As mentioned by Chuck and his links - the beauty of these sensors is
that you can connect quite a few of them together to 1-wire bus and
address them by their unique ID.
DS18B20 are really much better choice than USB thermometers when you
need to measure temperature at more than one/two places.
a) you need USB port for every USB thermometer - so you end up building
chains of USB hubs and being limited by length and cost of USB cables
b) 1-wire interface can connect many DS18B20 thermometers over long
distance and the cable is dirt cheap/thin.
c) DS18B20 are about order of magnitude more precise than USB Tempers
Chuck mentions many useful links which I will not repeat.
In short, you do not need to develop anything for this mature problem.
Just hook the stuff together on RaspberryPi, use the 1-wire kernel
modules (w1-gpio w1-therm) and read the values from /sys/bus/w1/devices
- you will see one directory for each device.
Have a look at this web page about the project scope/difficulty:
https://pimylifeup.com/raspberry-pi-temperature-sensor/
Here is datasheet for the sensor itself:
https://datasheets.maximintegrated.com/en/ds/DS18S20.pdf
I personally used the USB thermometers to debug some cooling zone
balancing problems in my data center. It is trivial to just plug the
Tempers to front panel USB ports on a servers in different cooling
zones/isles, read them by cron and push the values to a DB for simple
web CGI interface. I was OK with +-2.5C precision for trending and zone
balancing modifications. The thermometers were also great for remote
alarms when something went completely wrong - the cooling went down
once per 2-3 years kind of events caused by people shutting PCW,
clogged filters or ..... All that because the cooling system
monitoring/alarms/control were linked to some closed source systems at
fire and disaster response and not me ...
Hope it helps, Tomas
On Tue, 2017-01-17 at 11:54 -0800, Michael C. Robinson wrote:
> > I would highly recommend the Maxim DS18B20 as mentioned in the link
> > by
> > Chuck if you need something more accurate. They cost under $2 in
> > the
> > waterproof version.
> > Hat-down to the analog designers @ Maxim designing them so precise
> > within this wide temperature and voltage range (±0.5°C Accuracy
> > from
> > -10°C to +85°C @ Vdd=3-5.5V) <-- the probe + reference + A/D
> > convertor
> > are at the same hot/cold temperature and at variable voltage for
> > about
> > $1 per sensor delivered. Amazing, in my opinion.
> > I hope it helps, Tomas
>
> Using the GPIO pins would work if I needed one temperature, but I'm
> trying to monitor the temperature of the air coming out of 7 servers.
> Maybe I need different scripts or maybe I need a special out of tree
> driver.
> _______________________________________________
> PLUG mailing list
> [email protected]
> http://lists.pdxlinux.org/mailman/listinfo/plug
_______________________________________________
PLUG mailing list
[email protected]
http://lists.pdxlinux.org/mailman/listinfo/plug
