Supercapacitor balancing

[xani]

Hi.

So I've got a bunch of old 3000F maxwell ultracaps and decided to make a jump starter out of it. I tried making simple balancing circuit from just 3 basic 10A rectifier diodes in series for each of 6 caps, which got just about the forward voltage to get the whole thing in 12V range and below the 2.7V per cap

My reasoning was basically "I will charge it with 1 amp max so if few watts of heat goes thru the diodes that's not a big deal, they are rated at 10x that".

So I built it, started charging it (one of caps had already some charge in it), and it promptly evaporated.

What's more it was still going after I turned off charging current and only failed (by exploding the diode) few seconds after I've disconnected charging current.

Am I not understanding something about how the the supercapacitors charge ? It looked to me like the supercap somehow managed to charge above the diodes's forward voltage then put current thru them and burned them, but why ?

[gcewing]

Are you charging it with something that actually limits the current to 1 amp? If not, you have a recipe for dead diodes.

Also, you said that one of the capacitors already had some charge. What voltage was it charged to? If it was even a little above the Vf of its diodes, it could result in quite a large current through them.

You might want to put some low-value resistors between the diode chain and the caps to limit the maximum discharge current that can occur.

[xani]

Are you charging it with something that actually limits the current to 1 amp? If not, you have a recipe for dead diodes.

Lab power supply. That also have max current of 5A so even then shouldn't get to "wire melting" currents (as in "I tested the balancers alone under the 5A current and aside from getting hot nothing bad happened")

Also, you said that one of the capacitors already had some charge. What voltage was it charged to? If it was even a little above the Vf of its diodes, it could result in quite a large current through them.

1V so not anywhere near the forward voltage of 3 diodes.

You might want to put some low-value resistors between the diode chain and the caps to limit the maximum discharge current that can occur.

That was my original plan but after measuring the diodes I've seen forward voltage at 0.9V at 2A current and that was a bit too close for my liking to max cap voltage (2.7V).

Edit: now that I think about it, heating a diode lowers forward voltage so maybe that was the reason ? A bit of current they were passing over "full" cap got them heated up, which caused more current to flow from the cap itself, faster heating, more current, BOOM

[gcewing]

I think your diode fell victim to thermal runaway. As the diode warms up, the Vf for a given current falls, so more current flows, making it even hotter, etc. until it burns up.

That would explain why the diode kept heating up after you stopped charging. The Vf had become smaller than the voltage that the capacitor was charged up to, so it continued to discharge.

Generally, connecting a diode directly across a low-impedance voltage source (which a supercapacitor most certainly is) is a Really Bad Idea. 

Putting a current limiting resistor in series with each group of diodes. Choose the value so that the voltage drop across it at 1A, plus the Vf of the diode at 1A, is still within the voltage limit of the capacitor.

[xani]

Yeah, that's most likely, I completely forgot that the Vf goes down with the temp

[gcewing]

That was my original plan but after measuring the diodes I've seen forward voltage at 0.9V at 2A current

Why 2A? If the supply is limited to 1A, you shouldn't get more than 1A through any diode.

If the Vf is still too high at 1A, you might have to use a Shottky for one of the diodes, or use zeners of appropriate voltage instead.

[xani]

That was my original plan but after measuring the diodes I've seen forward voltage at 0.9V at 2A current

Why 2A? If the supply is limited to 1A, you shouldn't get more than 1A through any diode.

If the Vf is still too high at 1A, you might have to use a Shottky for one of the diodes, or use zeners of appropriate voltage instead.

I was just testing the diodes I had at different currents to see if I can bump up the charging current, no other reason.

The whole thing was kind of ad hoc as I had to get a starter for the weekend as I was trying to revive old car at my mum's place that she "forgot" to run over the winter and had near dead battery, but kinda didn't make it in time

I decided to just re-do properly and I will just get a transistor to open instead of fucking around with hitting right Vf

[Capernicus]

if you have the caps in series,  if its 4 caps you have to put 4x the volts in.