Waterproofing controller (possibly ideal product found)

[wannesd]

Hey ES,

I've now officially read every thread there is about waterproofing electronics.

Two things I've noticed: there seem to be a lot of people against potting the controller, for two main reasons:
1) Heat dispersion
2) Impossible to replace components when something breaks

First of all, let me say that air has a thermal conductivity of 0.026 W/m.K source
This is a very bad heat transfer medium of course, as demonstrated by people oil cooling their hubs.

I think that the hottest parts in a controller are the fets, which are heatsinked to the aluminum case.
There's also a drain down resistor which gets hot:

Many controllers use a big power resistor to drop the input voltage to the regulator. These can run pretty hot, so coating them in thick epoxy might not be a good idea. There are controllers on the market that are potted in resin, but some have proven to be highly unreliable. Tongxin sold/sell a tiny potted controller, but they fail because of local heating internally. Two that I've seen have had burnt areas around the current shunt, showing that the potting was acting as a good thermal insulating layer, to the detriment of reliability.

This I don't understand, as epoxy typically has a thermal conductivity greater than air, so it should help remove heat from the resistor, faster than an air gap (which is what it is now) would.

Anyhow, I found a material which I think would be perfect for potting controllers:

THERMALLY CONDUCTIVE GELS

Dow Corning silicone gels are soft and cure to form a cushioning, low-modulus, resilient, gelled material. Cured gels retain much of the stress relief capability while developing the dimensional stability of an elastomer.
Dow Corning offers a line of thermally conductive gels that couple the stress-relieving capability of a silicone gel with the ability to dissipate heat from devices.
These thermally conductive gels can be used as potting materials for transformers, power supplies, coils, relays and other electronic devices that require a low-modulus material for thermal dissipation.

They can also be used as ingredients in formulations for thermally conductive gel sheets.
These silicone gels cure without exotherm at a constant rate regardless of sectional thickness or degree of confinement.
Dow Corning features thermally conductive gels that have controlled volatility, including one UL 94 V-0 approved product.

Specific versions of thermally conductive gels contain glass beads designed to guarantee a minimum bond line, ensuring a reliable electrical insulation.
Those materials find their use as liquid gap fillers and can favorably replace thermal pads.

Out of their line-up of gels, I find the SE4445CV to be the best suited: easily removable afterwards due to the fact it does not harden, but becomes a gel like substance, and a thermal conductivity of 1.26 W/m.K, 48 times better than air!


What do you guys think about it?

 

[wannesd]

http://www.ellsworth.com/dow-corning-compound-se-4445-encapsulant-gray-210ml-cartridge/

Not cheap though...

 

[wannesd]

Nobody enthousiastic about this? Guess l'll try it and see if it's any good.

 

[liveforphysics]

I've used it. Neat stuff. Do not have any grease or oils on what your trying to pot or it cure inhibits badly. I imagine it would improve durability and likely performance in many ebike controllers.

Many EV controllers and other automotive controls use this type of material for potting.

 

[wannesd]

Nice. Know of any cheaper suppliers than the one I linked?

 

[cwah]

can't you use thermal paste?
http://www.aliexpress.com/item/Cpu-thermal-conductivity-of-silica-gel-thermal-grease-waterproof-cpu-electromagnetic-furnace/1090490494.html

 

[whatever]

I wonder what sort of thermal conductivity inexpensive silicone has? perhaps worth trying some of the grey coloured ones?

 

[alsmith]

Make sure that any silicone sealant you plan on using is not electrically conductive- it can be formulated either way.

 

[bee]

This I don't understand, as epoxy typically has a thermal conductivity greater than air, so it should help remove heat from the resistor, faster than an air gap (which is what it is now) would.

I suspect there is still some air movement around the resistor due to convection.

Has anyone tried filling a controller with mineral oil?

What about waterproofing wiring? Mineral oil wicks into wiring quite well and might prevent corrosion for those riding in the winter.

 

[friendly1uk]

This I don't understand, as epoxy typically has a thermal conductivity greater than air, so it should help remove heat from the resistor, faster than an air gap (which is what it is now) would.

I suspect there is still some air movement around the resistor due to convection.

Yes, potting stops convection currants, leaving just conductivity to take the heat away. So if you pot it. it is surrounded by material almost the same temperature, slowing heat transfer from the component. If you allow air movement around the component it is unlikely to have it's heat dissipation inhibited so much.

 

[bowlofsalad]

Mineral oil might not be a good option, there are other heat exchange fluids that might be a better bet. Sealing a controller correctly might be difficult, you really don't want the controller to slowly drain of fluid because if it did, that controller would be destroyed pretty quickly. The oil residue would act as an insulator, the heat would build up. I imagine if you wanted to fill the controller with fluid, you would orient the controller vertically as it would make a slow leak from the wiring holes easier to avoid. The issue would then be if fluid still leaked, eventually a fet or two would be exposed from the oil and cook itself.

There are lots of options for non-conductive heat transfer pastes, grease and pads. I have been considering using some similar option for this myself. In fact, though my understanding is limited concerning EV controllers, I am surprised to hear that there aren't at least some cheap thermal pads between heated surfaces. It really sounds like the controller is more of an oven. Perhaps it is planned obsolescence. The fets and plenty of other parts in the controller will have a significantly shorter life span if ran continually hotter.
http://www.newegg.com/Thermal-Compound-Grease/SubCategory/ID-85?Order=PRICED&Pagesize=100
http://www.frozencpu.com/cat/l2/g8/c487/list/p1/Thermal-Thermal_Pads_Tape.html

I don't know what sort of gap there is between the fets and the aluminum casing, that would be useful information. Maybe the gap varies between controllers, I imagine so. Even if the gap is exceptionally small, even if the aluminum is touching a fet, it would still be extremely wise to have some thermal transfer material in place between the heat exchange surfaces. This wouldn't be good just for the fets. The air inside of the box would stay cooler this way, making sure anything else inside of the controller wouldn't be exposed to such extreme heat.

I have been wondering if spraying the controller with boeshield t-9, removing the side covers and placing something like bug screen over each side would be suitable. I don't know how I would work all that out exactly, maybe leave the sides alone that involve the wires and use the other two sides. You may also consider just drilling a bunch of holes in some fashion.

 

[IBScootn]

What about making your own mineral oil gel? I found this quote:

http://www.gelcandlemaking.com /

"Candle gel is made of 95% mineral oil, and 5% polymer resin. This resin is a powder that, when mixed with the mineral oil, transforms it to a more solid state. Very similar to Jello!
Penreco is the company that manufactures the gel and holds the patent on this technology. This is the only brand of gel I recommend using because of it's quality and consistency. Versagel™ comes in 3 types: CLP (low density), CMP (medium density) and CHP (high density).
The thicker or more dense the gel is, the more fragrance oil it can hold. "

 

[IBScootn]

Here's the recipe to make your own mineral oil gel:
http://m.wikihow.com/Make-a-Gel-Candle

 

[Chalo]

It seems like potting the controller in wax would provide heat removal through phase transition, along with sealing. And the wax could be gently heated to remove it for controller repairs.

 

[d h zehrbach]

An issue with the FET cooling is that many controllers have a wafer of plastic between FET and the heat sink that actually insulates the FET from the heat sink. Gels and oils help by just bypassing the heat sink and taking the heat to the outer case through the medium.

What would foil the gels would be if there is any shrink back from touching the controller outer case. While the gel mass would make more mass to absorb heat spikes, if it is not touching the case firmly, it will not work as well as planned.

This is compounded by terrible conductivity from the heat sink to the outer case. You can make an improvement just by adding more bolts to the heat sink to bind it better than the normal few screws in a straight line. Adding them outboard of the existing rows makes sure the full sink touches the outer case. So if there were say 5 bolts in a row, add 5 more on each side of the first row in a staggered position to grip all of the heat sink.

I often see several inches of heat sink bar sticking past the bolts on both ends of the bolts. Adding bolts there helps a big amount.

I see a lot of controllers buried under the bodywork, under the seat, etc where they see no or limited air movement. In a device where even 15 degrees shows up on MOSFET operation, its better to make sure the controllers see the maximum and the coldest air possible. (That is why all ZEV Electric have the controller on the bike side)

All ZEV controller cases are black anodized as they shed heat better than plain aluminum by about 7% according to Alcoa. (works on motor cases too. Do inside and outside )

I made one controller a few years ago with two big rubber hose connections 1.5 inches in diameter with the hose running up into the tail of the bike. A computer fan up under the seat sucking (to keep out water) with the intake sucking through an air filter under the tail to keep out water seemed to work well. Cheap and easy at least.

Our normal ZEV controller has 26 MOSFET but this new double stator 36 Kw motor for the LRC and the T-10 required 30 MOSFET when we did the "load the bike with sandbags and run long grades" testing. I do not like the larger case so we are scratching our heads to try to cool the smaller 26 MOSFET unit and make it work better. I do not think it will take much.

D H Zehrbach
Z Electric Vehicle

 

[esoria]

i use spray like this....
In 2 year of rain, i open a lyen controller and is like new, no oxide, nothing...
http://crcindustries.com/auto/?s=18411