Waterproofing

[FlyingSquirrel]

good point, reducing the amount of components that are fried is already a partial success
but id have to look into efficiently doing this, with the amount of connections that lead to the less power hungry parts
ie the flight controller usually has a minimum of 6 connections (not counting ground)

and how would i ideally go about calculating the remaining current, if lets say the main power shorts (through the water) to another component?
since, to reach a safe current i would have to drop from 140 to 1A, in extreme cases 0.5A (assuming a short without any resistance ofc)

[Ian.M]

You woudn't use a solenoid to disconnect -  that's a job for a fuse and crowbar circuit or for a electrically triggered pyro-fuse.  You could also use a MOSFET based disconnect, with gate circuit power drawn from the load side so that you have to connect an auxiliary battery to its output to reset it, on a potted or epoxy coated PCB so leakage cant cause it to reconnect.

If you cant disconnect the motor feed, you risk loosing all your ESCs to water damage - so solving the low mass high current disconnect problem is well worth doing.  Even a few tens of mA can cause significant board damage in under an hour - it only needs about 3000 Coulombs to electrolytically dissolve a gramme of copper, or slightly over one Coulomb per square mm of 1oz copper track, so anything other than a total disconnect is unacceptable.

[FlyingSquirrel]

hmm so if i understand you correctly, id use 5 (1 per ESC and 1 for the flight controller) N channel MOSFETs (ie IRF4104SPbF, or is this overspeced?)
and without a pull down, allow an initial impulse to power on the circuit/s
then common the gates and allow "intruding" water to create a short to the ground, which would power off the MOSFETs

what ive been wondering:
- could i also connect the gates to the carbon frame? (as far as i remember it has a resistance of roughly 30ohms)
- should an additional short exist to a power rail, wouldnt that basically short the battery? so id also have to add a 140+A fuse to the battery

and could i also use a P channel MOSFET (without pull up) instead? to power off when any short occurs and use a switch or reverse diode to discharge it once the short is cleared
or is the N type scenario more reliable?

[Ian.M]

Ideally, you'd want to cut all power as close to the battery as possible, so you'd have a single power cutoff module the main battery connector plugs into.   You'd also need to do the best you can to waterproof the battery itself, especially to prevent water penetration into the pack and to keep the balance connector dry.

I'd be reluctant to include the capability for the module to initially turn on on battery pack connection or turn back on without an external power source as it increases the risk that shock and immersion will cause it to fail to cut off or later to turn itself back on.   Having to apply external power to a pair of contacts to activate it isn't unreasonable  - just touch them with a 9V PP3 battery.   I suppose you *could* do something with a hall sensor and a magnet . . . .

[Assafl]

Reed switches is what's used in encapsulated electronics (even in heart pacers they use a reed switch). Cheaper and far more reliable than Hall Effect Switches (and don't need power to operate).

I was looking for high power reeds to see if it is possible for the reed to disconnect the battery - but none seem to be over 5A carry currents. There is a safety reed from telemechanique that seems to be 100A rating - but it is big and red and over 100$. Maybe it has a relay inside?

So it can control a very low Ron FET, or it may be possible to use a reed switch as a "use once" fuse (replaced every time it falls into the ocean). Even then, having that type of current through a 5A rated switch is too much (even for me).

Sealing techniques:
Another "sealing" technique, one favored by linesmen, is vaving an inverted "can" or "box", open only at the bottom. Sensitive electronics is inserted from the bottom, and cables  are arranged to be "scattered" so that wicking of water up cables is limited (or coated in that yucky hydrophobic grease). In your case, that would mean the copter should only crash upright (which I don't know if doable) and should never tip inverted.

The best sealing technique, that one we used most often in the navy, was waterproof aluminum boxes that had waterproof glands. We used Bendix/Amp really expensive glands, but there are cheap plastic ones available. Now the key to using such a box is that the gland has an Oring that is a perfect fit to the cable. Another Oring sealed the gland to the box (or it was a flat profile O-Ring). Sometimes, the connector itself was the gland (REALLY EXPENSIVE!!! - but may find NOS on ebay). We used specified lubrication to ensure the Oring was seated properly.

The trick is the cable. The cable has to be round. Not fairly round - absolutely round. It has to have a surface that is sealable like PVC, and more importantly, needs to be sealed at both terminated sides - or you must use a cable that sealed against water ingress. The latter cables are injected with hydrophobic grease that is substantially more annoying than silicon or petroleum grease. Otherwise water would wick up the cable into the box.   

As examples:
For systems like active sonar, the cables were a core of individual transducer HV pairs, each core sealed. The area between pairs was fabric strands with the yucky grease. this core was overmolded in real tough thermoplastic. Above that two alternating direction layers of steel cabling was wrapped (to carry the tons of weight of the transducer array). These were carefully terminated to large load carrying terminations.   

Smaller side scan sonars use a substantially smaller version of same, and newer generations put a lot of the electronics in the transducer. In an active SONAR that is impossible since the amplifiers are huge.

I don't think any of this is necessary. Just conformal coating and getting rid of the potential energy of the battery upon impact should be enough. BTW - What about the motors? If salt water gets into the motors I assume that would destroy the bearings?

[Ian.M]

Reed switches frequently don't survive hard impact schocks.  Even assuming the glass body is well enough protected and supported not to break, the reeds can bend leaving the switch permanently open or closed.

If the battery is disconnected quickly enough (seconds not minutes for salt water), everything else is salvagable if the motors and boards are immersed in clean fresh water as quickly as possible then cleaned and dried within 24H.  If you let anything dry without cleaning, the increased salt concentration will almost certainly FUBAR it

[Raj]

Silicone dip?

[Raj]

Has anyone ever tried using polyester resin - as in what is used for fiber-glassing and gel-coating boats, etc in a marine environment?

I've used it for surfboard repair for years and it's pretty easy to work with.  It seems like brushing a thin layer on a pcb might work well for a permanent, light weight waterproofing. Of course later repair or rework would would be a problem.

I don't know what it's electrical conductivity is but I'd think it would be very low. Thermal conductivity is likely high enough to allow adequate heat exchange with a thin layer.

Resin isn't shock resistant

[Ian.M]

Also Polyester resin is known to have adhesion problems to Epoxy.  Its OK for laminating but is a lousy encapsulant or adhesive.  Many marine DIY users and a few companies have found that out the hard way . . .    Epoxy resin would be a significantly better choice, but applying a thin enough coating without the use of dilutants that may compromise its water resistance would be difficult.

However, I see no reason to use anything other than a modern thin film hydrophobic coating.  The extra weight and thermal resistance to ambient are both negligible and as long as there's fume extraction, such coatings don't interfere with rework *EXCEPT* if epoxy PCB repair or other glueing is required.

A further thought on the disconnect:   If local laws permit amateur pyrotechnics, and the O.P. has access to a lathe, he could make an explosive disconnect similar to an explosive bolt.  When fired it would expel the mated male bullet connector in the battery cable  from its female end.  Construction would basically be: machine a M-F bullet connector extender out of brass rod, with a cavity in the center to hold an electric igniter, and some black powder sealed in place with a wad smeared with electrical grade silicone.     The igniter wires would be led out through small 45 deg angled holes in the side of it, sealed with a small blob of silicone inserted before the igniter and the whole assembly insulated with thin silicone tubing.  I would expect several to be damaged during development, but once the right powder charge weight has been found, the brass shell should be reusable - just clean and rebuild with a new igniter, powder etc.

[FlyingSquirrel]

ive ordered some appropriate MOSFETs and will try looking into trying to make an XT60 extension, that will cut the main power upon water contact
reg. the initialization with the 9V battery - couldnt an unlucky short also trigger the MOSFET? making sealing these terminals crucial

unluckily i have no control over the orientation in which the drone would crash, i usually do my best to have it upright when i anticipate a crash
but ive already had it in every other orientation too - since unlike the typical photography/video drones, these are meant to flip/roll/twerk in every direction (even fly upside down)

and i wont be able to legally get access to gunpowder here (maybe 1ce a year when they sell some fireworks), ive even failed purchasing general acids (H2S04, HCl etc) with more than 10% concentration (thats how knobby they are here)
but you gave me another idea - maybe i could use a 0.25W resistor as a backup for the MOSFET, designed to fry similar to a fuse

[FlyingSquirrel]

heres a little update to the waterproofing its self:

so far what i found to be the most robust solution for the ESCs is:
- conformal coating; ive done 2 layers of spray and 2 layers by dripping and spreading evenly, without brushing (later after soldering)
- self fusing tape; over the entire board and cables, with roughly 5mm of overlap (pressing it closed around the wires)
- more conformal coating; where the wires penetrate the self fusing tape (for additional safety)

this seems to work perfectly for the ESCs, since the wires all enter on one side and exit on the exact opposite side
but i doubt that this will work for the FC or PDB (without using countless layers of the self fusing tape), because theres wires and connectors in all directions (so still experimenting here)

what ive left out so far mainly for weight reasons, but could be added later (in case the above solution should eventually fail)
- o-rings around the individual wires (as Assafl mentioned, but i havent got for all the required gauges yet), to help seal them off
- steel/aluminum protective lacquer, which could be added around the tape to help protect against abrasion (be ware of conductivity!)

[Koen]

The hell are these answers ? There's solutions. It's a known and solved issue. Some products are even sold at major electronic distributors for very reasonable prices. Don't over-complicate it, don't try your own mixture of solutions. Look how the pros do it and copy it.

[FlyingSquirrel]

and how do the pros do it...?

note:
- weight and maintainability (so potting wont do)
- vibrations and high G-forces
- a lot of grime involved (dust, sand, dirt, mud, grass, branches you name it)

messing with the chemical integrity of the mentioned substances was never a subject

the only solutions i found in the sector so far are either unreliable, have to be refreshed very often, are not maintainable or they mainly consisted of marketing (without actual details)

ps: as mentioned, the best coating the major electronics distributor in the country recommended finally ended up cracking and delaminating

[CatalinaWOW]

I hope this all works well for you.  You might want to do some test flying where the parts are easily recoverable.  ESCs do tend to be marginal on cooling, and the multilayers of conformal coating haven't done any favors for heat transfer.  A lot will depend on your flying style, but the test flights will give you the best chance of learning at the lowest cost.

[Assafl]

so far what i found to be the most robust solution for the ESCs is:
- conformal coating; ive done 2 layers of spray and 2 layers by dripping and spreading evenly, without brushing (later after soldering)
- self fusing tape; over the entire board and cables, with roughly 5mm of overlap (pressing it closed around the wires)
- more conformal coating; where the wires penetrate the self fusing tape (for additional safety)

If you leave wiring exposed - it will wick (capillary) water under the self amalgamating tape. The tape will then prevent the water from evaporating - causing more damage.

You can try to dip the ESC in water and measure resistance - you will find that it rakes it no time to get in.

As a rule: 99% waterproof is worse than no waterproofing at all (water resistance being different than waterproofing). Because water will eventually get in - and the lagging will make it stay.

[Ian.M]

Which is why you either have to start with a brand new freshly cleaned board, use a high adhesion conformal coating, and pull a vacuum on it then release before its started to cure to get good penetration under components to totally seal all exposed metal apart from the off-board connections,  or use a hydrophobic nanofilm surface treatment so you don't get large blobs of water on the board and capillary action doesn't wick it under parts.

Wrapping it in *anything* other than a foiled pouch hermetically heatsealed round the wiring, containing a dessicant sachet, will make the problem worse.  Failure to keep the boards clean will make the problem worse.   Unless you fly in rain or fog or over beaches, most of the dirt should be dry and easily cleaned with a soft brush and a vacuum.  Don't let it build up, especially if rain is expected.

If you've been flying at sea coasts near salt spray, you've got a major problem.  As the salt dries, it forms localised pockets of extremely concentrated brine which can cause rapid corrosion anywhere metal is exposed.   To prevent problems, either your electronics needs to either be well enough sealed to survive a hoze-down while running without any water penetration (a lot of extra mass) or you need to strip the craft and wash all the boards and motors as soon as possible after the flight then, if there's any sign that the hydrophobic coating is under-performing, wash with IPA, dry thoroughly and reapply the coating.