Thermopile SMBUS vs I2C

[Mosaic]

Hi all:
I want to build a multi thermopile remote temp sensor to track Lead acid battery cell temperatures simultaneously, during charge/ discharge cycles. (6 cells per batt).

The range of interest is about 25C thru 50C which governs temp compensation for Charge voltage.

I am considering this sensor:
**broken link removed**

Also would SMBUS work with the I2C module in a PIC, wiki kinda says it should.

Consider the 12F1822 or 16F1823 as the PIC.
https://www.microchip.com/wwwproducts/Devices.aspx?product=PIC12F1822

EDIT:
Actually this device,
https://www.melexis.com/Asset/IR-sensor-thermometer-MLX90615-Datasheet-DownloadLink-5477.aspx

Seems the better solution, also uses SMBUS. However it offers PWM which might make for a simple LP filter and a single pin PIC ADC read as the method to grab the temp.

 

[Diver300]

As far as I can tell, the SMBus is very similar to the I2C bus, and you can almost certainly use them together. The SMBus has an interface timeout, which I2C doesn't. That is actually useful, as if you want to restart communications, all you have to do is to wait 0.1 seconds or longer.

With any bus communication system, you will probably need an oscilloscope to get it working.

The results will be better than using something with a PWM output, which needs a lot more wires, and all the problems with analog readings. A bus system is better for multiple readings, as you can parallel all the sensors, use very few wires, and not have to worry about signal quality, because all the analog signal processing is done inside the sensor.

Having said that, the Melexis device does have the advantage that you can alter the min and max temperature for the PWM range, so you can make it much more sensitive over the range you are interested in. The downside of the Melexis device is that you have to programme it, either for PWM and range, or for different addresses if you are using the SMBus. You need to use the SMBus to programme it to work with PWM, so you've got to write the code to get that to happen, whether you use it once or millions of times.

The TI device has the advantage that the address is set by wiring selector pins to ground, positive, SDA or SCL so you can have all 6 sensors on one bus without having to programme them first.

I hadn't come across the infra-red temperature sensors before. They are quite neat, but they are not as accurate as contact thermometers. There are lots of digital thermometer ICs. You could look at the Dallas DS18B20 which only needs one wire plus ground. Also each one has a unique address. Normally, you have some way of reading the address before assembly, so that you know which sensor is which, but in this application it may not matter if you don't know which cell is the hottest, just that one is too hot.

 

[Mosaic]

Thanks for the effort into the answer. I feel the Melexis part is more suitable since it also offers the option to select a field of view. If I were to go with contact thermometers, I would prob just go with NTC thermistors and adc sample them as a low cost solution.

 

[Mosaic]

I'm going have a crack at upgrading the field of view to give a greater range. The spec sheet on the Melexis device gives the dimensions of the tubular mirror extension required.

I'm going to try polished brass tubing cut down to size. (2cm x 11/32" dia). If that works, it cuts the cost of a 5 deg. FOV sensor to 1/4 the retail value and makes isolating cell temperatures a bit simpler.

Ideally I'd want to use stainless steel, but locating 11/32" diameter thinwalled tube seems impossible.