Introduction: Waterproof a LM35 Temperature Sensor

Here is a instructable to waterproof a LM35 for use on a tethered ROV using a automobile 12V battery as a power source. This came out of a need for the MATE ROV Competition. The LM35 series are precision integrated-circuit temperature sensors, whose output voltage is linearly proportional to the Celsius (Centigrade) temperature. Here is the datasheet -

Step 1: Cat-5 Used for the Tether

I used cat-5 cable for the sensor as it will shield the sensor signals on the twisted pair (and I had a spool sitting around.) It will also easily attach to the tether.

I used six wires (three pair) for the sensor (Vs+, Vout, Gnd). I twisted colored pairs together to give me the three conductors.

Step 2: Solder the LM35

Before soldering, I added short lengths of heat-shrink tubing wrap to the conductors, and two larger ones over the cat-5. Then, I soldered the LM35 to the cat-5 cable and added a 10K resistor to the Vout (center) to stablize the signal for the length of cable.

I then shrunk the wrap over the soldered joints using a torch (on it's lowest possible setting.)

Step 3: Copper Cladding

I used 1/4 inch O.D. copper tubing (the kind for an automatic ice maker for your refrigerator) to make a small copper cover for the LM35.

I was able to just squeeze the LM35 into the copper tubing, then I squished the tubing just above the sensor in the vise...being very careful not to squish the sensor.

Step 4: Operational Sensor...but Not Waterproof

At this point, the sensor is fully operational in the air, but it is not waterproof. I decided to give it a quick test to make sure everything still works.

That's 16.7 mV (16.7 C or 62 F) on the voltmeter. Everything seems to be working.

Step 5: Capping the Probe

I wanted to have the sensor on a probe so it would stick out forward of the ROV a little. Here I'm building the cap out of 1/2 inch copper pipe and cap. I drilled a 3/8 inch hole in the cap to accept the sensor with the squished copper cladding on it.

Step 6: Silicone Sealant to Waterproof

The copper cap was hammered into 1/2 inch PVC pipe to make the probe. Then, I used aquarium, silicone sealant to seal the gaps as I pushed the sensor into the probe.

The entire length of the probe was filled with sealant to complete the waterproofing. Note - However, you should probably fill it with a non-conductive grease. Then, clean the end of any grease and seal it with the silicone sealant. The sealant will never set if it is not exposed to air.


paulbuzz made it!(author)2016-05-31

What I actually did was remove the wires from inside the outer insulation, soldered them to the legs of the LM35 heatshrinking then carefully feeding the three back into the outer insulation.
The LM35 just visible.
Then I applied the heatshrink over the LM35 and cable and as I said, squashing the heatshrink with pliers whilst it was still hot.

I no longer have an aquarium but I understand the importance of not having metals inside the tank, especially sea water.
This way surely avoids this most especially copper being a big no no, most especially where invertebrates are present.

solarmax made it!(author)2016-05-31

I didn't make this for an aquarium. This was for an underwater robot competition about 8 years ago. I agree that this would not be good for an aquarium.

paulbuzz made it!(author)2016-05-31

I see, I got the impression some had thought about using this for an aquarium.
Your build, then, actually inspired me to add this for those interested in monitoring their aquarium. :)

paulbuzz made it!(author)2016-05-24

I would not use metal inside the aquarium at all if possible.
I suggest using the larger diameter heat shrink over the LM35. Heating it and shrinking then while the heatshrink is still hot Pinch it with a pair of pliers.
In my example the probe is not immersed in water but the length of the heat shrink could be longer to ride higher up the cable. I would also add a small cable tie to better seal the heatshrink at the other end of the cable, also while warm.

Greywolfg made it!(author)2016-05-30

I agree. I like the heatshrink method better. But good instructable, solarmax.

solarmax made it!(author)2016-05-30

Thanks. I completely agree. If I were to do this again, I would use the hot-glue, shrink-wrap method.

Greywolfg made it!(author)2016-05-31


FieldingBlue made it!(author)2013-06-01

I like this. I've thinking about it for my fish tank for a while. I had not thought about cladding in copper though.
May I suggest that some heat sink compound added to the LM35 just as you insert it into the copper pipe would give an improved temperature transfer. Just a thought. Thanks for sharing.

NinjaCrow made it!(author)2013-07-04

Not sure I would go with copper if you are going to use it in a fish tank. I'm not sure how tolerant freshwater plants/fish are but its probably not the best thing for them, I know it can be tank ending with saltwater. Since copper has a tendency to kill nearly all invertebrates save horseshoe crabs. Well I assume horseshoe crabs don't have a problem with it since their blood is copper based.

FieldingBlue made it!(author)2013-07-04

Actually that's a really good point. Maybe stainless steel but that's no where as easy to work with, well at least from the perspective of a electronics tech. It certainly explains why I cannot ever seem to keep fish healthy for any great length of time. :)

mr+fat made it!(author)2012-08-02

Great idea! Just one thing the sealant will eventually cure inside the tube without exposure to the direct air, I have had hundreds of silicone tubes go off over the years of my work even un-opened tubes can go Hardin the right conditions

zhenglu891028 made it!(author)2012-04-08

Cool! Very Nice!

qark123 made it!(author)2008-05-30

Hi, would you please explain the function of the stabilizing resistor in more detail? I understand that there is no current flow through it, so it doesn't change the temperature value, but what does it do?

solarmax made it!(author)2008-06-16

Check out the datasheet at . Figure 2 shows the resistor from Vout to Ground.

JohnTom made it!(author)2010-08-09

Fig 2. shows the resistor connected from Vout to -Vs. This is not Ground, but a negative bias voltage to allow you to use the full range of -55C to +150C Your probe will measure +2C to +150C The resistor is not necessary in your circuit.

etscheid made it!(author)2011-11-22

solarmax, I think you meant Figure 3 of the LM35.pdf (& series, like you have it shown in the pic). JohnTom, I believe the series resistor is good to have there.

qark123, if you do DC analysis, there's practically 0 current. But, there is current in transient analysis. The problem is that a long cable can have a lot of capacitance, and when the little lm35 needs to change the Vout, it has to charge or discharge that capacitance to change the voltage. This series resistor will help limit how much current it drives at once. Another other (& maybe more important?) reason for an external resistance (instead of the device just having a larger output resistance of it's own), is in case there's any transient reflections from driving into a transmission line like a cable. The reflected voltages can cause a potentially harmful (to the lm35 or receiving device) voltage level. Finally, like solarmax referred to by saying 'stabilize', since the R slows the driver down, avoiding/minimizing reflections in the cable, the sampling won't see an incorrect voltage due to reflections either.

Anyway, you can see that avoiding transmission line effects is a good thing. The net effect is that a larger R like this will make it a tiny bit slower for a new Vout voltage to show up since it's being slowed by the resistor. But, you'd probably have to have a bigger R &/or much bigger capacitive load before the slower responsiveness was noticeable.

dr.knowitall made it!(author)2011-08-06

actually, 16.7mV would result in 1.67C thus giving us a reading of 35F.

you convert mV to V by dividing mV by 1000. which would give us .0167V.

the LM35 is .01V/C, so we then divide our .0167V by the .01 and we get 1.67V/C then you can convert to F from there.

skillett made it!(author)2008-04-16

maxim/dallas has way cool CHEAP parasite power temp sensors in TO-92 cases and you can get FREE samples. link:

They also have a bunch of software to read from them. they can be tagged with XML. if needed. you can put a bunch of them on a wire and they can tell temps for each one due to a 64 bit serial number. you can cobble together a serial interface for a couple of bucks instructions here:

there is a opensource app called digitemp for/linux/winDOZE
link here :


daleth made it!(author)2011-04-06

Wow—these look absolutely perfect. Since 2.5 years have elapsed since this comment, is there an Arduino library for these yet? ;)

solarmax made it!(author)2008-04-16

Thanks! I'll check those out.

skillett made it!(author)2008-04-16

PS. The DS18S20 can be programmed to set off an alarm at a specific temp too. It does centigrade AND farenhight

mooseo made it!(author)2010-03-04

This is definitely slick, but may be overkill in many situations.

I've had good luck just sticking them in adhesive-lined heat shrink tubing. Heating that will seal around the cable and then I crimp the outer end of the tubing with a pair of pliers while it is still soft. The plastic won't have the same temperature conductivity as the copper, but if you aren't looking for fast response, it's OK.

Another thing I've done is use the 8-pin SMT version of the LM35 and for the resistor. They end up being a smaller package that slips into the heat shrink. Don't drink coffee before you solder up the leads.

handidad made it!(author)2008-10-14

An easy method of sealing the squished end of the 1/4" copper tube would be to solder it. But do this before inserting the temp sensor.

LinuxH4x0r made it!(author)2008-04-13

Could you give more info about the sensor and how to use it with a voltmeter? Thanks Btw, great ible

solarmax made it!(author)2008-04-14

The LM35 (and LM34 for F temps) takes a voltage in (from 4Vdc to 30Vdc) on the on the left lead (see the photo above) and ground on the right lead. Then, it adjusts the output voltage of the center lead (referenced to ground) to whatever the temperature is (10.0 mV/°C) - from 0C to 50C in the case of this little guy. Very ease to work with...but, I'm glad I bought extras. I broke one of the leads while manipulating the sensor during this procedure. Best thing is the voltage input range of this can be a button-cell or a car battery...output is consistent.

BTW, some guys have taken the output of this sensor (10.0 mV/°C) and multiplied it (using a LM324 op-amp) to give a 0-5V output. Then taken that and plugged it into an analog input of a robot controller. (.pdf)

LinuxH4x0r made it!(author)2008-04-14

Thanks for the info. I might buy some of these.

solarmax made it!(author)2008-04-14

You're welcome.
BTW - $1.69 USD at Jameco -

LinuxH4x0r made it!(author)2008-04-14

I might just get some free ones from ti Thanks though

joejoerowley made it!(author)2008-04-13

Cool! Very Nice!

Ora made it!(author)2008-04-13

I had no idea that there were such simple temperature sensors, this a very well documented instructable, and has inspired me to use these in an upcoming project or two.

About This Instructable




More by solarmax:Flower DOTKLOKAdding a dimmer to your DOTKLOKWaterproof a LM35 Temperature Sensor
Add instructable to: