waterproof sensor ds18b20 is not waterproof

I bought this :

It was cheap only US$ 1.15. It worked just two hours underwater. :frowning:
AliExpress seller says he has tested the product. But speciferar only that he tested the electrical data.
I do not think he understands what Waterproof means.
Has anyone here in the forum had the same experience waterproof sensors. I bought three so I’ll try to make watertight .The leaked one had connecting wires got rust.
see attachment.

I guess it's a case of what do you expect for $1.85? I note with faint interest that the item in the link doesn't look much like the one in your picture. The examples in stainless tubes are probably reasonably waterproof but, if you want to get serious about it, you would still put them in a thermowell of some sort.

The picture I sent is after I opened it , when it stopped working. I’ll send one to you might understand

You spent very little money and got low quality. Perhaps for a lawyer it should be waterproof when it is sold as waterproof, but I think you got what you paid for.

For better quality, buy them at Adafruit (are those better ?), or make them yourself.

I understand you have pulled the sensor apart. I don't think you should have been able to do that but it seems that the $1.85 versions are not sealed into the tube. I guess you could now reconnect the sensor to the cable with heatshrink over the joints, put a dab of thermopaste in the bottom of the tube, reinstall the sensor, and then follow with some epoxy sealer.

You will then have a sensor as good as my $3.85 ones from eBay, or a $9.95 one from Adafruit. These should be quite satisfactory in a fish tank and mine test OK in glass of water, but, as I said, if you want to get serious, you go about the job in a serious manner. I have seen screw-in DS18b20 on eBay but they are rare and cost about $10

Look at the definition of waterproof for your watch. Most will say that it is suitable for use in the shower but not for swimming.

You get what you pay for.


Is the metal needed ? Does anyone know if ms-polymer is okay ? I want to do a project some day, and I'm thinking about spraying the soldered metal up to wires with fixating spray or lacker, and make a droplet shape of the whole thing with ms-polymer. Would that work ?

I don't think the metal is necessary, it's just the obvious choice, and I can't see why you would ever considering using something else - particularly at the price. The tube is robust, more thermally conductive than any polymer, and sized to ensure the minimum amount of less thermally conductive goo inside.

Mmmmm, I'm not convinced. The TO-92 is plastic, and extra 1 or 2mm polymer is just a little extra (extra isolation, that's true). The cheap one from Ebay has air isolation between the metal and the TO-92, and if water ever will get in, it is in contact with the wires. Did someone open the metal tube of the Adafruit sensor ? My conclusion: Make one yourself, or open the metal tube to see if it is any good.

I have to admit I haven't checked all fifteen of my sensors but, of those I have looked at, not one has air isolation between the TO92 and the case. All are filled with solid sealant, and it is a fair bet that the ones I have not looked at are the same. You can see the sealant when you slide back the cover between the cable and the tube. I have often tested sensors by doing the bleeding obvious, putting them in a glass of water, and never had a problem. An extra 1 or 2 mm of plastic may be an extra 100% cover over whatever does the work, and that can't possibly be a good idea, although it isn't as bad an idea as having air isolation. It appears that that is what the OP had, air isolation, which makes me think that he bought a factory reject and maybe the price confirms that.

Clear. Thank you.

I think it's a tradeoff between the thermal conductivity of the metal tube and its thermal mass. I've pulled temperature sensors out of those tubes and coated them with a thin layer of epoxy. The epoxy isn't a good conductor but there's a lot less of it to heat up. I've found the response time improves with this treatment. But it is more fragile.

I submit that anything significant there is as likely to depend on the starting gradient as anything else. Most of my sensors are in short tubes but I need some that are in long tubes. I can't tell any difference in their performance, but those in long tubes have twice the thermal mass. I guess this is because the sensor is down at the bottom of the tube, and most of the metal is essentially just going along for the ride.

I have used a lot of these temp sensors and never had any problem as far as being water proof. So you probably just got a bad one. However, since I use it in saltwater, the metal corrodes after about a year in saltwater. I have since switched to the one sold by sparkfun that does not use metal. it costs $10 last time I bought one.

Nick_Pyner: I submit that anything significant there is as likely to depend on the starting gradient as anything else.

Response time, as I meant it, is independent of gradient. It's the time it takes the sensor to change by some defined fraction of the sensor-environment difference.