Temperature sensing for multiple cells

George,

We do not have a configuration to multiplex the thermistors to a single input. The bq40z50 supports four thermistors. If the gauge that you are using does not support four, then you could just use one and place it on one of the four cells.

Tom

Tom,

Thanks a bunch! I never thought of using the bq40z50, but it looks like it will work fine.

Tom,

I just got blind-sided by our customer in that they now want to go to a 5-cell series stack, thus making the bq40z50 unusable. Now I need a monitor that can support 5 thermistor inputs. Note that I cannot use your previous suggestion of using only 1 thermistor and attaching it to one of the cells, as the individual cells are required to have thermal isolation between them and I need to know if any cell is in danger of going into thermal runaway.

George,
The bq78350 and bq76920 chipset will be our best option for 5S, but it only supports one thermistor. You could place it on one of the cells.

Tom

Tom,
I have already considered the bq78350/bq76920 pairing with the thermistor on the center cell of the stack. I (and more importantly my management) am concerned that this will not provide accurate measurement of the stack temperature since the individual cells are thermally isolated, at least side-to-side, from one another. Will thermal conduction from the cell ends through the air provide sufficient heat transfer to average out the individual cell temperatures while still preventing a cascade effect from thermal runaway of any one cell?

George

One thermistor should provide enough information to make gauging adjustments for temperature changes. It will limit that device's ability to detect safety thermal events that may occur in one of the isolated cells.

What is the measurement time for the thermistor input? I was thinking about using a slow-speed clock and a ring counter to sequentially connect the 5 thermistor inputs to the TH pin of the bq76920 and have it read the thermistors sequentially. I thought I saw somewhere in the literature the rep rate of the TH measurement was in the neighborhood of once every 2 seconds. I realize that this would mean the temperature for the worst case cell would only be updated once every 10 seconds, but it would give a little more comfort to the customer.

Does this make sense, or am I nuts (I realize the two may not be mutually exclusive, but I had to ask)?

George,
The bq76920 should measure temperature every second. This device is supported by another group, so I will move the message to that forum for a better answer.
Tom

George,

This seems like it should work. The bq76920 is really just using the impedance on the TS1 pin as the bottom-half of a resistor divider and reading this voltage; the temp on bq76920 is actually measured on a 2-second interval. You may want to multiplex every 2.5-seconds or so in order to ensure that you'll always get a reading from a particular thermistor (as due to clock frequency variation you might miss or double-read a thermistor if multiplex exactly 2-seconds).

Allen

Allen,

One more question:  What is the timing requirement for a thermistor reading, and how fast does the bq76920 respond?  For example, what if, in my multiplexed thermistor setup, one thermistor is sensing significantly higher temperature than the other 4?  Will the AFE immediately respond to the high thermistor and take appropriate action (shutdown) and then try to restart if the next reading is within limits?  It would not be as much of a problem if the other thermistors were reading temperatures within the hysteresis band of the temperature input (i.e. trip temp = 45°C, resume temp = 40°C, one thermistor is indicating 48°C and the other 4 are between 40 and 45), since I would expect that the sensor would not be able to reset until it gets a reading below the resume point.

Is this something I should worry about?

George

Alan (or anyone else who may have input),

I have been again asked by my management regarding the scheme to multiplex the various thermistors in order to gain sequential readings of the pack temperature.  My question now is regarding the response of the pack to an overtemp seen only on one of the 5 cells.  Will the pack shut down on the first indication of an overtemp and then try and restart if the next cell in the sequence is not in an overtemp condition?  I realize that in the event of a single cell reaching the OT set temperature, the other cells in the pack are likely be above the OT clear (depending on the amount of hyseteresis I build in), but for sake of assumption I need to know.

Does the controller react instantaneously to an overtemp indication, or does it average multiple (2-second) readings before issuing a shutdown command?

George

George,

The controller will not react instantaneously to an overtemp indication. You have both a threshold setting and a delay setting for thermistors. The temperature is sensed during a quick A/D conversion once per second. If the threshold is exceeded, it will continue checking each second to see if the high temperature remains. If the high temperature is still present after your programmable delay time in seconds, then the protection will activate. Otherwise, if the temperature is no longer above the threshold, no protection is activated. Also the delay timer is reset whenever the temperature is found to be below the threshold.

Doug