Potting Electronics

Thread Starter

scubasteve_911

Joined Dec 27, 2007
1,203
Hi all,

I am working on an autonomous underwater vehicle project, which isn't for work. I am using an Intel Atom motherboard for all my computing needs, a picture is located here http://www.pressebox.de/attachment/113306/Intel_D945GCLF_Press.jpg

The problem is, how can I keep this board working, even with 5000PSI of pressure? I was considering getting an aluminum box, then standing the PCB off, then potting it. But, I need to add heatsinks to some ICs and have the aluminum go out into the water. This is because the epoxy is a good insulator of heat, even the thermally conductive epoxies are poor. If the heat doesn't escape, then it will create pressure pockets and crack the components/resin.

Another option I thought of is to get a high strength aluminum container with a fluid such as mineral oil to insulate, yet conduct heat.

I know that you should select a low durometer potting compound for potting electronics, since it will not stress the components much. But, in my case, I need a fair amount of compressive strength, so I need to go with a harder resin (about 75 shore D).

I am removing as much as possible from the PCB, since there are some unnecessary connectors, etc. All I am using is one hard drive, a ram module (pain in the butt, since it is tall!), and several USB ports. I am even considering replacing the aluminum caps with tantalums, since it's a solid dielectric that can withstand a bit more pressure.

Any suggestions?

Steve
 

SgtWookie

Joined Jul 17, 2007
22,230
I see two crystal cans on the board. Those are mostly air inside. They'll be smashed flat like toothpaste tubes with your mineral oil solution.

The board itself is problematic; the card edges and even the vias aren't a hermetic seal. Water or fluid under 5,000 PSI pressure will force it's way between the PCB layers, and you will very shortly have a dead PCB. Even ICs have voids in the plastic cases. The water or fluid will find it's way in.

What you really need is a hermetically sealed pressure vessel. That's going to be ridiculously expensive.
 

Thread Starter

scubasteve_911

Joined Dec 27, 2007
1,203
Wookie,

Thank you for your time and expertise,

I was considering potting the entire PCB with an epoxy resin, not dipping it. The epoxy would act as a barrier for the pressure. Basically, the pressure will try its best to compress the entire brick. The pressure will induce strain on the material, then deform slightly according to the bulk modulus. What matters is that the compressive strength is not exceeded and the pressure distribution along the components. This is what I am uncertain about. Is there some drop of pressure as it is transmitted through the material? Kind of like a resistive drop to a potential?

I am clearly an electrical trying to do a mechanical's job.. Can anyone help?

Steve
 

SgtWookie

Joined Jul 17, 2007
22,230
Well, the vertical compression is one thing.
This search:
http://www.google.com/search?hl=en&q=FR-4+strength
got a lot of hits.

The horizontal compression is another aspect. I'm afraid that with 5,000 PSI exerted from both ends and both sides, the "brick" approach would exert a tremendous shear load on the SMT components unless the epoxy used had the exact same characteristics of the epoxy used in the board.
 

John Luciani

Joined Apr 3, 2007
475
At Vicor the DC-DC converters were potted in a hard epoxy for thermal conduction as well as rigidity (they have since switched to a thermally conductive silicone compound).

When the rigid epoxy was used sensitive components were conformal coated with a silicone compound (I do not recall the mfg). The conformal coating reduced the stress as the rigid encapsulant expanded and contracted. IIRC the rigid epoxy was made by Castall. I
do not remember the mfg of the silicone compound.

What ever solution you decide on I would recommend temperature cycling your
assemblies to qualify the process.

(* jcl *)
 

Thread Starter

scubasteve_911

Joined Dec 27, 2007
1,203
John,

Thank you! I was actually considering dipping the entire PCB in a more flexible resin (shore A 80 ish), then potting it. Conformal coating seems like it can have the same effect, but I suppose I will need to drench it to get in all the nooks and crannys.

I will research the company and post back with results. Thanks again!

Steve
 

Thread Starter

scubasteve_911

Joined Dec 27, 2007
1,203
Thanks again guys! I will certainly post my success/failures so that people can learn from them.

When doing projects, I plan a couple months ahead (or at least try to) to get rid of 'show stoppers', so I won't be posting for a while.

Steve
 

Externet

Joined Nov 29, 2005
2,200
Hello Steve.
Yesterday I finished a simple gadget that will work at much less depth, and used wax. Other times I have used silicone.

For high pressure; and if the size of that board allows; I would explore cutting a shorty diving tank in two, lathe thread male and female ends and fit a o-ring. Unsure if the i. d. will fit your board. A compression fitting at the neck will allow cabling, and no need for potting.

For sure room for hard drives, batteries and more...
Just an alternative to consider.

Miguel
 

Thread Starter

scubasteve_911

Joined Dec 27, 2007
1,203
Miguel,

Very interesting idea, I have considered something similar. In my particular project, I must minimize all drag sources, which really limits what I can do. If I didn't have this restriction, I would certainly attempt to make a pressure vessel.

I have seen tanks wrapped with carbon fiber to keep the pressure contained, so, I suppose that laminations of carbon fiber over aluminum would be ideal. These tanks were rated for 4500 PSI! Although, I guess it would be a bit easier for the carbon fibers to contain a pressure, rather than restrict a compression..

I'm just trying to find a good line on reasonably priced epoxy. And a company that doesn't need a 10 gallon minimum order :)

Steve
 

Externet

Joined Nov 29, 2005
2,200
What plans do you have for the cpu fan?

A 4000 psi rated tank will easily take twice the pressure. The safety margin is huge, from outside or inside. Do not see any need for carbon fiber reinforcements.

If you go the epoxic-kind-of-compound route, make sure it is of a type that can be dissolved with some 'available' fluid as acetone instead of a fancy $400/gal something.

The way I have potted some submarine things is first putting the circuit board in a balloon, condom, rubber glove, etc (or several layers) of the proper size to fit comfortable and pour the potting whatever on top of the balloon.
Direct wetting of the board (embedding all crevices) gives me the goosebumps. Can be done only once and any later changes (jumper, fine tuning, modification...) means another entire new circuit, which can be co$tly.

Silicone and nylon ties at the neck of the balloon will seal the cables entry. To give the balloon a less sharp/more uniform surface, the board can have its crevices filled with some plastic beads.

Smaller than a bathroom tile:
http://www.linuxdevices.com/news/NS9703792352.html

scubaMiguel
 
Last edited:

Thread Starter

scubasteve_911

Joined Dec 27, 2007
1,203
All of the cooling must be passive, epoxy isn't the best lubricant :p Due to the fact that the water will be, at most, luke warm, and it's high specific heat, all I need to do is extend a heatsink into the medium.

I am pretty much heatsinking every IC on the board, because I am worried about thermal-mechanical interaction. I'm really tempted by that board, it's about 250$ with all the heatsinking. I paid 100$ for the intel atom. The atom actually has decent performance, the biggest downfall is the southbridge?, which is actively cooled. The atom is happily cooled passively with a TDP of 4W.

I was thinking about molding some sort of urethane for connector thermal relief. Your idea seems attractive, but can it withstand ~5000PSI? I am aiming to go 4Km deep, but that's just a goal. I figure, if I keep my sights on 4Km, I should be able to get around 2Km. That extra stretch equals a lot more engineering and concerns. If the lake trial (about 500PSI) works out, then I should be getting more money to make it better.

Thanks for your input scubaMiguel, haha

Steve
 
Top