NiMH battery paralleling

I would be worried about a cell failing short-circuit. Isn't that the normal failure mode? Then the other strings will discharge themselves through the string with the faulty cell.

Leaving aside right or wrong question, it is pretty typical solution for battery packs to have cells hardwired in parallel for both charging and discharging. First, the cells are connected in the parallel one to another, and then the paralleled pairs are connected into one string.

Vladimir Vassilevsky

DSP and Mixed Signal Design Consultant

On a sunny day (Wed, 18 Jul 2007 10:13:26 -0400) it happened Spehro Pefhany wrote in :

I think the main thing to watch out for is when one chain is empty, that the max charge current is not exceeded. So when the others charge the empty ones. Any chance of reverse connection? Diodes would make me feel safer. Or something with MOSFET swicthes perhaps to reduce losses?

Hi, John:-

With NiCd cells shorting is the usual failure mode, I am not sure about NiMH. The end-of-life is defined as a decrease in storage capacity under specified conditions. I don't think I've seen one that has failed completely (unlike NiCds), in my own use of them for digicams etc.

That's my idea behind putting "or" diodes in there, so if the voltage drops on one string due to a dead cell (open or short) then the survivors should share the load (more or less).

But even without diodes, in case of a catastrophic (maybe a chain reaction failure due to overcharging from other cells) the internal fuse & vent in the cells chould go (but before that, the external fuse should isolate the problem string and break the current). I suppose in the case of a polyfuse it would cut in and out with some rep rate, which isn't good.

Best regards, Spehro Pefhany

Hi, Jan:-

They'll be charged in individual strings (one charger per string), so at least that's not an issue.

Shouldn't be, but putting diodes in the right location could act as insurance. It wouldn't be catastrophic if string was reversed.

Yes, the old bass-ackwards MOSFET trick. It would add a handful of parts since the string voltage is more than the gate can stand and there's a lot of strings. For lower voltages it would be nice.

Best regards, Spehro Pefhany

You can get away with paralleling many types of rechargeable cells for the simple reason there is a "dead band" before charging starts. For instance a flat lead acid cell at 1.9 volts will not start to charge until the terminal voltage is raised to around 1.95 volts. 1.92 volts or so will do nothing.

This doesn't happen with NiMH (or Nicads). The tiniest increase in terminal voltage will result in a charghe current.

This in itself is, of course, not an issue but it becomes one when it is realised that the terminal voltage of different NiMH cells varies slightly for the same state of charge.

Two cells in parallel will never have the same open circuit voltage for the same state of charge so when connected in parallel the one with the higher voltage will discharge slightly into the lower one.

During the use of the battery pack this means that the cell with the lower terminal voltage (yes lower, not higher - because it will be in a higher state of charge for the same voltage and therefore able to produce more current) will be worked harder during discharge and charge and result in much less than double the capacity of a single cell on its own.

That's why they should not be connected in parallel.

The diodes are a must.

Hi, Gibbo:-

Sounds reasonable. Done!

Thanks.

Best regards, Spehro Pefhany

If a single series string can handle the load, I'm assuming that what you are after is the higher capacity, not increased peak current due to the parallel strings.

I think your diode/fuse solution will work OK. There are some multiple battery controllers (search Maxim, I think) that switch between two or more battery packs. They allow for hot swapping single packs, control charging, and provide some battery status signaling, IIRC. Needless to say, there will be additional $$ involved.

That could be a problem. But the diodes will fix that. You might not even have to polyfuse the individual strings if you use diodes.

The isolation diodes alone will allow hot swapping.

Looks like combining NiMH in any way is an art form, and you're taking a chance of early failure even with the series string, especially if the customer has access to it. Google turns up a few papers describing UPS backup systems using hundreds of NiMH arranged into parallel banks but they don't go into any detail regarding matching and monitoring methods.