PDA

View Full Version : Cell temperature sensor



DaveAK
01 January 2011, 1150
The TI Eval board that I'm testing cells with comes with individual cell temperature sensors that consist of a very thin PCB, a 0603 1000pF capacitor and a SOT-23 dual diode, (one of these (http://www.fairchildsemi.com/ds/MM/MMBD4148SE.pdf), SE configuration, pin 3 open.) The circuit is simply two diodes in series, with the capacitor in parallel. One of the PCB traces is made in to a rectangular pad.

So my guess is that the pad creates a resistance which will vary as the temperature changes. Does that sound about right? But what does the double diode do? I've never seen that before.

What I want to do with my LiPo cells is to make a 4SnP battery. Each 4S part would have it's own sensor, so you'd end up with n sensors. If you understand how these sensors work maybe you could answer this question for me. :) Could I connect n number of these sensors together in either series or parallel and successfully be alerted should one of the 4S packs rise more in temperature than the others? Would serial or parallel be better? (Parallel would be prefered.)

I realize that I might not be able to indicate secifically which 4S group was overheating, but I can live with that if I can at least indicate a problem. If I knew the value of n I could measure each indiviually to indicate the location of the problem, but these are going to be modular packs so n will vary.

magicsmoke
01 January 2011, 1239
Hi Dave, It's almost certainly the diodes themselves that are acting as the sensor.
The conduction of a diode varies very precisely with temperature.
Different techniques involve measuring the voltage across the diode with a known current or measuring the current with a known voltage.
From diode type to diode type there may be huge differences in performance, from one to another of the same type they're usually pretty close.
Anyway, the most oft used approach is to pair either of the above techniques with a micro controller which will compare the results with a previously 'mapped' table.

Usually better to measure the current - better noise performance.
Capacitor to filter noise (some diodes could be used as a noise source!).

Happy New Year
Rob

DaveAK
01 January 2011, 1342
Thanks Rob! I never knew that about diodes. And these are of course hooked to a microcontroller, into which you can load various profiles depending on type of cell. That would probably include a current/temperature mapping table.

So with all of that in mind I'm not sure how I'm going to get this to work. Maybe by monitoring each parallel string, by configuring the mapping table for n number of cells in some way.

DaveAK
01 January 2011, 1847
Any reason it's a dual diode?

magicsmoke
01 January 2011, 2012
Any reason it's a dual diode?

simply to double the resolution. If one diode increased by say 10mV over temerature then two would be 20mV.



So with all of that in mind I'm not sure how I'm going to get this to work. Maybe by monitoring each parallel string, by configuring the mapping table for n number of cells in some way.

The map will be for a particular diode(s) and is irrespective of the number of cells.
Some uC have current sensors built in, either multi channel or multiplexed.
You could also do any number of discrete solutions but to be honest, if you've got the use of a uC then I'd just use one of the digital solutions such as ..

http://focus.ti.com/docs/prod/folders/print/tmp101.html

I didn't pick this one for any reason other than just as an idea of what's out there.

Rob

DaveAK
01 January 2011, 2157
Unless I can find anything similar then I'll go with the Maxim 1 wire solution as it allows for an unlimited number of devices on the bus.

magicsmoke
02 January 2011, 1207
quick note.
In a vid that billmi linked to, http://www.ri.cmu.edu/video_view.htm...60&menu_id=387 the presenter mentions the irony of the temp sensor, that he stuck on a cell, failed and burned the cell. He recommended a fuse in the supply lines to the sensors.

DaveAK
02 January 2011, 1213
quick note.
In a vid that billmi linked to, http://www.ri.cmu.edu/video_view.htm...60&menu_id=387 the presenter mentions the irony of the temp sensor, that he stuck on a cell, failed and burned the cell. He recommended a fuse in the supply lines to the sensors.
Thanks. That reminds me to watch that vid again. I only got a few minutes of it last time I tried. I haven't decided on the actual configuration yet but I do want to incorporate appropriate fuses, would never have thought to put it on the temp sensor though! Still figuring everything out, but hope to mock something up soon.