Balancing series stacks of supercapacitors during charging/discharging

[ocset]

Our contractor thinks that the balancing of series stacked supercapacitors during charging is needed because of differences in leakage current. However, we keep telling him that charging of series stacks needs balancing due to the tolerance differences in capacitance from cell to cell. Do you agree? (When the charger goes onto "float charge", thenn the leakage current effects take over)

Also, the problems inherent to charging batteries in parallel with a single charging current source do not exist for paralleled supercapacitors. This is because if one paralleled supercapacitor hogs all the charging current then it will not overheat like a battery could. Do you agree?

The great advantage of supercapacitors in series stacks, that is never mentioned anywhere, is that if series stacked supercapacitors are discharged, then even if they have massively different capacitance values, no single capacitor gets a damaging reverse bias voltage whilst the others have a positive voltage......this makes supercapacitors better than batteries...because a series stack of cells can see one cell go negative voltage whilst the others are positive voltage...destroying that cell.
Why is this great superb advantage of supercapacitors never mentioned..anywhere?
 

[drussell]

The great advantage of supercapacitors in series stacks, that is never mentioned anywhere, is that if series stacked supercapacitors are discharged, then even if they have massively different capacitance values, no single capacitor gets a damaging reverse bias voltage whilst the others have a positive voltage......this makes supercapacitors better than batteries...because a series stack of cells can see one cell go negative voltage whilst the others are positive voltage...destroying that cell.
Why is this great superb advantage of supercapacitors never mentioned..anywhere?

Uhhhh...

Why would capacitors in series behave any differently than series cells in a battery?  If one discharges more than the others in the series string, it most certainly will see a reverse "charging" current.  You would need a reverse protection diode across each capacitor to prevent this if it is a possibility in your application.

[ovnr]

Series: You need balancing due to a combination of factors; it's not a single thing. If all the leakage currents were the same while the capacitance differed, they would self-balance when full. Vice versa, if all the capacitances were the same and the leakage currents differed, they would not self-balance. Just add appropriate balancing resistors (or active balancers), and leave some voltage headroom.

Parallel: Sure, but with limits. You can damage a supercap with too high charging current.

Discharge: It's not mentioned anywhere because it's wrong. You certainly can make it go negative. With minor (+- 20%) capacitance differences and modest currents/string lengths, it should be largely fine. If you're pushing it up against the limits, you will want parallel diodes for each cap.

[ocset]

Discharge: It's not mentioned anywhere because it's wrong. You certainly can make it go negative. With minor (+- 20%) capacitance differences and modest currents/string lengths, it should be largely fine. If you're pushing it up against the limits, you will want parallel diodes for each cap.

Thanks, but you really cannot get one capacitor in a series stack to go negative in voltage  whilst the others are positive in voltage when you discharge a series stack.
The attached ltspice simulation prooves this.
ltspice is a free download.

[ocset]

Here's another point....
Take three 1F capacitors each charged to 1V...that is, they contain 1 Coulomb..each.
Now put them in a triple series stack....
How much charge is in the triple series stack capacitor?
Answer is 1 Coulomb.
Do you agree.

[mikerj]

Discharge: It's not mentioned anywhere because it's wrong. You certainly can make it go negative. With minor (+- 20%) capacitance differences and modest currents/string lengths, it should be largely fine. If you're pushing it up against the limits, you will want parallel diodes for each cap.

Thanks, but you really cannot get one capacitor in a series stack to go negative in voltage  whilst the others are positive in voltage when you discharge a series stack.
The attached ltspice simulation prooves this.
ltspice is a free download.

Slight fail in your modelling; you aren't balancing the capacitors during charging so each capacitors final voltage is inversely proportional to its capacitance.  If you modify the circuit to charge each cap up to the same final voltage you will see that the smaller value caps definitely do get a reverse polarity during discharge.

This should be obvious anyway from basic physics.  The caps are in series so each one has the same discharge current, i.e. charge is being removed at the same rate (Q=It).  A capacitor that starts with a small amount of charge (i.e. smaller capacitance in this case) will clearly have all it's charge depleted more quickly than one with a greater amount of charge.

[rs20]

Our contractor thinks that the balancing of series stacked supercapacitors during charging is needed because of differences in leakage current. However, we keep telling him that charging of series stacks needs balancing due to the tolerance differences in capacitance from cell to cell. Do you agree? (When the charger goes onto "float charge", thenn the leakage current effects take over)

One and/or both of you seems to be making the fundamental logical fail here of assuming that any given thing is done for precisely one reason. The "my proof is correct and looks different to yours therefore your proof must be incorrect" fallacy.

As already pointed out, unequal leakages and unequal capacitance values can both cause damage, independently of each other. So you're probably both right but having an argument because you can't see that.

[Berni]

Supercap stacks DO need balancing.

While the huge capacitance tolerances you usually see on them(offten worse then +/- 20%) is taken care of that the bigger capacitor changes up to less voltage so that everyone contains the same amount of charge. This makes it work reasonably well if you charge them up quickly and then soon after discharge them completely.

But they also have internal leakage currents. So as the capacitors sit charged for days or weeks some will self discharge faster and that unbalances them. This could then cause the least leaky cap to get even more voltage upon the next charging or on a full discarge the most leaky cap would go negative.

Supercaps can be pretty fussy in the long term about there voltage, driving them too close to there rated voltage can cause failure a year or two down the road. Its probably perfectly sufficient to add series resistors over the caps to even out the lekagaes, but there are also supercap charger chips that can do active balancing.

[ovnr]

Discharge: It's not mentioned anywhere because it's wrong. You certainly can make it go negative. With minor (+- 20%) capacitance differences and modest currents/string lengths, it should be largely fine. If you're pushing it up against the limits, you will want parallel diodes for each cap.

Thanks, but you really cannot get one capacitor in a series stack to go negative in voltage  whilst the others are positive in voltage when you discharge a series stack.
The attached ltspice simulation prooves this.
ltspice is a free download.

I made my own - schematic: https://nxr.no/p4es, result: https://nxr.no/g1yw

As you can see, the blue and turquoise lines go below 0v. Blue due to leakage causing the voltage to be uneven in a steady state; turquoise due to differing capacitance, which will be the main issue as you can't balance it out.

Here's another point....
Take three 1F capacitors each charged to 1V...that is, they contain 1 Coulomb..each.
Now put them in a triple series stack....
How much charge is in the triple series stack capacitor?
Answer is 1 Coulomb.
Do you agree.

Yes, in an ideal world where the capacitors are identical. Coulombs are units of current, not units of energy. This is just like putting batteries in series; three series 2000 mAh AA batteries will still be 2000 mAh, but you will have 9 Wh instead of 3 Wh.

[David Hess]

Supercaps can be pretty fussy in the long term about there voltage, driving them too close to there rated voltage can cause failure a year or two down the road. Its probably perfectly sufficient to add series resistors over the caps to even out the leakages, but there are also supercap charger chips that can do active balancing.

As I recall, increasing the charge voltage by 0.2 volts decreases the operating life by 10 times.

[ocset]

Thanks i quite agree that supercaps need balancing to even out the leakage current situation.
And if a series stack of capacitors, which has been leakage current balanced, is suddenly discharged, then all the capacitors will go through the zero volts at the same time, which i believe you agree with?......non-idealities accepted

[ovnr]

And if a series stack of capacitors, which has been leakage current balanced, is suddenly discharged, then all the capacitors will go through the zero volts at the same time, which i believe you agree with?......non-idealities accepted

If the capacitors have the exact same capacitance, then yes, they will reach zero volts simultaneously if their original, steady-state voltage is the same (as it would be if they were balanced properly).

Since you're not going to find caps that have the exact same capacitance, no, they will not all magically reach zero volts at the same time.

[rs20]

And if a series stack of capacitors, which has been leakage current balanced, is suddenly discharged, then all the capacitors will go through the zero volts at the same time, which i believe you agree with?......non-idealities accepted

What exactly do you mean by "leakage current balanced"? If you mean "equal equivalent parallel resistances", then yeah, that translates into the steady initial state where every capacitor is at the same voltage, and the answer given above by ovnr is correct under that assumption.

However, if I am to interpret the phrase "leakage current balanced" literally, that's a pointless truism because Kirchoff's Current Law guarantees that the currents are all equal through any series string of components.

This may seem pedantic, but I'd actually decline to answer your question until you clarify what you mean.

[ocset]

Thanks, my reply#3 shows the sim which prooves that if there is no leakage , then any series stack of caps, whether or not equal in capacitance, when suddenly discharged, will reach zero volts at the same time as each other.

[ovnr]

Thanks, my reply#3 shows the sim which prooves that if there is no leakage , then any series stack of caps, whether or not equal in capacitance, when suddenly discharged, will reach zero volts at the same time as each other.

Have you actually looked at your sim? Noticed how the 10F cap has a voltage of 200 mV, compared to the 1F cap which is at a bit over 2V?

Yes, if there is no leakage, you have a capacitive voltage divider that works down to DC, and the caps will reach zero volts simultaneously. However, in the real world, there is no such thing as "no leakage", no matter what you do. And if you balance out the leakage current so that each capacitor has roughly the same voltage across it while the capacitance differs, you will end up with caps going negative.


Just to clarify: Are you trying to argue about how things work in the magic land of ideal components and butterflies, or in the real world? There are important differences between the two, like "being physically possible or not".

[mikerj]

Thanks, my reply#3 shows the sim which prooves that if there is no leakage , then any series stack of caps, whether or not equal in capacitance, when suddenly discharged, will reach zero volts at the same time as each other.

Guessing you couldn't be bothered to read my reply which shows why your simulation is flawed?

[ocset]

Slight fail in your modelling; you aren't balancing the capacitors during charging so each capacitors final voltage is inversely proportional to its capacitance.  If you modify the circuit to charge each cap up to the same final voltage you will see that the smaller value caps definitely do get a reverse polarity during discharge.

Thanks, the amount fo balancing current to do this is very small.
We wouldnt modify the charger so that different capacitances in a series stack got charged up to the same voltage as each other.
I am not sure why this would be a wanted thing(?)

[IanB]

We wouldnt modify the charger so that different capacitances in a series stack got charged up to the same voltage as each other.
I am not sure why this would be a wanted thing(?)

Because you would wish to regulate the voltage on each capacitor so the voltage doesn't go over the maximum permitted. Let's say you have 2 V supercaps and put 5 of them in series. If any cap goes above 2 V it will be damaged. If you charge from a 10 V supply with balancing, you ensure the maximum voltage on any capacitor is 10/5 = 2 V. If you don't apply the balancing then you could have some capacitors at more than 2 V with other capacitors less than 2 V. Without balancing you have no control over this.

[jbb]

Also note leakage currents and capacitances will change with time, age, cycle count etc.

There really is no escaping the need for some balancing somewhere.