PDA

View Full Version : OT: Need simple electronics help



Steve Steven
07-14-2007, 08:40 PM
I just recently put in a ground source heat pump, and I want a way to check the inlet and outlet temperatures of the water flow. I have a set of valves on a mettalic "Tee" in each line, which gives me a metal place to measure the flow temps. There is enough of a difference to readily notice by hand. Temps would be in the 45deg - 75deg F range.

What I would like is a way to measure by thermocouple using a cheap HF $4 digital VOM as a readout, I remember reading about this some years ago. I have a commercial Chromel/Alumel thermocouple (type K) which could work I think.

Can anyone point me to a simple circuit that would do what I am looking for?

TIA,

Steve

J Tiers
07-14-2007, 09:09 PM
Your BETTER way is to obtain two of the national Semiconductor LM35 sensors.

Yes, they will need a power supply, although it can be a 9V battery. It can be turned on with a push switch when you want a reading.

The HUGE advantage is that these sensors are linear with temperature, which the thermocouples are not, and have a much larger output. The basic part has an output of 0.01V per degree F or degree C, both types are available.

With two resistors, the output can be scaled to any volts per degree that is wanted, up to the voltage rating of the part.

You can get a direct reading of temp on your voltmeter display, no funny conversions needed (aside from decimal point). By reading between outputs, you could get differential temp, etc, in one reading.

The parts are a few cents in manufacturing volumes, so cannot be very expensive in low quantity. Digi-Key, etc.

darryl
07-14-2007, 09:14 PM
I'm not familiar with a K type thermocouple, though I could look it up and get the specs. A thermocouple is going to generate a voltage depending on it's temperature, so measuring the voltage and changing the temperature of the thermocouple will give a variable voltage reading on a meter. Because a thermocouple is such a low resistance device, pretty much any meter could be used and the readings would still correspond closely to temperature.

There won't be much voltage, probably in the low millivolts, so it's likely that the lowest voltage range the meter has would have to be used. You could calibrate by using ice water and boiling water- note the voltage levels accordingly, then interpret the range of thermocouple voltages as temperature. If, for instance, 10 millivolts corresponded to ice water, and 60 millivolts was for boiling water, then 35 millivolts would correspond to halfway between, or 50C.

I'm going to try finding the specs on that and them maybe I can give closer figures.

darryl
07-14-2007, 09:23 PM
I'm not familiar with a K type thermocouple, though I could look it up and get the specs. A thermocouple is going to generate a voltage depending on it's temperature, so measuring the voltage and changing the temperature of the thermocouple will give a variable voltage reading on a meter. Because a thermocouple is such a low resistance device, pretty much any meter could be used and the readings would still correspond closely to temperature.

There won't be much voltage, probably in the low millivolts, so it's likely that the lowest voltage range the meter has would have to be used. You could calibrate by using ice water and boiling water- note the voltage levels accordingly, then interpret the range of thermocouple voltages as temperature. If, for instance, 10 millivolts corresponded to ice water, and 60 millivolts was for boiling water, then 35 millivolts would correspond to halfway between, or 50C.

I'm going to try finding the specs on that and them maybe I can give closer figures.

nheng
07-14-2007, 09:39 PM
Using the LM35 (LM34 for deg. F) or Analog Devices TMP35 is a good idea. The TMP35 will operate down to 2.7 volts but is only in deg. C. 2 AA cells would probably give you 5,000 or more hours of continuous operation. The LM34 would need 4 cells but so do most toys :D

Using two of them, each with its own floating battery pair (and separate ground), and a few switches, you could wire the two outputs in series with the lower temperature reversed in polarity. This would give a dirt cheap and simple HOT-COLD differential readout. Flip the switch to read HOT or COLD separately.

Den

deltap
07-14-2007, 09:41 PM
I'm not sure if a $4 VOM will give much better results than holding your hand on the pipe. HVAC manufacturers use low cost thermistor sensors for the majority of readings- 10K ohms @ 70 F. For service people they supply tables for voltage and resistance that correspond to temperature. For greater accuracy platinum sensors are used. One low cost solution would be to tape a $20. digital pocket thermometer to the pipe and insulate it. An oil filled well in the pipe would give better results.

nheng
07-14-2007, 09:50 PM
I haven't seen the $4 wonders but you'd have to work pretty hard to make a meter less than 2+ digits of accuracy.

The LM34 has a 1 deg. F accuracy. Using two of them would give a differential accuracy somewhat better than this.

Deltap is correct about insulating whatever you use for a sensor as this is where you will lose more accuracy than anywhere else.

He also mentioned digital pocket thermometers. In my xmas stocking, I got a very nice Polter digital cooking probe that cost only around $14. It is switchable deg F or C and has a fairly wide range. I told Santa that I would be using it to probe IC temperatures and it might look different when I was thru with it. :D Den

darryl
07-14-2007, 09:51 PM
Just read up some on thermocouples. Type K has .001 volt output at 77 F, so that doesn't lend itself well to direct measurement with a voltmeter. The only other way I can think of that doesn't involve any electronics is to use a thermistor, and measure ohms with a meter. It would need to be calibrated, of course. Looks like the better way is with a temperature sensing chip, as others have already talked about.

One thing about using a meter to measure ohms- this reading requires the meter to pass some current through the device, a thermistor in this case. That represents a power consumption. Measuring voltage doesn't take nearly so much power, only what it takes to put the readings on the display, and some updating of the data. What this all means is that if you use a thermistor as a sensor, the meter's battery life will be lower than if you were to be able to take a voltage reading from another sensor type.

So I suppose it's best that you go to a simple electronic circuit and have a power supply to run it. Maybe you can get the meter to run off the same supply.

Dawai
07-14-2007, 10:28 PM
Parralax (basic stamps) has some great thermocouple monitoring circuits and sensors on thier site. Programming is pretty simple and the software is free.
That would give you some easy ways to control the flow, when needed, when warm enough, etc..

Opto-22 cards with thier 5volt ttl control and isolation to control 120 volt objects works great.. INdustrial stuff that never seems to break. I have cycled them opto output modules fast enough to control electronic ignitions on a home made harley distributor once, also a tattoo coil driver for a tattoo machine.

Steve Steven
07-14-2007, 10:48 PM
Well, this is really great! I had no idea such a thing as the LM34 was available. Not being electronically inclined, I can read a data sheet however. It looks like the TO92 package is cheap, and might be suitable.

How would you suggest to mount it to the as-cast bronze T in the line? Could I just epoxy it to the T? I would need a 5 to 20 Volt supply ( 9 V battery?) and set the output to a milivolt meter, is that right? Could I use the HF meter, or would I be better using a analog meter, either a actual meter or an anlalog VOM? It looks like the case is the actual temp sensing element, I may be better to get the TO46 metal case version, twice the price but might work better in this instance. Mabye I should strap or wire the case to the T, and insulate the package for better heat transfer.

Looks simple enough, I can set it up with ice water and boiling water to get a basic voltage standard, I would be happy to get it within +-5 deg, i am mostly interested in the difference anyway.

I printed out the spec sheet from DigiKey
http://www.national.com/ds/LM/LM34.pdf
and it will be perused in detail later.

Please comment more, I appreciate your comments greatly!

Steve

nheng
07-15-2007, 12:18 AM
Steve, Save that ice for a cold drink. These chips are accurate to 1 degree and linear to +/- a half degree over their ranges. You may want to stick with the to-92 cases instead of the metal so there's no fear of a connection to the piping. You could mount it in place with a little squirt of expanding foam insulation and solve two problems at once. Tack the device against the spot you want to monitor with super glue or even a temporary fun-tack type adhesive wad. Encapsulate it with the expanding foam and it should be pretty well protected from the ambient temperature.

You do this all on your own of course as this information is all presented as an educational example ;)

J Tiers
07-15-2007, 12:23 AM
Concur that it is accurate as-is.

Epoxy to the pipe, though, then cover with non-corrosive foam. That way you insulate from environment and not from the pipe.

With common supply, which is no problem, you can direct-read the differential temperature, which is what I gather you are after anyhow.

nheng
07-15-2007, 01:41 AM
"With common supply, which is no problem, you can direct-read the differential temperature"

JT is correct on this. When I suggested wiring them in series and reversing one, I had visualized how to do it but didn't finish thinking it thru :eek: All you need to do, of course, is to measure across the two sensor outputs.