Tír Na HiFi

@jrling - thanks for the report - what wattage 1K resistor do you use between caps? I presume this has to be sized according to what instantaneous current is likely to be drawn. I've seen reports of series Maxwells without these resistors holding their balance?
The bleeder resistors are 1/4 watt in my set-up. The current flowing through is tiny as the cells should be in balance all the time. I did it because I was advised to do so by an expert and it gives peace of mind. Out of balance serial Ultracaps could be v bad news and definitely not good for SQ as they would be fighting each other.

The bank switching is not a runner as far as I'm concerned - it requires a regulator on the output to maintain a constant voltage which defeats the purpose
Agreed and far more complicated and expensive. Double the cost and a bit more. And would sound worse rather than better as they are degrading the output voltage all the time as they discharge.

I've seen talk of chemical reaction noise in batteries but when I first looked into A123 LiFePo4 batteries the characteristic that allowed such low internal impedance & hence high current delivery was the large surface area of the cathode by using iron nanophosphate i.e. very small particles. Along with other techniques they ensured that the ions move very quickly across the electrodes - hence the high power delivery. Anyway, all this means that there is very much less chemical reaction noise particularly when low currents are being drawn typical of DACs.
I have read the same and rate the A123 cells highly. Never had a problem with any of them ever. Their storage voltage is fantastically constant.

The idea of putting a battery behind the capacitors doesn't make a lot of sense to me either as we still need all the relays/switches to ensure the battery is isolated & prevented from being drained - do the ultracaps bring a significant upcheck in sonics?
Well you are right but I just disconnect the whole PSU at end of session. The batteries keep the Ultracaps nicely topped up for their small discharge current. But the whole extra benefit comes from combining the batteries and Ultracaps to give an even more tough PSU than Ultracaps on their own. Difficult to measure benefit, but noticeable in my case - but I am using the PSU on an audio PC not a DAC, so who knows on other applications? At least it doesn't cost much to try.
For a 5V supply two Ultracaps like in Ian's board can run happily at 2.5V each in series or 2.7V to give 5.4V if that works for the application to give a little voltage sag compensation in the PSU chain.

@jkeny,

Ian has not reported that online yet, which is why I was a little reserved on the details. He was charging it via one of his LT3042 reg boards. I don't have much more info than that.

I agree that the constantly-charged UC is the most convenient option and I'm interested in getting it as close as possible to the pure UC solution. One thing on the setup Abartels uses is that he does not use a regulator at the output... he charges the bank up to the max he feels is safe to use on the target circuit, swaps it in, then lets it slowly drop to the voltage he feels is the lowest allowable, then swaps in the freshly charged bank. Bank switching is set via timer, not voltage measurements. I personally would prefer a constant controlled voltage level, which the constantly-charged setup provides. It not only provides the benefits @jrling mentioned, but I suspect there are benefits for the powered circuits too.

@jrling, thanks for your report, very interesting info AND real-world too, like that from @jkeny and many others here!

Greg in Mississippi

@jrling - are you using 1K resistors across the caps as in this diagram?

jkeny - mine are in a case and quite a long process to take a picture that would tell a thousand words.
But this is what I copied (with 5 not 6 Ultracaps in series)
Hoe that answers your question. It works perfectly.

Jonathan

That's great, thanks

and here's a link to Maxwell's technical paper on series cell balancing -
http://www.maxwell.com/images/documents ... ancing.pdf

jrling wrote: ↑Sat Jul 07, 2018 2:48 pm and here's a link to Maxwell's technical paper on series cell balancing -
http://www.maxwell.com/images/documents ... ancing.pdf

Great article, thanks!

Attached is a little board I built to charge three batteries, provides 10V and 120 ma.
Should provide a little more current if needed, the 7.4 ohm resistor sets the 317 to provide 160 ma.

With this circuit, if you provide at least 13VDC to the 317 input, it will charge three batteries in series at 3.3V each.

You just connect the batteries up in series, connect the ground point to neg of the first battery,
Q1 emitter to + on 1st battery
Q4 emitter to + on 2nd battery
Q3 emitter to + on 3rd battery

I used a 4 pole switch (just needed three) between the transistors and batteries, so I can keep the batteries from draining if power is not provided to the 317.

Circuit has a kind of class a aspect, with the 317 connected like this, it always draws 160ma of current. If there is no load, then this current goes through Q3, Q4 and Q1, and I used MPSA92's in a pretty small to92 package, so they get a little warm. First version I had a smaller R1, so it drew more current and the MPSA92's got a little too hot for my liking, so I adjusted to what is shown.
BTW, if there is a load to draw the current, then the 92's will stay much cooler, but I wanted to make sure it works ok either way. Whatever the load draws does not go through the 92's.

For more current, using a PNP transistor in a to220 package should work. If you heatsink the to220, should be able to get as much as you could need.
You might also need to do something with the 317. It does claim to be able to put out 1.5A, which is quite a bit of continuous current.
The nice thing here is the batteries will take care of instantaneous current, you just need to make sure you can supply more average current than is used.

This is being used to power a DAC, a ling DAC designed by Richard/abraxalito

Excellent, Randy, thank you!!