practice may be different?

[nige2000]

Suppose only way to know is to test bench it at different current demands measure individual cap voltages incl power up and down scenarios

[jrling]

jkeny - you are right about the difficulties of charging Supercaps from zero if left to discharge. And for you as a manufacturer the situation is quite different.

I use LiFePO4 batteries charging the Supercaps for reasons of SQ mainly. I think the combo of the two in parallel gives the best PSU I could think of. Really low ESR and therefore fast to react to current demands and with batteries enough reserve power to keep everything fully on the target voltage. And, in my case, no other active components such as diodes, regulators or active balancing circuits - so as unadulterated as one can get. And cheap!

Actually I also keep a Maplin XM20X charging the batteries/Supercaps always on. The power consumption at idle is negligible, but it keeps them all at exactly the 'right' voltage; I am guessing that also means that the Supercaps never go more than a mv or so below the charged voltage and so will last pretty well indefinitely. But I am only a year in!