Charging ultracapacitors

[mtweeman]

Hello everybody!

Lately, I've became interested in ultracapacitor charging. I've been doing some electronic stuff before, but this is completely new to me.

Can I get any advice which way should I go in that matter?

I've been doing some research myself, but without any results which I'm satisfied of. So far I know that there is a need of having constant current to charge ultracapacitors, so perhaps I should have current stabilizer. However, there are many of them available and I'm not quite sure which one would be the best in this particular case.

Any help would be appreciated.

[Zero999]

It needs to be switched mode, otherwise you lose over half of the energy.

More information is needed.

[German_EE]

An ultracapacitor presents almost a short circuit to any voltage source charging it, especially if the capacitor is close to 0V. You therefore need to charge it using a constant current circuit so that the voltage source can cope with the load.

[DanielS]

Simple enough: you put together a regular buck regulator with per-cycle current limiting. This way, it will act like a current source until it reaches its operating voltage and become a voltage regulator at that point.

Even ancient PWM controllers like the TL494 can do this.

[mtweeman]

Thank you for quick responses!

Perhaps I should be more precise from the very beginning.
What I intend to do is to build a high power (around 200W) system. The circuit has to load 58F, 16V ultracapacitor bank in about 40secs.

I'm looking for a solution how to do that. Some examples, ideas (block diagrams) would be helpful. What I expect at the end is circuit's reliability mostly (I'm open for suggestions if I should add something to that list). If any solution that can be simply bought come to your mind, let me know - can be inspiring, too.

Hero999, what information do you need exactly? I will be happy to provide it.

English is not my native language, so please, use full names of devices, etc., so I can easily understand what you are talking about.

[DanielS]

Well, Q = C*V and if you want to charge in 40 seconds, Q/40 gives you a 23.2A constant current.

So you need a DC-DC converter able to provide 16V output (if the bank itself is rated 16V, you might want to aim slightly lower like 14-15V) with a continuous 23A short circuit output current limit.

[German_EE]

58F @ 16V is a lot of power, be careful.

[eneuro]

0.5*C*V^2 gives around 7.4kJ, so desent power a few kW in 1 second 
Anyway, probably you miss something-is it one bulky 16V ultracapacitor ? 
or rather many 2.5V volt in series?
This means that you probably nee to balance voltage on each cap in series to keep it below 2.5V and just connecting DC-DC can be not the safest way to charge those caps in series close to its nominal volage 

Do you have 16V ultracap or string of 6-7 2.5V in series? 

[mtweeman]

I mean module like this: http://www.mouser.com/ds/2/257/Maxwell_16V%20Small%20Cell%20Module_DS_1015371-6-341171.pdf
so eneuro is right on that one. That would be series of ultracapacitors.

You said that DC-DC can not be the safest way for charging. What options do I have? Which one do you recommend? Thanks in advance.

[eneuro]

That would be series of ultracapacitors.

For sure it is 6   caps each 350F, so 350F/6~58F 

I haven't got in hands this thing, but in its datasheet they wrote whatever do they mean:

Cell Voltage Management - Passive

'
Since in specs is 16V maybe they protected those cells somehow and this passive cell management does a trick?
Anyway is it possible access indyvidual cells?
Maybe someone else had this thing in hands 

[DanielS]

I haven't got in hands this thing, but in its datasheet they wrote whatever do they mean:

Cell Voltage Management - Passive

'
Since in specs is 16V maybe they protected those cells somehow and this passive cell management does a trick?

The spec sheet says they balance voltage using a resistive divider.

They likely match the cells before assembling the module to make sure their (dis-)charge characteristics are reasonably well matched so the resistive network does not need to do much equalizing beyond prevent one cell from rising significantly above the others due to float-charging.

[mzzj]

Thank you for quick responses!

Perhaps I should be more precise from the very beginning.
What I intend to do is to build a high power (around 200W) system. The circuit has to load 58F, 16V ultracapacitor bank in about 40secs.

I'm looking for a solution how to do that. Some examples, ideas (block diagrams) would be helpful. What I expect at the end is circuit's reliability mostly (I'm open for suggestions if I should add something to that list). If any solution that can be simply bought come to your mind, let me know - can be inspiring, too.

Hero999, what information do you need exactly? I will be happy to provide it.

English is not my native language, so please, use full names of devices, etc., so I can easily understand what you are talking about.

200 watts average if you charge with constant power*. Or constant current 23Amps *16v ~400W peak power

*not really feasible, you would need to charge with 200Amps @1 volt and 400Amps@0.5 volts etc.

[DanielS]

200 watts average if you charge with constant power*. Or constant current 23Amps *16v ~400W peak power

*not really feasible, you would need to charge with 200Amps @1 volt and 400Amps@0.5 volts etc.

It is 200W average under constant current too since the voltage, and incidentally power, increases linearly with charge so the average power is half the peak

Inconvenient numbers aside for constant power, constant current is simply easier to work with.

BTW, the capacitors themselves are only rated for 19A RMS if you allow a 40C temperature rise (or 12A for 15C), so even constant-current at 23A for 40 seconds might warm things up quite a bit. The charge time might need to get bumped to 50+ seconds at 19A or less to keep temperature rise in check and get better service life.

[eneuro]

The spec sheet says they balance voltage using a resistive divider.

So, it will work up to some charge current limit and above this, those bloody resistors in voltage divider simply will not be able dissipate fast enougth additional charge while there is no active overvoltage protection, I guess   

They can select caps, but... aging will degrade capacity in some of those cpas in series and without overvoltage protection at higher charge currents simply resistive divider might not work no more 

But ok, maybe this passive cell voltage management works for higher charging current too, while they says about DC-DC step down chargers with current limit there:
http://www.maxwell.com/images/documents/an-010_charging_ultracaps.pdf

Anyway, I do not like it at all and prefere overvoltage active detection/protection  on each cell cap in series.

BTW: Maybe manufacturer wants customers buy those ultracaps packed in series more often, so it says nothing about other methods of charging caps in series-I smell  there, but without details what kind of resistors is used for this passive cell voltage management do not rant this, but no chance I wouldn't buy such module-rather built one from separate caps with added small PCBs for overvoltage detection/protection to let know charger if any cell is overcharged to stop charging and let switch on power resistor on this cell to discharge to save level 

[HighVoltage]

I am pretty sure that they use a cell balancing with resistors and diodes in this Maxwell 6-pack
But this is not very efficient.
In some of the larger Maxwell cells they use their patented active cell balancing circuit, which is much more efficient.

[eneuro]

In some of the larger Maxwell cells they use their patented active cell balancing circuit, which is much more efficient.

Probably you mean circuit mentioned in this document Maxwell Active Cell Voltage Management Electronics ? 
Looks like nothing fancy and I used to use TL431I to detect overvoltage condition and triger cell discharging, so maybe their patent pending is their specyfic circuit to perform similar task quoted from document above:

Each cell voltage is individually monitored against a nominal reference voltage. When the reference cell voltage is exceeded the voltage management circuit actively works toreduce the voltage of the cell to below the trip level.

Tried find this Maxwell patent (pending?) by ended so far with this patent: Battery cell-balancing method and apparatus

I'm really interested in using such a few 350F 2.5V ulracaps in series  in recent small tennis robot project while do not want to mess with flamable Li-on, so probably will use full bridge SMPS with 6 outputs and mentioned overvoltage detection/protection PCBs I've alfready done for 12V lead-acid battery protection and rather piece of cake to detect overvoltage and stop charging given cell as well as discharge if needed-no rocket science and no Maxwell patents needed just  and  to avoid     

[mtweeman]

I've been using many other forums over the years on different matters, but I'm really surprised and thankful that so many of you decided to involve into my topic. Thanks for that!

Much useful information was given and helpful materials, too.
I think that this conversation leads into "build your own capacitors bank" direction. I was also considering it. I just thouught I should give a try to whole bank sold by supplier first. Could we focus on bought separately capacitors then and building my own circuit to charge them? Any tips, thoughts - more than appreciated. If I'm going to charge one cap after another then a charging time will increase, because I have to charge 6 caps "separately". Loading them in parallel is not making it easier, because I'll need higher current. Any clues how to do that without losing time? Charging time is one of the most important things for me on whole case.

[eneuro]

I think that this conversation leads into "build your own capacitors bank" direction.

This passive cell voltage management in this product you linked might work quite well, since when read more in linked patent description where they introduce to passive battery cell voltage management-there is at least resistor with transistor and probably some kind of diode or zener to or P-channel  mosfet with Vgs 2.5V to detect cell overvoltage condition and disipate heat in resistor-so basicly this what I've already done and rather call it active while there are not only resistors on caps but some kind of control.
Unfortunatelly, I had no chance to make teardown of this 58F 16V ultracapacitors pack and while I have no idea what level of heat can be disipated at which rates, rather looking for custom solution where when we detect during charging overvoltage on one cap more often (first) than on other while we are close to its nominal 16V than we can simply let know via optocoupler our charger that this cell is overcharged and for example stop charging to avoid overcharging this probably worst cell, so we'll end up with battery not charged to full nominal voltage (eg. 15.5V) but we know which cell is a problem and since we have custom made caps in series we can identify this worst cell and replace with brand new without need to buy another one fully assembled 6 cells product you linked in your posts 

So, basicly it was rather methodology confusion-I was thinking I have active battery/cap management when monitoring each cell and discharge (eg. using power resistor), but it looks like this aproach is called rather pasive, while active is considered with a lot of additional wiring and switches to allow redirect current between cells and balance its voltage without disipating at heat which decreases efficiency in the case of passive voltage cell management.

For me passive is fine since it is simple to implement and can put in custom solution as huge power resistors as needed and still detect which cell is out of specs and how much.
Since I do not know what expect in Maxwell product you linked -I did't found how this passive management is implemented in hardware, I can't put it into circuit simulator and see how it will perform in my application-I'd rather develop own solution, especially while additionally need very different package-ultracaps need to be put in series but in classic horizontal (not vertical order) position to fit into custom 3D printed housing