Hi all. I want to run a few remote temperature sensors to use with the arduino. I was going to use some 10k ntc's with a voltage divider close to the sensor. I could either do that, or just measure the resistance at the end of the cable (max run will probably be about 30-40 meters) and have the divider there?
A third option would be to use some 3 wire sensors but I have never used them and I will be using 2 spare wires that form the wiring for my alarm system, so I would rather just stick to the 2 lines.
It will depend on the location, if there is too much enviromental noise then an analog output device will be more susceptable that a digital output like a ds18b20,
also with an analog output the wire resistance needs to be accounted for
both methods are quite cheap and easy to implement, maybe the best way to find out is to set it up and test it out
winner10920:
P.s a ds18b20 can be used in a 2 wire parisitic power mode, which is slower but is only two wires
Not so good for long distances though, probably better off commoning grounds with the alarm and using the spare pair for +v and signal if the 1 wire is used. Pity the 1 wire devices are only accurate to .5 degree c, quite a drift.
I would be inclined to run separate wiring in for the sensors, unless this is all prewired and you're trying to use existing spare cores?
The wires are not run yet, but to keep things simple i was planning to use the spare 2 cores in the 8 core cable.
would those 1 wire devices be more accurate than a standard voltage divider? I suppose it would depend on the resistance of the wiring I'm using? I guess I could common the grounds that shouldn't cause a problem should it?
The 1-wire devices are digital output and so don't have any accuracy issues with the wire length. As noted though they may have other issues with long wires.
The LM134 produces a current output, 2 wires and should be no wire length issues.
Using the suggested circuit as shown in the data sheet, at 25C you should read 3v from the sensor.
It produces 10mV per degree K (same as C but with an offset) and the Arduino ADC reads approx 5mV increments, so I think you'll get about .5C resolution, better than the sensor's accuracy by the look of it.
I found them in a few places but they aren't cheap, around $6 by the look of it, I don't know why they would be so much.
I would think your best off with a ds18b20, definetly cheaper than that, 9-12 bit resolution, and could have multiple off the same 2 wires, and programming is simple, especially with the dallas one wire library
Thanks guys, the thing is I've got 100's of 10k leaded ntc sensors so I might just use them, and try and calculate the drop over the cable? Not sure if I should have the divider set up at the thermistor end or the measurement end?
To save wires you should have it at the other end, by the arduino, other wise you need 3 wires to the sensor
you will need to measure the resistance of the wire going to the sensor to get an accurate reading, or take a reading at a known tempature at the board then wire it to the longer wire and measure it at the same known temp, and get the offset directly by the difference
winner10920:
P.s a ds18b20 can be used in a 2 wire parisitic power mode, which is slower but is only two wires
Not so good for long distances though, probably better off commoning grounds with the alarm and using the spare pair for +v and signal if the 1 wire is used. Pity the 1 wire devices are only accurate to .5 degree c, quite a drift.
I would be inclined to run separate wiring in for the sensors, unless this is all prewired and you're trying to use existing spare cores?
In practice I've tested several batches of DS18S20's and found a standard deviation at room temperature of less than one LSB (so about 0.05 deg C) - that 0.5 deg is worst case across whole temperature range.
Lurch:
Pity the 1 wire devices are only accurate to .5 degree c, quite a drift.
In practice I've tested several batches of DS18S20's and found a standard deviation at room temperature of less than one LSB (so about 0.05 deg C) - that 0.5 deg is worst case across whole temperature range.
Probably worth pursuing then. I am in the process of building some sort of multi zone heating control and to save inputs the 1 wire devices are ideal, wasn't too keen on the accuracy though looking at the datasheets. Bit more confident in ordering some now.
Yeah I would definetly recommend them, I haven't tried long cable lengths with parisitic mode but I know that in their regular power mode they work great even in a car, which is a pretty noisy enviroment
The best part about them is that there are no adjustments needed for wire length or interference, they do all the data conversion and report back
so for a network of them it saves alot of time, also the ability to expand off any node is great
There are many advantages to using 1-wire temp sensors so I think I'll agree with the guys on that, I understand they don't work too well over long wires using parasitic mode though so I'd plan to supply power.
