SparkFun Forums

[pov_nelson]

Ya I just talked to them and they wanted like 500$ per bin and 800$ for the monitor. What a joke!!! What kind of monitor would i use to check the temperature? What kind of cable wire should I get and what are the conectors i should use to connect the wires to the monitor?
Thanks
Kevin

[rgbphil]

I got the part number wrong in the last post sorry.

I still think a direct temperature to voltage sensor is better for the OP. Fiddling about with thermistors and micros etc is all very well, but perhaps a little complex. Don't forget that a thermistor needs calibrating and a gaggle of circuitry (albeit this can be as simple as a resistor), it also doesn't give a voltage/current reading directly related to temperature. Also the response to temperature will not be linear, except for a narrow range of temperatures depending on the sensing circuit.

I'm assuming there is no automation, ie a computer looking at the sensors giving alarms etc and you want to just look at the temperatures. If you want alarms and what not, then the thermistor/micro approach would be better.

Otherwise:
Temp to voltage sensor : http://jaycar.com.au/productView.asp?ID ... rm=KEYWORD
Panel meter : http://jaycar.com.au/productView.asp?ID ... rm=KEYWORD
Some cable. I suggest some shielded three or four core cable. Connect the shield to ground, one core to the temp sensor and the other to a 5V supply.
Something like: http://jaycar.com.au/productView.asp?ID ... BCATID=543
a 5V supply
Miscellaneous connectors and terminals.
Sorry the part numbers etc were from an australian site....I just had a quick fish about.

You could pot the sensor and wiring in some epoxy so you'd just have the cable poking out of the blob of epoxy and run it down to the panel meter+PSU.

A more professional job would have, as others suggest, a network of sensors (saving on cabling costs) hooked to a computer or micro which could give alarms and what not.......it depends on what you want.

Phil

[pov_nelson]

Thanks
Just a couple of questions. I was going to go with the one wire sensor but then you said that they are a little complex. I am new at most of this so what problems do you see me having in setting them up. Do you have to program anything or just wire it in and then hook up the digital meter and get the temperature? Does the panel reader that you recomended give you the temperature or what is the number? Does it matter how long the cables are? Some to the bins only need about 20 feet but some may need 100 feet.
Kevin

[esklar81]

Kevin,

As you are not (yet ) comfortable doing things with Arduino or other microcontrollers, the easiest way for you to start (if you have a computer that's portable enough for you to be comfortable carrying it from bin to bin) is to get a USB to 1-wire bridge, a few of the sensors and an old modular phone cord.

I have not used the 1-wire driver software that Maxim/Dallas offers, but you can download it from their site and have a look. My guess is that they've made it pretty easy to use, as they make their money selling sensors. (The software's available free from their website.)

On the hardware side, it appears you'll need to find out which of the 4 conductors on an RJ-11 connector is to be connected to which pin of the 3 pins on each sensor, then solder the sensors into the appropriate places in the cable. You'll need to insulate the sensor pins from each other and provide enough mechanical protection for the assemblies to withstand the intended use. (Those tasks are likely to be just about the same with any sensor technology, although using digital communication makes the wiring practices less critical than they would be if you used analog communication, such as with thermistors.)

For mechanical protection, how about threading the assembled cable/sensors through some food-grade plastic tubing? I'd avoid metal sheathing, because the metal is likely to conduct heat from sensor location to sensor location.

Eric

[pov_nelson]

Eric what is your opinion on Phil's suggestion?
Kevin

[esklar81]

Kevin,

I'm somewhat concerned about noise, particularly in your use environment. OTOH, it's a cheap solution (if you have a cheap source of shielded cable) and probably worth a try. When you're constructing the circuit as shown in the application note, remember to keep the resistor far from the sensor, otherwise your measurements will be distorted by the resistor heating the sensor.

You should be able to find a supplier of the sensor much closer to you than the one to which he linked.

On a minor note, I'm not sure the panel meter to which he linked has a suitable range. However, you should be able to use your own voltmeter, at least initially.

Eric

[lyndon]

I'm with rgbphil. When you're a beginner, it's best to take small bites first. Trying to do a large complex project often leads to discouragement and wasted time and money.

A few sensors like the trivial-to-use LM35 (provides a linear 10mV per degree F), a rotary switch, and a cheap panel voltmeter (can find those LCD ones for $20 all over the place) and you're in business. Use the switch to select which sensor you're looking at. Good quality instrumentation cable to the sensor will eliminate noise. Later improvements can be your own custom built, Arduino based meter with recording capability. You can add a multi line display to see all temps at once, etc.

But starting with an easy to use sensor and a meter is a great way to get your feet wet.

[rgbphil]

esklar is probably right about the panel meter.....I just threw in the parts list as a starting point.

You probably want something that reads from 0-5V...this should give you a range up to around 47degC....if you want higher (I have no idea how hot those silos get) then you'll need a higher panel range and a higher voltage supply....do you have an idea of your max temp?. Also whats the accuracy you want. I'm guessing around 1-5 degC would be ok, the LM335 can do 1degC without any calibration. Anything better needs a bit of fiddling, check out the datasheet for details.

There should be plenty of moving coil (ie a needle) meters around that do that, otherwise some panel meters can be configured to have adjustable ranges by putting links in at the back of them.

As for noise, again the other suggestions are good, basically the thicker the cable the better, particularly if it's shielded. When using shielding, try to make sure all wires from shields are connected to the same place (ie don't string the grounds around and make a line or a 'ground loop', and by generous with the thickness of the ground wiring. However bearing in mind you app is pretty simple, there's no need to get too precious about these sorts of things.

Another noise avoidence trick would be to put a reverse bias zener diode (say 7V) across the power and the temp output wires to ground. If there is a spike induced by nearby machinery the zener diode will conduct the extra voltage to ground and keep everything protected. These should be well under a dollar each. I'd put zeners at the sensor end of the cable, and the meter end.....though if you are using a moving coil meter you could probably avoid the sensor at the meter end...they're very robust.

Phil

[pov_nelson]

Thanks Guys
How thick of wire are we talking about and is there any limits to the length of wire i can run before the interferance is too high?
Kevin

[AndyC_772]

I'm afraid I really don't see why there's so much to worry about regarding electrical noise.

Temperature is a very slowly varying quantity, and the change in resistance you get from a thermistor is quite large and easily measured. All that's needed to remove the noise - and frankly, I can't see a grain silo being an especially noisy environment anyway - is a big enough capacitor at the measuring device. Remember: noise can be filtered out of an analogue signal, provided the signal is constant, but once it exceeds the margin available on a digital signal then you'll get errors.

Turning the resistance measurement into a temperature reading is, as others have said, the tricky part. Personally I don't think it's any more difficult than trying to get a "direct" temperature reading at each and every measurement point sent back to a display, but that's just a matter of opinion.

If you can find a thermistor whose output is fairly linear over the range you're interested in, it might be possible to turn its output into volts/degree with a couple of op-amps, which could then drive an off-the-shelf meter. At least then no programming would be necessary.

[pov_nelson]

Hey Guys
I think I am going to get a few of each and just play around with them for a while. I have recieved lots of recomendations for parts and I am a bit confused on what exactaly I need. What are the best yet most cost effective sensors and what is the exact type of wire I should use for each?
Kevin

[pov_nelson]

My Laptop has both an ethernet port and a phone port in it. Do i still need to get a ds9490R adapter or will it work with what I have?
Kevin

[pov_nelson]

I also want to play around with an Arduino so what hardware would i need to get started.
Thanks
Kevin

[fll-freak]

Kevin,

If you wish to pursue tinkering with an Arduino and thermistors, then the following is likely a good choice:

http://www.sparkfun.com/products/9950 (currently on back order)

http://www.sparkfun.com/products/9729 (makes it easy to connect wires)

http://www.sparkfun.com/products/298 (a wall wart to run it away from your computer)

http://www.sparkfun.com/products/250 (some generic thermistors)

http://www.sparkfun.com/products/9393 (a small lcd display to print the temperature measurements)

Wire. Best choice would be to see what you can get surplus in your area for a test. You might be able to scrounge some plenum cat-5 cable from a construction site. It is not shielded, but the twisted pairs would help your signal integrity. It also has good mechanical properties (but not strong enough to handle the downdraft of 1/2 million pounds of grain!).

You could make a very simple prototype with the above. If you can get this small scale system working (even with just a few feet of wire) then you might be ready to move up to a full scale model. Kind of like the Mythbusters!

Let us know how this works out for you.

PS: I was right that there would be many ways to skin this cat.

[SpikedCola]

I would also like to recommend the use of 1-wire temperature sensors for this application. Your requirements are some of the main reasons 1-wire was created. All of your sensors would be on a data wire, for easy reading. And there's a lot less chance of noise & signal length having a deteriorating effect (vs an analog sensor).

There is a Dallas Temperature library for Arduino that you could implement with all your sensors in about half an hour. /AND/ Dallas samples the temperature sensors for free. What more could you ask for.