SparkFun Forums 

Where electronics enthusiasts find answers.

Have questions about a SparkFun product or board? This is the place to be.
By pov_nelson
I am tring to design a cheap temperature reading system to check the internal temperature of some grain storage bins on our farm. The bins vary in sizes but most are around 20 feet wide and 20 feet tall. Ideally I would like to hang a wire from the top middle of the bin down to the bottom and get a temperature reading every 5 feet or so. Then send those readings to a digital meter that would give me a temperature readout of the diferent sensors located in the bin. Is there any suggestions on what would work.
Thank You
By fll-freak
Fun project!

There are likely to be as many solutions as there are people who will answer this question. It will often come down to what people are familiar with.
If all you got is a hammer, everything looks like a nail"

The solution will also depend on your skill level and budget!

Your biggest constraint is the length of cabling needed. Putting a cheap thermistor (resistor that varies with temp) at the end of 20+ feet of cable is likely to induce so much electrical noise that the signal would get lost. Unless you used coax, but then the cost might go way up.

I would be tempted to put a tiny micro (like an MSP430 $1.50 each) at each measuring station. Many of these chips have a built in temperature sensor so you would nothing little more than the chip itself and a few resistors. These devices can talk on a simple serial bus (I2C). As long as you ran the clock real slow and used twisted pair wire (like Cat-5 ethernet cable) you should be able to run that distance. At the other end of the chain, you would have something like an Arduino (or even another MSP430) to collect the data. You could possibly convert it to an analog signal to read from a multimeter, you could display it on an LCD at that location, or you could use a wireless system to transmit the data to your home computer.

It will be interesting to see what other ideas you get!
By AndyC_772
There's no problem with noise pickup if you're measuring a quantity that varies slowly, like temperature. The cable will certainly pick noise up, but you can filter it out again with capacitors. Extremely low bandwidth isn't a problem if the quantity you're measuring varies slowly enough. I'd certainly rather try and make a resistance measurement from a long way away than try to run an I2C bus the whole distance. I2C is robust, but not infinitely so.

I'd have thought one of the biggest problems would be making a sensor arrangement that's physically robust enough to be buried in a grain silo. You'll want a bundle of cables that terminate in a thermistor every so often, but you'd need to protect the thermistors and solder joints somehow. Heat shrink sleeving might help, and maybe some glue to protect the joints. Even so, I'd expect filling or emptying the bin to pull quite hard on the cable, so make sure it's strong enough or you have some way to keep some slack in it.

If the cables are very long, their resistance might be significant and will need to be allowed for. You could either measure the resistance of the wire to each thermistor individually and use that to correct your readings, or perhaps use a true 4-wire resistance measurement.
By rgbphil

If you don't want any automation etc, and just want to hook up a multimeter, or hook up some fixed panel meters then have a look at the LM335/LM337. These output directly in mV/K. So at 25C you have 2.73+.25 = 2.93V.

For long lengths you might want an op amp with a fixed gain. Either way, they are cheap and don't require any micros.

By pov_nelson
Thanks for your replies.
The cable wires would be 40 to 100 feet long depending on the size of the bin. I would like the wire to end up on the bottom outside of the bin so I dont have to climb to the top or even run the wires to a central location by the group of bins.
What are the different types of read outs available. Can you make it work like the digital meat thermometer that I have where you just plug in the wire and it gives you a digital read out. Or run all the wires to a computer so all of the current temperature readouts show up.
Any Thoughts
By fll-freak
To keep the wires as short as possible, I would suggest starting it at the top of the silo, then run down the center and then out the side. You will likely want to run this electrical wire against a length of rope wire like they use to support small antennas. This will take the stress of the electrical wire when the time comes to empty the silo.

As mentioned before, you can get any type of display you want. Anywhere from old analog needle displays, LCD/LED displays like in a digital voltmeter or fever thermometer, all the way up to a display on your laptop.

Again, much of this will depend on your wallet and your skill level. I suspect that industrial companies all ready sell such a system but a a steep price. Grain silos must measure temp and humidity to keep the grain from spoiling.

On the cheap, I would experiment with a 50' length of cat-5 plenum cable, a thermistor, and a voltmeter. The thermistors are dirt cheap (see The cat-5 plenum might run $200 for 500 feet. Each cat-5 cable has 4 sets of twisted wires. That would give you four sensors if you keep a seperate ground for each one (likely a good idea).

You will need to learn to solder and dress your connections with shrink tubing and a good strain relief.

Now you can simply measure the resistance from the other end of the cable. Now you compute the temp from a table or equation. If this all works, you could try your hand at a small device to display the temps for you in whatever way you want.

Gave you give us an idea of your budget and skill level?
By groggory
I would make up a small 'measurement PCB' for each spot that will be measuring.

Pretty much use the reference design on these ... _sht10.htm ... er_cap.htm

Then put that in a small project box so you can mount it easily...or slip it into some pvc tubing ... or anything else that comes to mind.

Are you sure about 'hanging it from the middle'? Sounds kind of delicate.

I would have a microprocessor at the top of each grain silo that reads all the sensors that you have in that silo. Put an XBee on it. Then have that sensor wirelessly transmit its data to the mothership for processing (another micro + xbee). That mothership would have an SD card on it to store all the data it collects.

Then when you feel like it you pull the SD card, bring it up in excel, and see what's been goin on.

Honestly, you are like a case study for XBee. That is almost exactly what they were built for.

Hell, if you really wanted to be awesome you wouldn't even use any micros in the field. Drive everything off the xbee. It can do the cyclical sleep, data collection, transferring, mesh, everything.
By esklar81

I suggest you take a look at the "1-wire" temperature sensors. These have the advantage that the conversion to a digital temperature signal is internal, so you don't have the analog signal degradation issues that an RTD or thermocouple would present.

1-wire is rather slow, as modern interfaces go, but I don't see why you'd want to read each temperature more than once a minute, and a 1-wire system should be capable of running much faster than that. (If your grain temperature is rising more than 1 C/ minute... :shock: ) In fact, your application is similar to one of the purposes for which 1-wire was developed: building temperature management.

You will, however, need to provide something to drive the 1-wire bus. There are devices that you can use to do that from a PC.

Alternatively, you can provide a microcontroller to poll the sensors and display, store, or transmit the data. (Depending on how fancy you get, you could even give your storage bins their own web pages and check the temperatures from just about anywhere in the industrialized world.) If this is your first microcontroller project I (at the risk of setting off a religious war :wink: ) suggest Arduino. Arduino is intended for people with little or no experience in digital electronics and there's a huge community of support available.

An application note that Maxim provides indicates that you could get the range you need from a single-line network.

Have Fun,
By pov_nelson
Thanks again
I have a basic understanding of wiring, soddering and other electical stuff. I have a volt meter. I do most of the basic electrical work on our farm. My budget is value based. We have around 50 bins that if it is cheap enough I would like to put something in all or most of them. I dont know if I am even close but I could probably afford to spend 25 maybe 50 dollars per bin.
Some of the hopper bottom bins I could have the wire coming out of the bottom of the bin but the flat bottom bins it needs to come out the top so the wire does not get cought in the auger when we clean out the bin.

By esklar81

Your budget of $25-50/bin seems reasonable to me.

For the approach I had in mind, the sensors are about $4 each (a bit more onesy-twosey, a bit less in quantities of 10-99, $3.40 at 100 pieces (20 bins' worth)). To cover a 20 ft run at 5 ft intervals, you'll need 5/bin.

I looked a bit more closely at the application note to which I provided a link and found a claim of being able to manage a 200 m line if you use two pins of a microcontroller and a FET for each line.

I have, of course, no clue as to how your bins are laid out. But, if you can find a location so that each bin requires a run of no more than 200m, you can put each bin on its own line and share an Arduino among, say, half a dozen bins. That would let you share the controller and display costs (ballpark of $75/set) among the bins. Add a few bucks for cabling, and you're still well below $50/bin. If you can get adequate performance with a more complex network, the controller and display cost per bin will go down even further.

By pov_nelson
Everything sounds great so far. It looks like there are two main options I have. The 1 wire temperature sensor or the thermistors. Which one should I use??? Will they stand up to the conditions. How long are they expected to last?
By esklar81

Although thermistors are less expensive ($1.25 vs $4.25/point), I'm still leaning toward the 1-wire sensor approach. This is primarily because the resistance of the thermistors that SparkFun sells are of the order of kilo-ohms in the temperature range of interest. That means the currents to be measured (assuming single-digit voltage excitation) are of the order of mA or less and, therefore, quite subject to noise under the measurement conditions of this project. That noise concern could be addressed by putting an instrumentation amplifier in each thermistor location, but that would probably remove most of the cost advantage while increasing labor.

If the initial intention is to be able to walk up to a bin and "take its temperatures" (rather than monitoring all of the temperatures of all the bins at the same time), the cost of the project can be reduced to ~ $100 + ($25/bin) by
a) putting a string of 1-wire sensors in each bin and terminating the cable with a suitable connector, and
b) building a handheld box with a battery, microcontroller, and display.
This is probably a good way to start, as it eliminates the concerns with long wires and will not require changing the sensors if you decide to tie them together in the future.

The thought occurs to me that you'll need to identify which sensor is at which height and build a lookup table. As each of the sensors has a unique ID, though, you can have the microcontroller display the bin to which it is connected and the temperature for each height in the bin. (1-wire protocol provides a means for the controller to determine, when you connect the controller to the string of sensors, which sensors are connected.)

I took a quick look at the link you provided and I wouldn't be at all surprised if they are using the 1-wire sensors. Do you have any idea what their pricing is? (In fairness, they are providing a sheath (although I'm not sure that's a good idea from a measurement perspective) and retraction.)

Have fun,