Building an inline temperature sensor

[DarkenReaper57]

Instead of dishing out 30-40 bucks for an inline temperature sensor with a LED display, I opted to make one myself. I bought this LED thermal display http://www.jab-tech.com/customer/product.php?productid=2642&partner=froog.

Now, initially I was going to follow this guide http://www.ocforums.com/showthread.php?t=304440 (it is about halfway down) but surprisingly, I could not find small chunks of 1/2'' ID copper pipe or 1/4'' OD copper pipe at lowes or home depot. They only sold them in large spindles at 15 bucks+ a pop.

Instead, I bought a 1/2'' ID tee and I intend on using that. The only problem is (if it is a problem) that the part that the thermistor goes into is rather large, comparatively. If I fill the end up with JB weld, then drill a small hole for the thermistor, then apply more JB weld to the thermistor, that should be fine, right?

The only other possible problem is the thermistor itself. It doesn't look like the one in the above guide. I just want to make sure the tip will be fine in water. I can even lightly coat the thermistor's tip with JB weld if needed.

Here is a pic of the sensor:



So, what do you guys think? Is my plan ok?

 

[BellaCroix]

Did the same thing in my test loop I built 8 months ago to try my hand in watercooling before laying down the big bucks for German gear.

The only exception is that I used aquaium silicone sealant. When I used JB Weld last time I noticed it heated up a little when I mixed the epoxy and was concerned that it would warp or melt the plastic barb. I'm probably totally off on that thoery but I figured it was better safe than sorry, plus aquaium sealant stays flexible.

Instead of drilling though the silicone though I just glued the fitting in place and let the sealant set over a few days.

I used the same fitting you pictured... no problems at all. Be sure you don't have the probe flapping around much in the waterflow. Again, I'm sure it's not a problem if it is but it looks a little more professional if it's not.

The next time I do it I'm going to put the probe further into the fitting then use the same silicone glue to attach the probe to the inside of the fitting to lay flat along with the waterflow.

Edit: I read the guide in that link you posted and he indicated he was concerned about using silicone to seal it because of the pressure created from his 1100+ GPH pump. They use silicone sealant on aquaiums over 150 gallons. I'm no hydraulogist but if it can hold the weight and pressure of 150 gallons of water, I feel safe in using to seal a line with what I image is less overall pressure. It's probably not a good idea if using copper since it really wont bond to the metal and the pressure will just force it out like a cork... but if you're using a plastic barb you should be okay.

 

[DarkenReaper57]

Well, the only reason I am thinking of filling the end with JB weld and redrilling after it is dry is to make sure the sensor is centered. It will probably be a lot easier that way, but I could be wrong.

I didn't realize that JB weld heated up...how hot does it get? I do have some liquid nails that I can use instead if needed. I don't want to damage the sensor or the tee.

How far did your place the sensor into the pipe? Did you make sure the little sensor ball touches the water, or is just the thin film surrounding it sufficient? Can the film be trimmed at all? Did you lightly coat with silicone?

Oh, did you use a regular tee, or make your own with a smaller end for the sensor? I'm hoping that the larger end won't be a hassle for me.

 

[ikellensbro]

Look for a 6" or so long piece of 1/2" copper pipe witha a round/point end, they're use when setting up plumbing for showing where water lines are, and should set you back a buck or 2 if you're still interested in using copper pipe.

Your idea sounds like it should work ad long as your thermistor can function submerged in fluid with antifreeze/whatever you put in your loop. I may just see when I get my P160 if the thermistors work underwater & try using one with a T line. For filling the gape in the T-line, try using a cork or rubber plug with a notch cut in it for the thermistor line, then fill the top cut with a little aquarium sealant/JB Weld/Goop/etc. If I try this, I'll give you all the credit for coming up with the idea

 

[Top Nurse]

Having disassembled an Innovatek in-line temp sensor I can give you some pointers on doing it getto style. First of all the temp sensor used in the Innovatek product is the same sensor you showed in your pic. They are called NTC sensors and are fairly inexpensive and available from a wide variety of sources. My suggestion is for you to take a small piece of round copper stock. Say about 3/16" to 1/4" by about 1" long and drill out the center leaving one end closed. Try and leave maybe about 10-20% of the length undrilled. Now take your sensor and push it into the hole you just drilled. What you want is for the sensor tip to be at the bottom of the hole.

Now go find yourself some scrap cable TV coax and strip off the shield. What you want is that insulator that surrounds the center wire. Get about 1/2" of the stuff and you will use this insulator to wedge the NTC sensor so it stays at the bottom of the hole against the copper wall. Now you will add some silicon sealant to the hole just to keep the wire from getting pulled out of your sensor tube. Now you just need to figure out a way to get your copper round into the water stream a bit and you will be one dude.

 

[tcc]

I plan on putting a temp sensor in my WC setup. But I went the lazy way

Bought a 1/2" ID barb on both ends and a temp sensor that is used for fish tanks. Going to drill a hole in the barb dead center and epoxy the sensor in the hole. Voila!! And it's back lit too

 

[wwparrish]

agree on the silicon, better than JB weld. If you check the sweat fittins you can find a cap for your tee to use on the end the sensor wire comes out. After the part of the tee is filled with silicone you can drill a small hole for the wires and expoxy the cap on to prevent any possibility of the silicon "plug" from coming out.

Paint all of the sensor that will be in the water with clear nail polish. I have had the "tape" that covers the sensor of the style you show, delaminate (come apart) after a while. The then exposed thermistor will absorb water and fail a while later. The little round bead type with no tape should also be "painted".

 

[DarkenReaper57]

agree on the silicon, better than JB weld. If you check the sweat fittins you can find a cap for your tee to use on the end the sensor wire comes out. After the part of the tee is filled with silicone you can drill a small hole for the wires and expoxy the cap on to prevent any possibility of the silicon "plug" from coming out.

Paint all of the sensor that will be in the water with clear nail polish. I have had the "tape" that covers the sensor of the style you show, delaminate (come apart) after a while. The then exposed thermistor will absorb water and fail a while later. The little round bead type with no tape should also be "painted".

The cap thing won't work because I bought a barbed tee .

To avoid buying more stuff, do you guys think it will work to just fill the end with JB weld, let it dry, drill a small hole for the thermistor, apply more JB weld, then stick it into place? I'll heatshrink the top where the sensor goes into.

 

[zer0signal667]

The cap thing won't work because I bought a barbed tee .

To avoid buying more stuff, do you guys think it will work to just fill the end with JB weld, let it dry, drill a small hole for the thermistor, apply more JB weld, then stick it into place? I'll heatshrink the top where the sensor goes into.

Yes, I think that's a fine idea. I'm also voting for coating the thermistor in something. I would use some type of epoxy just to make sure it's not going to come off and fly into your cooling loop. Oh, and JB weld won't get hot enough to damage anything. There's a slightly exothermic reaction going on when it cures, but it's not going to heat up nearly enough to cause a problem.

 

[DarkenReaper57]

Yes, I think that's a fine idea. I'm also voting for coating the thermistor in something. I would use some type of epoxy just to make sure it's not going to come off and fly into your cooling loop. Oh, and JB weld won't get hot enough to damage anything. There's a slightly exothermic reaction going on when it cures, but it's not going to heat up nearly enough to cause a problem.

Coating the thermistor in JB weld should work, right? If not, I can always use liquid nails. The guy in the guide used JB weld, so I don't see why it would be a problem...

So, how far should the sensor go into the pipe? Until the little bead sensor itself can touch water, or is the thin film surrounding the assembly sufficient?

 

[zer0signal667]

Coating the thermistor in JB weld should work, right? If not, I can always use liquid nails. The guy in the guide used JB weld, so I don't see why it would be a problem...

So, how far should the sensor go into the pipe? Until the little bead sensor itself can touch water, or is the thin film surrounding the assembly sufficient?

JB weld is fine, it is an epoxy. Have the bead on the tip submerged in the water.

 

[wwparrish]

due to the flow in the pipes and the excellent heat conduction of water, you dont need to get the sensor down into the main flow. Just get the bead out of the "plug" and you will be good to go. The less of the sensor you have "hanging" in the water the better. The reason I mentioned the nail polish is that it can be applied in a very thin layer and any female will have some handy usually at no cost. JB weld would work fine too. I like to make sure whatever I put on the themistor is very thin so that the sensor will respond quickly to temp changes. In the long run whatever you coat it with will heat up to the water temp and give you a decent reading.

I might suggest another thing. test your sensor. before installing it dip it in a container of ice water. Fill a cup with ice, as much as it will hold, then fill with water. Let it sit for 5 minutes. dip in your sensor an inch into the middle of the container and let it stabilize , it shoud read 32F or 0 C (assuming it will read that low, check your documentation). I wont go into the physics of why that is accurate but it is. Any difference from 32 or 0 will tell you the accuracy at that temp and while not perfect will tell you if you need to factor in the error in your actual working temps.

 

[DarkenReaper57]

due to the flow in the pipes and the excellent heat conduction of water, you dont need to get the sensor down into the main flow. Just get the bead out of the "plug" and you will be good to go. The less of the sensor you have "hanging" in the water the better. The reason I mentioned the nail polish is that it can be applied in a very thin layer and any female will have some handy usually at no cost. JB weld would work fine too. I like to make sure whatever I put on the themistor is very thin so that the sensor will respond quickly to temp changes. In the long run whatever you coat it with will heat up to the water temp and give you a decent reading.

I might suggest another thing. test your sensor. before installing it dip it in a container of ice water. Fill a cup with ice, as much as it will hold, then fill with water. Let it sit for 5 minutes. dip in your sensor an inch into the middle of the container and let it stabilize , it shoud read 32F or 0 C (assuming it will read that low, check your documentation). I wont go into the physics of why that is accurate but it is. Any difference from 32 or 0 will tell you the accuracy at that temp and while not perfect will tell you if you need to factor in the error in your actual working temps.

Thanks for the suggestion on checking how accurate the temperature sensor is. I'm going to test that now.

Does anyone know if the thin film surrounding the sensor can be trimmed without damaging it? I want to cut close the the actual sensor head so less is sticking out into the water.

EDIT: I did the ice cube test thing, and the lowest it would read was 3.1C Documentation says it will read as low as -10C, and accurate to 1.5C. It seems to be more accurate than that, however, since it detects room temperature to .2C when compared to a thermometer in the same room. I do live in Florida, so perhaps the ice water (even though I used a large glass) doesn't quite hit 0C because it is melting so quickly, releasing latent heat during the process?

 

[ikellensbro]

You could try checking your temperature & compare the results with a thermometer for accuracy.

 

[DarkenReaper57]

You could try checking your temperature & compare the results with a thermometer for accuracy.

That is basically what I did in the above post .

 

[wwparrish]

sure you can trim it, its just tape, just be very carefull to cut just the tape.

/giggle check it with a thermometer, who checks the thermometer for accuracy ? Lab thermometers are a bit pricey. Again I am not going to explain the physics but an icewater bath is EXTREAMLY close to EXACTLY freezing (unless you live on a mountain top or bottom of the sea).

 

[wwparrish]

hmm reread your last post, nope as long as there is any ice left in the water it sould be very very close to freezing if you let it sit long enought to stabilize. My best guess is that the error is not linear. Since its not a hard thing to do, try it again with lots of ice, little water and stir it for a couple of minutes. I must admit its an odd result. I have done this many times and have not seen that much of a discrepecny in that the error is still about the same when compared to room temps. I do have a lab thermometer but its not a super good one. /shrug Well I am making a mountain out of a mole hill, you will find the best use for the termo after the initial "thats cool" wears off is to notice when things start going wrong or maintence is needed. What I am trying to say is that whatever it reads, it reads but you want to look for unexplained changes like the temps creeping up in a couple of months due to dust in the rad or some such thing. Dont worry too much about it.

 

[DarkenReaper57]

hmm reread your last post, nope as long as there is any ice left in the water it sould be very very close to freezing if you let it sit long enought to stabilize. My best guess is that the error is not linear. Since its not a hard thing to do, try it again with lots of ice, little water and stir it for a couple of minutes. I must admit its an odd result. I have done this many times and have not seen that much of a discrepecny in that the error is still about the same when compared to room temps. I do have a lab thermometer but its not a super good one. /shrug Well I am making a mountain out of a mole hill, you will find the best use for the termo after the initial "thats cool" wears off is to notice when things start going wrong or maintence is needed. What I am trying to say is that whatever it reads, it reads but you want to look for unexplained changes like the temps creeping up in a couple of months due to dust in the rad or some such thing. Dont worry too much about it.

Ok, I'll retry it again with a larger glass and more ice. Also, something that seemed interestintg (and counter-intuitive) to me - the sensor detected warmer temperatures at the bottom of the cup than roughly 1 inch below the surface. I suppose this does make sense in a way, as lakes freeze top to bottom, and warmer water is at the bottom, etc...

What is the error range usually at 0C for a normal thermometer or sensor, from your experiences?

 

[wwparrish]

susprising small, even the cheap ones are within a deg C. I hate to get you off on another tangent but most of the chips inside the display are very accurate, its the components in the calibration circuit that are usually of low toleracnce, for instance instead of using a percision 1% or better resistor they will use a 10% tolerance etc. Often they have provisions inside the module for adjusting/calabrating but its not done at the factory because time is money. I doubt the compunurse has that, I never looked in one of those and havent used one. I use a lot of the dual temp indoor outdoor ones and they run in the $20 and up range.

use a styrofoam cup it really should not matter but your results astound me and I would hate to have to burn my physics book, Its old but it cost a fortune at the time

Anyway, again, please dont worry about it too much. oh, and major respect your way for doing it the old fashioned way, good show !

 

[reillyp]

I just bought a cheapie lcd + thermocouple at frozencpu for the external cooler I am building; I plan to either use part of a heat pipe from left over equipment (if the pipe is actually sealed - I looked at one and it was not) or scounge up a small piece of copper I can drill / dremel.

I plan to thermal paste the thermocouple into the hole, seal it (hopefully with some kind of sealant that I can undo if I want to change the setup later), and then epoxy the tube into a cap on my radiator - ie measure the radiator water temp. In the forums I keep seeing people say that there is only 1C difference or so throughout a usable loop, and my main interest is in temp differentials not, say, the water temp right before it goes into the cpu block...

 

[wwparrish]

That will work fine reilly, my comment would be that the "sheild" with AS you are putting around the temp probe will make the response time to temp changes slower. However as you mentioned temps in a watercooling system dont change much after the system comes up to heat and dont usually change rapidly anyway. The way you are doing it has some advantages in protecting the probe from the water. As I mentioned in an earlier post in this tread coating the temp sensor (and any wire lead exposed to water) with the clear coat nail polish (its a high grade plastic polymer) will seal it very well and allow you to put the probe directly in the water. It will simplify construction and react a tad faster to temp changes. In the end it will probally not make a lot of difference.

 

[reillyp]

Uh... girls have to replace their nail polish occaisonally. I really don't want to have to drain my system to touch up my nails...

 

[zer0signal667]

Uh... girls have to replace their nail polish occaisonally. I really don't want to have to drain my system to touch up my nails...

Girls' fingers see much more rigorous use than the tip of a thermistor in a watercooling loop