Thermistor control for ni-cads

[zeusmorg]

 I am building some ni-cad battery packs, and I wish to add thermistor control to this scheme.

 I've come up with a solution that uses one thermistor, however I want to use one in each individual pack, and have the charger cut back if only one exceeds temperature,

so does anyone have any thoughts on how I can accomplish this? I'll be using 10k NTC thermistors, the battery pack will be a nominal 48v system. The charger system will be rated @ C/10.

[BigBreaker]

Thermistors in parallel, each on a different ni-cad pack,  will have a resistance as a group that is dominated by the hottest one, IE the one with the lowest resistance.  It's a soft min type of transfer function though...  you may want something different.

The resistance of paralleled resistors is the product of the individual resistances divided by their sum.  If you do some test examples you will see what I mean about the lowest resistance dominating.

[Bruce S]

Zeusmorg;

   I'm thinking that if you're using a charger that is built for the total of the packs to keep them all at or near balanced. It would be better to have a thermistor for each pack in series with the others. This way if one overheats, then the whole charging stops. This will keep them entire packs from a possible cascade effect of over current to the rest of the packs. Once that thermistor cools then the charger could restart.

If you're looking to control the charge to individual banks, then you could use a gang of OpAmps as Feed-back control for each individual bank. This could be a better route to go since you could still charge to other banks while allowing the errant one to cool.

OR the thermistor could be used with say a LM317 on the adj side to automatically limit voltage/current to the individual banks.

Just a few hang-over thoughts.

Keep Smiling;

Bruce S

 

[zeusmorg]

I'm planning on charging the pack whole @48v..

[laskey]

The problem is of multiples.  If you use a series or parallel circuit of thermistor you're okay with 2 or 3 or 5 thermistors, but if you start having too many then the circuit's ability to detect the change in the state of the resistance becomes muddy.  Say you have 10 10k thermistors, say at temp the thermistor goes to 1k. I you have 10 in series the total drop is 9k out of 100k which won't be all that easy to detect.  If they are in parallel then the circuit steady state resistance is close to 1k already, drop one to 1k and you get 523 ohms.  Easy to detect, but the more thermistors you add the closer the steady state and the drop states become.  So my thought is on the small scale use the parallel circuit, but if you've got tonnes of packs then switch to some kind of digital circuit like cascaded NAND gates, that takes many, many, many thermistors down to one output, and any one of them changing state could turn a relay, or power transistor, or FET on and off.

Hope that mess makes sense,

Chris

[zeusmorg]

 Not really, I'm electronically challenged.