> there is a Raspberry Pi that has been running underwater in the lobby of HZO’s building for 525 days and counting. [...]
> We’re not sure which treatment our tiny green computer got — HZO produces a number of specialised coatings. Perhaps the Raspberry Pi got a parylene coating, as that seems to be a highly water-resistant and submersible option. They also offer plasma-applied coatings that are apparently a good low-cost option if you’re not going to be chucking your tech into the deep on a regular basis
How so? The contacts themselves are probably fine without the coating, and which other parts would plugging and unplugging a cable into a port (re)expose?
A conformal coating is like putting a circuit board in a plastic bag. In this application they applied it around the connectors. If you remove the connectors, the seal is broken. If you break the seal and submerge the board in water you will have a short.
Usually in these types of applications, you use DEUTSCH connectors https://www.te.com/usa-en/products/brands/deutsch.html?tab=p... that have an o-ring to seal the connector and allow it to be exposed to the elements and provides a seal that also allows removing the connector.
How would you do regenerative braking train couplers between engines and battery cars? (That at present have just an air hose to hold brakes open per Westinghouse, but no power wiring or data cabling or a spec yet)
Would standard EV charger specs work for coupling battery train cars to engines and/or regenerative braking cars?
Maybe at least 1 Gbps and PoE+ to support a mesh network for sensors?
Presumably hyperloop teams have solutions for this?
The problem with electronics in water is that water is an electrical conductor.
So the circuit diagram for the board becomes "everything connected to everything else all at once". Which means, at best it doesn't turn on. At worst it creates a short and current flows from one point to another, usually blowing up some part.
Electronics and water therefore are seldom friends.
Distilled water does not conduct electricity. It is the dissolved ions in the water that allow electricity to conduct. You may be able to clean the PI to such a level that nothing dissolves into the water when it is submerged.
AFAIK electrolysis will force you to utilize a scrubbing system anyways once you've got current passing through the water.
But getting conductivity acceptably low is much less demanding than getting it this extremely low.
E.g. some CO2 lasers are cooled with purified water to limit the current through it.
Even ultrapure water conducts electricity. It says right here [1] that its resistivity is 1.82×10^9 Ohm * m.
This is a pet peeve of mine. We get told of conductors and insulators in school, but it doesn't work like that. Pretty much everything except a perfect vacuum conducts electricity. It's just that some materials conduct electricity so poorly that it can be rounded off to 0 in most circumstances, but there can still be measurable current flow.
Pure (eg. Distilled) water isn't a very good conductor at all. It's only when you add contaminant ions that it starts to conduct electricity in any meaningful sense.
Most people's experience with water rapidly destroying electronics is probably with highly ionised water such as sea water or water in the washing machine which has washing powder added.
Of course, it's worth pointing out in this case that as per the comments HZO claim they used water straight from the tap, so it's probably still an impressive enough achievement.
I'm curious how they went about sealing the external connectors (usb, hdmi), because surely those would need to be exposed and with the voltage present would act as points of corrosion. The pins on the Pi connectors are quite tightly spaced too. Maybe the AC nature of most of the signals helps there?
Electronic are dirty, your keyboard has a lot of junk, even salt from chips and pretzels, sugar from sticky fingers, etc. Spilling even distilled water on your keyboard dissolves all of that and makes the water conductive. Luckily some better laptops have channels (drip holes) for water to drain without affecting the board, but relying on bad conductivity of distilled wated in real-life scenarios is really not safe :)
I considered digressing into this in my post, but while completely true, I concluded its also irrelevant to the point being discussed. Even tap water has sufficient contaminants to make it an excellent conductor.
Outside of very clean, very lab, conditions, water conducts really well.
But your point is a nice aside, and I'm glad you brought it up.
Incidentally -drinking- pure distilled water tastes funny.
Plus even if you use distilled water, it'll probably find something to disolve a bit of if you keep in on your electronics for long and then it'll conduct a lot better.
Once you have your ultra pure water, and are willing to commit to maintaining the apparatus that keeps it that way. There is a second problem, extremely pure water is surprisingly corrosive, A laymans explanation is that it would be happier with a few ions, and will gleefuly strip them off any exposed metal.
Which is a shame. It would be nice if water could be used directly in an immersion environment. Much better thermal properties than anything else, and no mess,
There are a number of conformal coatings that can be used to keep a circuit board dry at fish tank pressures. Parylene is my favorite as it isn't too harsh on the board materials and coats even underneath the surface mount components. Doesn't usually stick to the board though so it's more of a perfectly sealed bag around the board. It's good for basically forever.
If you unplug the connectors you rip a hole in the "bag" though and the connectors will corrode. The way around that is to use expensive hermetically sealed connectors that are designed to operate underwater. Each side of those will cost a few hundred USD for boring round connectors. I have never tried to run e.g. HDMI through a bulkhead but such a connector would probably be impressively expensive.
Anyway the best thing about Parylene is you have to send your equipment out to have it applied, no do it yourself here. The spray on coatings are usually sticky, messy, and not that great overall in comparison.
>The way around that is to use expensive hermetically sealed connectors that are designed to operate underwater. Each side of those will cost a few hundred USD for boring round connectors.
In this respect, I highly recommend Cobalt series of connectors from Blue Trail Engineering: https://www.bluetrailengineering.com/professional-products - they cost "just" tens of dollars each and are very good (I used them); quite popular with ROV community.
As a small note, IP67 sealed connectors vary greatly in price, with excellent high pin count options like the Ecomate Aquarius line being in the low tens of dollars. The Deutsch/Amphenol AT automotive connectors are also very cheap yet provide a watertight seal.
I've seen some fairly generic looking many-pin connectors that are water tight. Is it just best practice to route everything through the 40+ pins and break out on the other side? Even power?
Is it possible one could coat electronics like this, then submerge them in (distilled maybe) water within an ice pack, freeze them, then use them and the heat melts the ice?
Maybe it could instead be within some type of very thin pouch/layer of dielectric liquid, like in immersion cooling, and then water/ice outside of that so the force of expansion from freezing could be distributed?
I’m not sure I’m following the end game here with freezing it. From what I recall ice is actually a pretty decent thermal insulator and since a given mass of ice is larger in volume than the same mass of water, you’d end up with a hole melting inside with… weird things happening as far as pressures go.
If the goal is to keep the processor nicely chilled, you’d probably be better off running liquid water over/through the block of ice and then keeping the coated board chilled in the ice water bath. If you’re trying to get below 0C, you should probably keep using a liquid with a lower freezing point instead of having the board frozen in a block.
> I’m not sure I’m following the end game here with freezing it.
My kids playing on the Mega VR headset and complaining about it getting hot is one example of my line of thought. Let’s say you have to absorb the heat generated by a given amount of electricity but want to do it in the most weight efficient way possible.
It turns out that regular old water has basically the highest latent heat of fusion at room temperature and pressure, which means a block of ice can absorb more heat by melting than pretty much anything else gram for gram.
What if the VR headset could just freeze itself overnight and then no fans or reliance on dissipating heat into the external environment but instead into the ice.
It’s probably stupid but I do know water is the best know molecule in terms of latent heat of fusion per gram at room temperature and pressure.
In water, thr cooling factor goes way up. Liquids are much better at absorbing heat than gasses are.
Allowing for the fact that you can cool down the water, and circulate it, means you can overclock the device a lot, without heat becoming a factor.
I don't know thd specifics of the coating, but its -unlikely- that it would change the thermal properties to any degree, and any difference would be dwarfed by the benefit of immersing in a liquid.
I meant in air. I made a while ago an Unmanned underwater vehicles (UUV), the on-board SBC (nvidia jetson), despite having a waterproof case I still did have heat shrink tube on it and the connectors as an extra measure, obviously it’s not as sophisticated as the OP coating tech, but the there were an around 45% thermal increase, would be interesting to see some benchmarks numbers on that!
My guess is that the effect in air is minimal, because the coating is so thin.
If the coating had impressive thermal properties (such a steep thermal gradient) , even at such tiny thickness, it would have other applications (where thermal insulation is desired.)
- what kind of heat shrink was it? There’s the usual thin plastic type and the gooey double-layer type. I’m actually not sure which one would be worse for heat… the gooey type seems like it’d be a thermal insulator but maybe with that much close contact it might be a decent thermal conductor. The regular thin plastic type would probably have been trapping a bunch of air around the heatsink (dramatically reducing the surface area for convection)
- how big was the waterproof enclosure? The stagnant air inside there seems like it would probably heat up pretty well. If the enclosure was plastic it would probably conduct that heat into the water quite poorly.
If you end up in a similar situation down the road, you’re probably way better off making an aluminum plate with a raised pad that you can mount the CPU to directly and use a gasket to allow the outside of the aluminum to conduct the heat directly into the water. Assuming you’re not going to crazy depths I’d expect that to be viable and you’ll probably get “better than forced air” performance out of it
The ones on the plugs/connectors were the usual ones, the one I put on the SBC was thick as I couldn’t find a thin big one, and I was afraid to damage the SBC while applying the heat, so maybe it was the double layer type? But I don’t know.
>how big was the waterproof enclosure?
It was big enough to house the SBC (jetson Nano) and some other components needed for the UUV, not exact size but probably ~20x15cm.
> better off making an aluminum plate with a raised pad that you can mount the CPU to directly and use a gasket to allow the outside of the aluminum to conduct the heat directly into the water
That’s actually a really smart idea, thanks!! I did have a heatsink (those small ones fit the cpu size) but it was under that heat shrink tube so probably made things worse.
> Assuming you’re not going to crazy depths
We didn’t exceed 50m depth, still I was concerned the pressure might leak some water into the enclosure and this I used the heat shrink.
53 comments
[ 3.6 ms ] story [ 110 ms ] thread> We’re not sure which treatment our tiny green computer got — HZO produces a number of specialised coatings. Perhaps the Raspberry Pi got a parylene coating, as that seems to be a highly water-resistant and submersible option. They also offer plasma-applied coatings that are apparently a good low-cost option if you’re not going to be chucking your tech into the deep on a regular basis
It’s a gen 4 board which is only a few years old. Did they do this previously with older pi’s or was that a mistake in the article?
Usually in these types of applications, you use DEUTSCH connectors https://www.te.com/usa-en/products/brands/deutsch.html?tab=p... that have an o-ring to seal the connector and allow it to be exposed to the elements and provides a seal that also allows removing the connector.
Would standard EV charger specs work for coupling battery train cars to engines and/or regenerative braking cars?
Maybe at least 1 Gbps and PoE+ to support a mesh network for sensors?
Presumably hyperloop teams have solutions for this?
So the circuit diagram for the board becomes "everything connected to everything else all at once". Which means, at best it doesn't turn on. At worst it creates a short and current flows from one point to another, usually blowing up some part.
Electronics and water therefore are seldom friends.
https://techiescientist.com/does-distilled-water-conduct-ele...
This is a pet peeve of mine. We get told of conductors and insulators in school, but it doesn't work like that. Pretty much everything except a perfect vacuum conducts electricity. It's just that some materials conduct electricity so poorly that it can be rounded off to 0 in most circumstances, but there can still be measurable current flow.
[1] https://en.wikipedia.org/wiki/Electrical_resistivity_and_con...
So no, you shouldn't really consider distilled water a conductor.
Most people's experience with water rapidly destroying electronics is probably with highly ionised water such as sea water or water in the washing machine which has washing powder added.
Of course, it's worth pointing out in this case that as per the comments HZO claim they used water straight from the tap, so it's probably still an impressive enough achievement.
I'm curious how they went about sealing the external connectors (usb, hdmi), because surely those would need to be exposed and with the voltage present would act as points of corrosion. The pins on the Pi connectors are quite tightly spaced too. Maybe the AC nature of most of the signals helps there?
Outside of very clean, very lab, conditions, water conducts really well.
But your point is a nice aside, and I'm glad you brought it up.
Incidentally -drinking- pure distilled water tastes funny.
Which is a shame. It would be nice if water could be used directly in an immersion environment. Much better thermal properties than anything else, and no mess,
If you unplug the connectors you rip a hole in the "bag" though and the connectors will corrode. The way around that is to use expensive hermetically sealed connectors that are designed to operate underwater. Each side of those will cost a few hundred USD for boring round connectors. I have never tried to run e.g. HDMI through a bulkhead but such a connector would probably be impressively expensive.
Anyway the best thing about Parylene is you have to send your equipment out to have it applied, no do it yourself here. The spray on coatings are usually sticky, messy, and not that great overall in comparison.
In this respect, I highly recommend Cobalt series of connectors from Blue Trail Engineering: https://www.bluetrailengineering.com/professional-products - they cost "just" tens of dollars each and are very good (I used them); quite popular with ROV community.
My apologies for an acronym. It's Remote operated underwater vehicle: https://en.wikipedia.org/wiki/Remotely_operated_underwater_v... (a small submarine, usually, with a long thin tether)
Could you share some vendors and pricing since it sounds like this is something you are experienced at?
You're very likely to destroy it over time, if not immediately.
If the goal is to keep the processor nicely chilled, you’d probably be better off running liquid water over/through the block of ice and then keeping the coated board chilled in the ice water bath. If you’re trying to get below 0C, you should probably keep using a liquid with a lower freezing point instead of having the board frozen in a block.
My kids playing on the Mega VR headset and complaining about it getting hot is one example of my line of thought. Let’s say you have to absorb the heat generated by a given amount of electricity but want to do it in the most weight efficient way possible.
It turns out that regular old water has basically the highest latent heat of fusion at room temperature and pressure, which means a block of ice can absorb more heat by melting than pretty much anything else gram for gram.
What if the VR headset could just freeze itself overnight and then no fans or reliance on dissipating heat into the external environment but instead into the ice.
It’s probably stupid but I do know water is the best know molecule in terms of latent heat of fusion per gram at room temperature and pressure.
In water, thr cooling factor goes way up. Liquids are much better at absorbing heat than gasses are.
Allowing for the fact that you can cool down the water, and circulate it, means you can overclock the device a lot, without heat becoming a factor.
I don't know thd specifics of the coating, but its -unlikely- that it would change the thermal properties to any degree, and any difference would be dwarfed by the benefit of immersing in a liquid.
My guess is that the effect in air is minimal, because the coating is so thin.
If the coating had impressive thermal properties (such a steep thermal gradient) , even at such tiny thickness, it would have other applications (where thermal insulation is desired.)
- what kind of heat shrink was it? There’s the usual thin plastic type and the gooey double-layer type. I’m actually not sure which one would be worse for heat… the gooey type seems like it’d be a thermal insulator but maybe with that much close contact it might be a decent thermal conductor. The regular thin plastic type would probably have been trapping a bunch of air around the heatsink (dramatically reducing the surface area for convection)
- how big was the waterproof enclosure? The stagnant air inside there seems like it would probably heat up pretty well. If the enclosure was plastic it would probably conduct that heat into the water quite poorly.
If you end up in a similar situation down the road, you’re probably way better off making an aluminum plate with a raised pad that you can mount the CPU to directly and use a gasket to allow the outside of the aluminum to conduct the heat directly into the water. Assuming you’re not going to crazy depths I’d expect that to be viable and you’ll probably get “better than forced air” performance out of it
The ones on the plugs/connectors were the usual ones, the one I put on the SBC was thick as I couldn’t find a thin big one, and I was afraid to damage the SBC while applying the heat, so maybe it was the double layer type? But I don’t know.
>how big was the waterproof enclosure?
It was big enough to house the SBC (jetson Nano) and some other components needed for the UUV, not exact size but probably ~20x15cm.
> better off making an aluminum plate with a raised pad that you can mount the CPU to directly and use a gasket to allow the outside of the aluminum to conduct the heat directly into the water
That’s actually a really smart idea, thanks!! I did have a heatsink (those small ones fit the cpu size) but it was under that heat shrink tube so probably made things worse.
> Assuming you’re not going to crazy depths
We didn’t exceed 50m depth, still I was concerned the pressure might leak some water into the enclosure and this I used the heat shrink.
https://en.m.wikipedia.org/wiki/Safety_data_sheet