Supercapacitor balancing?

Status
Not open for further replies.

ian field

Joined Oct 27, 2012
6,536
That does not sound possible (to me). The thing that seems most relevant is that voltage in a capacitor bank is a dependent variable - charge is the determining factor.

The charge on each negative plate equals the charge on each adjacent positive plate. For the charge on one to be reversed, they would all have to be reversed...right?


When you say that you have done this with smaller capacitors, do you mean you have set up a series bank, charged it up, then rapidly discharged it - and some of the capacitors have ended up with negative charge while others are positive?

I think this could only happen if you put in some wacky extra circuitry to make it happen.
Capacitors in general have wide tolerance, the lowest capacitance in the string will hit zero first, and may be back-charged by others that held more charge.

This often happens with Ni-Mh & Ni-Cd batteries - a weak cell will be repeatedly reverse charged by the others - they don't last long when that happens!
 

Thread Starter

snowdrifter

Joined Aug 13, 2013
43
That does not sound possible (to me). The thing that seems most relevant is that voltage in a capacitor bank is a dependent variable - charge is the determining factor.

The charge on each negative plate equals the charge on each adjacent positive plate. For the charge on one to be reversed, they would all have to be reversed...right?


When you say that you have done this with smaller capacitors, do you mean you have set up a series bank, charged it up, then rapidly discharged it - and some of the capacitors have ended up with negative charge while others are positive?

I think this could only happen if you put in some wacky extra circuitry to make it happen.
I had 1 cap in the 3-series bank that was reading higher than the other two. Threw a resistor across it to bleed the current out. When I checked them to be sure they were discharged, 2 caps were reading ~-.1 and the cap that was reading higher was reading ~.2v

I don't see how this is unreasonable. The capacitor's charge is caused by an imbalance of electrons between the two plates, correct?

As it discharges, the electrons shift from one plate to another. If you have another capacitor(s) continuing to push current through after it reaches the 0v mark, it would shift the electrons to the other side and give it a negative charge/reversed polarity







And as a matter of learning, why does the balance circuit need to pass hundreds of amps? The bleed resistors I had across each cap before kept them fairly well inline and they only had a draw of ~50-60ma. All the caps read within +/- .05v of eachother.

Only reason I don't want to stick with the resistors is the parasitic draw they have is an annoyance
 

LDC3

Joined Apr 27, 2013
924
And as a matter of learning, why does the balance circuit need to pass hundreds of amps? The bleed resistors I had across each cap before kept them fairly well inline and they only had a draw of ~50-60ma. All the caps read within +/- .05v of each other.

But also so I have some overhead since the capacitors will be charged with a 400 amp charge.
Your first post states that you will be charging the capacitors at a rate of 400 A. In order to maintain that rate to charge the capacitors, the voltage will be above 16.8 V. When the overvoltage protection turns on, it needs to handle the current flow (the current won't stop immediately).
 

Thread Starter

snowdrifter

Joined Aug 13, 2013
43
Your first post states that you will be charging the capacitors at a rate of 400 A. In order to maintain that rate to charge the capacitors, the voltage will be above 16.8 V. When the overvoltage protection turns on, it needs to handle the current flow (the current won't stop immediately).
Ahh ok I see what you are saying.


What I'm using to charge the caps is voltage regulated. It'll dump whatever power it needs (up to it's 400 amp output) to try and get the bank to 16.8v, regardless of what the individual cap voltages are. Hopefully that makes sense? Sorry if I wasn't clear on that before.
 
Last edited:

wayneh

Joined Sep 9, 2010
17,496
The offered solution is to never charge your system beyond the capacity of your weakest capacitor. If you need more capacity, buy more capacitors, or pick and choose them to cull out the low-performers.

There is no practical way to control the movement of charge around faster than the very low impedance connections you already have. Anything that could keep up with a 1000A discharge would cost more than the capacitors.

I should add that I'm just giving my position and I may be totally unaware of some important principle for this application, ie. full of it. That's why I mentioned modeling. I'd be much more comfortable if I saw this worked out in a simulation, one that estimates the magnitude of a potential "balancing" current. I'm assuming worst case, that the balancing circuitry might have to handle very large numbers. If someone can prove this is not the case, this could become much easier.
 
Last edited:

Thread Starter

snowdrifter

Joined Aug 13, 2013
43
Well that's the issue: The capacity of my weakest capacitor will change wont it? A higher drain rate will keep dragging the voltage down. Over time, to zero. Correct?
 

BLUESHIFT

Joined Aug 23, 2013
24
No, what WAYNEH is proposing is actually not a balancing circuit, so there is zero balancing current. I agree with him, for ridiculous charge currents, it's the right approach - simpler / easier / cheaper than anything else discussed.


The circuit should compare the voltage of each 'cell' (actually each parallel pair of capacitors) to a floating voltage reference (i.e. voltage across only the one cell, not stack voltage). If any cell exceeds your maximum voltage, charging stops for the entire bank.

Personally, I would set the limit at 2.65-2.7V.

So, when your 'weakest' capacitor hits its maximum voltage, everything stops. No charge current, no balancing current, nothing until you discharge the capacitors.



Sam Beck
www.blueshiftPDX.com
 
Last edited:

Thread Starter

snowdrifter

Joined Aug 13, 2013
43
While that may work well for protecting the capacitors, it wouldn't do anything in the event of a discharge, preventing a low-reading capacitor from flipping polarity. Hmm......



I ran into this on EBay
http://www.ebay.com/itm/2-7v-ultra-...650?pt=LH_DefaultDomain_0&hash=item2a25776b32

I'm not entirely sure what I'm looking at as far as the circuit goes, but I wonder if it would be possible to tweak this and get the voltage set point from 2.7v down to 2.4v. Then if that works, scale it up if need be. Thoughts?
 

wayneh

Joined Sep 9, 2010
17,496
How is that a viable option? I mean, does it work? Doubt it. It may work very, very slowly and make you feel good when you look later and see voltages you like. But aren't you concerned about the instantaneous voltages?

I think you should put an oscilloscope on your cells, one by one, during a charge and discharge cycle, and learn what the voltage profile looks like. Then you can begin to get an idea of how you might affect that voltage profile.
 

Thread Starter

snowdrifter

Joined Aug 13, 2013
43
I'm not as concerned with instantaneous voltage. A couple charge/discharge cycles ends up with all the cells reading roughly the same anyway. The issue I'm running into is after they sit for a couple days. That's when the voltages are different. The resistors seem to do pretty good at keeping them in check. It's just the parasitic draw they have that's the hang up.


I don't own an oscope :(
 

wayneh

Joined Sep 9, 2010
17,496
The issue I'm running into is after they sit for a couple days. That's when the voltages are different.
Some go down while others go up? I don't really understand.

But if your "balancing" currents are actually small compared to the charging and discharging currents (let's say that ≤10A is "small"), then the type of thing we discussed earlier is an option. Voltage references, comparators, MOSFETs, etc. For 10A and less, the needed parts are inexpensive.
 

Thread Starter

snowdrifter

Joined Aug 13, 2013
43
If they sit for a couple days, some discharge faster than others. Extrapolate that over a few weeks/months of use and I start to run into issues with that.

Right now my balancing current is ~55ma. It's really not that much, comparatively speaking
 

BLUESHIFT

Joined Aug 23, 2013
24
OK, my understanding is that there are two - or maybe three? - issues:


1. MOST IMPORTANT PROBLEM:
If charging current is 400,000 mA, and balancing current is 55mA, then your balancing circuit doesn't work at all. This is what's going to kill you capacitors the fastest.

I killed half of the 8x350F caps in an early prototype of mine exactly this way (10A charge current) - the balance circuitry eventually balances the capacitors, but effectively does nothing during the charge phase.

SOLUTION to 1:
wayneh's individual-cell-max-voltage-limit plan deals with this effectively.

THIS IS CRITICAL! Passive balancing works, but only if the balancing current is sufficient, so you would have to have .1ohm resistors for balancing to deal with your 400A.

(You are really talking about 400A, right? Where did you find a 16V*400A power supply?)




2. possibly important (I don't know how important) problem:
As the caps self discharge, an imbalance in charge between capacitors' leakage rates could cause trouble when they are charged back up, as they are starting with different amounts of charge across each.


SOLUTION to 2:

This is what your passive balance circuit deals with (resistors across each cap). You should should size them to dominate leakage current, I believe Maxwell's recommendation is 10x leakage. Leakage is 1.5mA from datasheet, so size resistors for 15mA max current.

Alternatively use an active-balance circuit, such as the one you posted an image of the other day (from engineering shock?) or similar.

Passive balance means continuous power loss across each cell, but simple. Active means power loss only when things are out of whack, but complex (relatively).




this is the part where my answers get fuzzy...

3. I don't think this is a PROBLEM:
Rapidly discharged capacitors could end up at a negative voltage.



SOLUTION to 3.
I don't think this is a problem if you deal with 1 and 2, but i could be wrong.

Specifically, I can't see a way for cells in the same stack to have opposite charges, because 1- is only connected to 2+ except for balance circuitry. If 1- is charged negatively, 2+ HAS TO BE CHARGED POSITIVELY.



That is, unless the balance circuitry and/or leakage screwed it up...



You would need to, as wayneH suggests, measure this and/or model it, I don't have a great idea what happens especially at very large charge current.

I think you could probably work it out on paper - assume different values for a 3-cell bank C1-C3 and try to set charges to get a negative voltage across one cap, or alternatively set two voltages positive and one negative and solve for charge distribution that achieves this...and then imagine how that could happen.


I think solving 1+2 solves your problem...but I could be convinced otherwise.


Sam Beck
http://www.blueshiftPDX.com
 
Last edited:

shortbus

Joined Sep 30, 2009
10,045
Maybe (probably?) I'm missing something in this whole thread. What is this cap bank for? Since when are capacitors, even super capacitors used to "store" energy like a battery,for an extended period of time?

I have always been under the assumption that capacitors were a 'short' term solution for energy in a circuit. To give a 'boost' for a need for more than the normal situation.

To me this thread sounds like maybe a disguised rail gun project.
 

BLUESHIFT

Joined Aug 23, 2013
24
I have always been under the assumption that capacitors were a 'short' term solution for energy in a circuit. To give a 'boost' for a need for more than the normal situation.
Historically you would have been correct, but supercapacitors are improving rapidly, as they have been for many years. Add to that cheap, efficient DC-DC switching regulators and suddenly capacitors are not just an overflow/buffer solution.


Current commercial models store about 1/20 of the energy of Li-ion batteries per Kg, but last about 1000x longer (in terms of time and charge cycles). So, depending on your power & energy requirements, capacitors are a viable option for medium-term energy storage today.

Specifically, supercapacitor energy storage works well today in applications that:
1. charge very rapidly
2. discharge very rapidly
3. have the physical space for bulkier/heavier supercapacitors
4. need to last thousands of cycles without maintenance
5. can handle the roughly 20x higher upfront cost (per J) of supercapacitors
6. can handle higher self-discharge/leakage rates


Laboratory-made supercapacitors have been reported that are comparable to Li-ion specific energy levels - 100 W*h/Kg. If these can be produced commercially (which I believe is likely as nano-scale carbon manipulation takes off) then they beat batteries in almost all criteria.

Plus, you know, flux capacitors are sweet.





Sam Beck
BlueshiftPDX.com
 
Last edited:

#12

Joined Nov 30, 2010
18,224
I don't own an oscope :(
Well, there's a large part of the problem. You're trying to do state of the art design work and don't have a proper meter!
That's like saying, "I'm going to set a new land speed record at the Bonneville salt flats without owning a speedometer (or anything better)."
 

wayneh

Joined Sep 9, 2010
17,496
Yes, at the least I would love to see some data acquisition, such as using a LabJack U3-HV, to profile the situation. If we eliminate the periods of charge and discharge, the necessary data rate doesn't require an oscilloscope. Maybe not even during those high-current events. I know from my own experience that you can log data into Excel every second or so using slow, noob techniques (i.e. what I can do).

One way or another it would be nice to see some data. It's easier to fix a problem you can see.
 
Status
Not open for further replies.
Top