Water-Sealing a Circuit for Submersion: LM35 Temp. Sensor Tips

[bemaitea]

Hello all,

I've just completed a home made digital thermometer via:

http://sci-toys.com/scitoys/scitoys/computers/thermometer/thermometer.html

My problem is that I need the LM35 temp. sensor circuit to be water tight as it will be completely submerged in PROPANOL.

My initial idea was to coat the leads in clear nail polish, isolating the exposed leads coming from the sensor. But since it will be submerged in isopropanol (which is a pretty good solvent), I was worried that the nail polish would eventually wear off and short circuit the system.

Any suggesstions?

This is what needs to be water-sealed:

Thanks guys!

 

[MATLABdude]

Propanol (or isopropanol) isn't a very good conductor, nor does it easily ionize, and you might be able to just submerge the sensor in it. Pure (as in distilled or deionized) water also isn't a very good conductor, but it tends to leach stuff from whatever it's kept in. I'd suggest hot gluing the sensor and then reflowing the hot glue (with a heatgun or blow dryer) such that you get a smooth, continuous, and water-tight finish, but I suspect the plasticizers may leach out (which will probably cause it to crack / discolor). Actually, this might be a problem with the plasticizers in the IC case, as well. Not sure what would happen if you just surrounded the sensor with (a thin layer of) silicone.

You could try any of the above in a small container of propanol for a few days (at the operating temperature) and see what happens.

 

[bemaitea]

Thanks for the input Matlabdude...

Unfortunately I won't have access to hot glue, but a silicon sealer I do have. I also have a wood glue, though I am certain that will fall apart in the propanol after a relatively shorter period of time.

I should also mention that the temp. range that I will be using this in is between 0C and -50C, so it'll be pretty cold.

 

[Proton Soup]

i'd use epoxy. but if you encapsulate the whole thing, the thickness of the coating will slow the time response. also, there's an app note on that page.

http://sci-toys.com/scitoys/scitoys/computers/thermometer/how_the_LM34_works.pdf

the part about the metal can... having ground exposed is probably no big deal if your power source is something like a 9V battery. if you're using DC from rectified mains voltage, i'd think twice about it.

 

[bemaitea]

Thanks for the suggestion Proton Soup!

I completely neglected epoxy! I will def. try that out.

And Matlabdude, I coated left-over wire with some vairous glues I have and left them in 70%propanol over night; today they are completely dissolved. This won't do as I will actually have the circuit submerged in 100% propanol.

I also just submerged the entire exposed sensor in propanol to see if it woudl short. You were right, Matlabdude, the propanol was def. a worse conductor than water, but it still shorted the circuit regardless. Though I suspsect that is because I'm using only a 70% solution. I wonder if 100% in a sterile lab setting would perform similarly.

I'm going to ACE today and will pick up some silicon and epoxy and give them a try.

 

[berkeman]

One tip -- be sure to clean the circuit very well before potting it. If there are any contaminants (like solder resin) that get encapsulated with the circuit, you will get dendritic copper migration across surfaces that have to conductors with a bias across them. Like between the pins where they go into the TO-92 plastic package. Over time, this will lower the resistance between the pins, and eventually lead to a short.

You need to have a very clean process if you are going to encapsulate your electronics...

 

[Proton Soup]

^copper does that, too? i knew about zinc, but not copper.

 

[berkeman]

^copper does that, too? i knew about zinc, but not copper.

Yeah, most definitely. I worked on a new encapsulated product about 10 years ago where we found the migration problem in an accelerated life test that we do (85C at 85%RH for 1000 hours). Figuring out the issues was very difficult, but in the end we got it fixed with a combination of assembly cleanliness and a proprietary process improvement.

After getting that all figured out (the Internet and search engines were still young...), we did find some cases where others had run into the same issues. I just googled copper migration dendrites, and got plenty of informative hits. Here's one:

http://www.cheee.engr.tu.ac.th/document/TIJSAT%202008/2.pdf

.

 

[Proton Soup]

huh. learn something new everyday.

 

[davidrit]

Small diameter heatshrink tube over each wire/lead with a larger diameter heatshrink over top may work. You may have to melt the one end to get a good seal. for example 3M FP-301 is rated down to -55 C and "excellent" solvent resistance.

 

[bemaitea]

I'm currently testing out epoxy in a solution of 70% propanol to see if it will soften or degrade over a period of a day. Since it takes 24 hours to cure, I accelerated the curing process by gently warming the epoxy over a flame. I cooled it off and let it sit for additional 10 min at RT then submerged in alcohol. After 2 hours it seemed as though the entire mass is still structurally in tact, though the outer layer seems to be degrading and scrapes off with my nail. We'll see how it is in the morning.

Both a liquid "cement" and wood glue dissolved readily in the propanol, though the cement held its own for a bit longer but eventually gave in too.

I may have to go the heatshrink route, though I'm hesitant to bring a direct flame near the sensor, so I'm unsure about melting the tubing.

The use of propanol is killing my options.

 

[OmCheeto]

The use of propanol is killing my options.

Have you considered placing the apparatus inside of a copper tube?
Copper can be sealed easily and has high thermal conductivity.
You might be able to wrap some thick copper foil around the end of the LM34 and use the spring action of the foil to press against the inside of the tube. Or wrap it until it is nearly the same diameter as the tube and press it in tightly.

You may want to talk to a chemist as to about how the copper might react with the propanol.
Or I suppose you could just toss a penny in your jar. They are copper plated.

 

[Proton Soup]

maybe ask the chemistry guys

 

[bemaitea]

Yeah, I've posted in the chemical engineering section, hopefully someone there will have a solution.

I keep checking up on the epoxy, it actually is holding up well (much better than teh otehr options).

I'll give a crack at heatshrink and see what it does for the system...

 

[bemaitea]

Have you considered placing the apparatus inside of a copper tube?
Copper can be sealed easily and has high thermal conductivity.
You might be able to wrap some thick copper foil around the end of the LM34 and use the spring action of the foil to press against the inside of the tube. Or wrap it until it is nearly the same diameter as the tube and press it in tightly.

You may want to talk to a chemist as to about how the copper might react with the propanol.
Or I suppose you could just toss a penny in your jar. They are copper plated.

Hmm...assuming I insulate the leads with some electrical tape before wrapping in the copper foil, that could work. But i'd like the apparatus to be as flexible as possible, so the stiffnness that it would induce on the wire might be an issue.

Once I get in the lab tommorow I'll have some things to play around with.

 

[Phrak]

Epoxy may have an issue bonding to the wire insulation, especially teflon insulation. Are the exposed leads submerged?

 

[davidrit]

On second thought, for the heat shrink melting the end may not work well unless you get heat shrink tube that is specifically designed for that. As an alternative you could just place a small plug of plastic or metal that tubing could close around.

I wouldn't use an open flame, best would be hot air gun/hair dryer or else a soldering iron held near the tubing. If you use the sodering iron, shrink an area let the part cool, then shrink another area. The IC packaging is designed to withstand high temperatures for a few seconds during soldering so don't worry too much.

 

[bemaitea]

Epoxy may have an issue bonding to the wire insulation, especially teflon insulation. Are the exposed leads submerged?

Technically only the sensor would need to be submerged, but its small enough that just keeping the sensor and not the leads submerged would be near impossible. Even if I were to position the sensor so that only it is sumberged, the propanol bath will be spinning, and there is a great chance that the leads will get wet.

On second thought, for the heat shrink melting the end may not work well unless you get heat shrink tube that is specifically designed for that. As an alternative you could just place a small plug of plastic or metal that tubing could close around.

I think multiple heatshrink wraps may be the best method, and is looking more appealing after my epoxy results. Only problem is I've already soldered the circuit! (damn thinking ahead!). I'll probably need to get shrink wrap large enough to go over the sensor to avoid de-soldering.

 

[berkeman]

The use of propanol is killing my options.

How about a different approach? Could you use a non-contact IR temperature sensor, and either sense the liquid directly, or have an object submerged that you measure the temperature of through the propanol? (I don't know how IR transparent propanol is, however.)

http://www.harborfreight.com/cpi/ctaf/displayitem.taf?function=Search

.

 

[Phrak]

Berkeman reminds me, that all this has already been done.

Typically, a thermal probe consists of a spun metal tube, or a tube resealed at the end, with a sensor (usually RTD or thermocouple) in the end. It can be steel, stainless, aluminum, or whatever, depending upon application. Various packing material is used inside (glass beads?) to thermally connect the probe to the tube. The top of the tube is sealed to a fitting.


Here's the first example I found.
http://www.vernier.com/probes/tmp-bta.html" The picture is on the small side of the scale. Some can be up to 8 feet in length.

edit: Come to think about it, the packing material is used to fill the tube about the sensor, as a filler, to reduce convection currents inside the tube, perhaps.

The temperature read by the sensor depends upon a numer of factors including the thermal resistance of the lead wires, the thermal resistance of the tube from hot end to ambient, the thermal resistance of the packing material, and ambient itself. I believe the submersion depth is specified by the manufacture when measuring over an open vessel. Heck, even glass lab thermometers have submersion depth marked on the glass.

 

[bemaitea]

Hey all! I found a solution (though we'll see how long it lasts!)

I went over to ACE last night and just roamed the aisles for things I might be able to use. Eventaully I came across some vinyl tubing that they have in the back (similar to the stuff that covers a CAT5 cable).

Esentially what I did was slip the sensor through the tubing and had its head poking out of the bottom. The diameter of the tubing was slightly smaller than the sensor itself, so it makes a pretty nice tight fit. Then on the other open end of the tubing, I packed it full of epoxy to cap it and prevent any sort of liquid to crawl up the tube.

Presto, prelimiary tests show it works just fine, and not leakage! I'll try to take a few pics for you guys when I get a chance.

Thanks for everyone that helped out, i really appreciate all the input!

 

[bemaitea]

Ahhh! More problems!

So the good news is that after a day in the lab and testing out the probe, it is water tight and doesn't seem to degrade in the propanol in the time span that we use it.

Bad news is that this thing will NOT read any temp. below 0C!

I've been poring over the application notes for this little bugger, but my physics is a bit rusty so I wasn't able to pick up on everything.

This is what the circuit should look like for this sensor:

This (crude) diagram is what my setup looks like:
(Please excuse the lack of a proper circuit diagram! Have mercy on a biologist! )

Am I doing somethign wrong? I'm certain that's how it's suppose to be set up.

Any input would be greatly appreciated!

 

[Proton Soup]

read page five of National's application note. you need both a positive(+Vs) and negative(-Vs) supply for that circuit. figure 10 is a workaround for full range and only one power supply.

 

[bemaitea]

read page five of National's application note. you need both a positive(+Vs) and negative(-Vs) supply for that circuit. figure 10 is a workaround for full range and only one power supply.

Ahh yes I didn't catch that at first, thanks for pointing that out proton soup!

So without having to get those extra parts for the single supply, what could I use for a -Vs? Another 9V?

edit: I did a google search and found this: http://www.kpsec.freeuk.com/powersup.htm#dual

Is that what I need?

 

[Proton Soup]

yes, two 9V batteries connected as shown there, and the 9/0.000050=180k resistor

 

[bemaitea]

yes, two 9V batteries connected as shown there, and the 9/0.000050=180k resistor

Great! I've got the 180K resistor already, just need to get the second supply.

Thank you so much for the help proton! I'll give it a whirl and see what happens!

 

[bemaitea]

This project just can't get a break!

In trying to refit the system, one of the pins coming out of the sensor has snapped at the root. No soldering will save this tiny pin. I'll need to order a new one.

Hey Proton, I've drawn up this diagram for the new system. Does it look right in terms of figure 11 with the 2 supplies?

Thanks alot!

 

[MATLABdude]

This project just can't get a break!

In trying to refit the system, one of the pins coming out of the sensor has snapped at the root. No soldering will save this tiny pin. I'll need to order a new one.

Hey Proton, I've drawn up this diagram for the new system. Does it look right in terms of figure 11 with the 2 supplies?

Thanks alot!

Don't you need the ground line connected to either the -ve terminal on the left battery or the +ve one on the right?

 

[Phrak]

Remove one battery. Connect it between Vs and ground. The positive goes to Vs.

 

[Proton Soup]

GND should be connected to the node where the batteries come together, plus to minus.

and they're not likely to discharge at the same rate, fwiw.

 

[Phrak]

GND should be connected to the node where the batteries come together, plus to minus.

and they're not likely to discharge at the same rate, fwiw.

Good grief, your right Proton. I snafued again.

 

[Proton Soup]

geseundheit ;)

 

[bemaitea]

OK guys, so assuming I attach the GND from the sensor to the node connecting the two batteries, where should I attach the GND from the multimeter? To the same node?

And thanks a lot for the advice guys, you've been very helpful!

 

[Proton Soup]

yes, GND is the reference point, 0V

 

[bemaitea]

Thanks proton, I'm going to wait for the parts to arrive then reassemble everything.

I'll let you guys know how it turns out.