Potting compounds

A customer is asking me to pot some boards, both to make a waterproof assembly and to obfuscate our design decisions.

Well, OK -- the customer wants it so it's the right thing to do.

The board assembly to be potted is over a meter long, and about 120mm wide. It'll get potted into a tray that forms the case of the finished assembly. The tray will be made from polyester/fiberglass laminate. The total height of the boards + potting will be on the order of 6-8mm.

Due to the use of some IR LEDs, the potting compound needs to be clear in the near IR.

So I'm looking for comments and suggestions. I already know the "you can't work on it if it's potted" comment -- that applies to everything below.

My thoughts on potting compound are:

• Good ol' epoxy, in some mix that's opaque to visible light, clear in the near IR, and very runny.

This should be a no-brainer, and it sticks to polyester well. But epoxy does have some scary features on its MSDS that give me pause (and means that I won't be able to talk my case vendor into doing the potting).

• Polyester resin, with some dye added for the right clarity/opacity combination.

This has the advantage of being 100% compatible to the polyester/ fiberglass, but polyester resin shrinks as it cures, and it's brittle. I suspect this is a Bad Idea.

• Some other 2-part mix?

• Some thermoplastic? Hot glue? Pitch? (I've got a dead fir tree in my field that apparently bled to death -- if I could render down the bark I could probably do half a dozen pottings with the scrapings from that, and I could call the finished product organic and free range). I know there's something that used to be common, but black tar would be a no-go because of it's optical properties.

--

Tim Wescott 
Wescott Design Services 
http://www.wescottdesign.com 
==================================================================== 

I've used a few products from Smooth-On for casting, polyurethane hard  
plastic and soft rubber, but never for encapsulation.  They have one product  
that may work,  
http://www.smooth-on.com/Castable-Epoxy-Res/c1295/index.html - their  
epoxacast 650.  Call them and see.  They have dyes to mix in for color  
choice and more opacity, but of course the question is going to be the NIR.  
You will ultimately probably have to just buy a test kit and try it.  Could  
you literally just put a short piece of soda straw over each LED so it isn't  
covered over by the encapsulant?  Either leave them in place or spray the  
outside with mold release and try to pull them out after the pour, maybe  
before things are fully cured. 

----- 
Regards, 
Carl Ijames

Dunno about visible opaque / NIR transmitting resins. One thing that will save your remaining hair: a thinnish layer of soft silicone for stress relief before your final potting compound. The forces generated by big chunks of epoxy as they cure are not kind to SMT electronics, or even lots of through-hole stuff such as pots and electrolytics.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

Unless your competitors are operating out of a *garage*, I think you'll find potting does very little to obfuscate a design. I've been involved in some "heavily counterfeited" industries, in the past. In one, we had an ongoing "friendly competition" in which we would periodically exchange "demonstrations" (prototypes) of obfuscation techniques. Potting, using 3D interconnection techniques, embedding metal chips and other XRay-opaque materials, etc. Invariably, we would each be able to "crack" the demos presented to us by the other party. (and this was just a "hobby" sort of activity in our organizations; not one that was aggressively pursued -- as there was no financial gain available!) Eventually, we moved to full-customs with draconian warranty/repair/replace policies to discourage folks from just buying those components as "spares"

These days, "college kids" have access to far more capable kit and techniques than we even imagined, back then! How deep are your competitors' pockets? You also risk driving the competitor to more aggressive *design* activities ("Why waste time sorting out what *they* did? Let's just look at the published functionality and aim to beat, that!")

But, be that as it may...

Why not just a good conformal coat with some suitable dye? Or, *paint* the assembly before conformal (clear) coating? That allows you to protect areas and components that you'd NOT like blocked.

I know one large manufacturer that simply "wiped off" the markings on key "novel" components.

So, your warranty repairs are now more expensive: warranty *replacements*!

I.e., you've increased your costs just to protect a design (with no guarantee that it's really "locked up"). IME, that's the downside of all protection schemes: you incur added costs, activities and don't practically affect your competitors results.

In the gaming industry, it was always amazing to see how easily the "knock-off artists" would copy 99.5% of the code -- without bothering to understand any of it! -- and still manage to tweak it enough that the *visible* product "appeared" to be different. Yet, the "protection schemes" imposed additional development and manufacturing costs on the original developers!

On Fri, 19 Jun 2015 15:39:21 -0500, Tim Wescott Gave us:

"obfuscate" and "clear" are mutually exclusive.

Grind chip ID surface, and clear cote... Done.

We dab secret parts with a real hard shiny green epoxy. One could rub off the wimpy laser-marked part numbers first, but I suspect it's not necessary.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

I think you'll find those goals are mutually exclusive.

100cm x 12cm x 0.8cm = 960 cm^3 volume. The density of epoxy is about 2 gm/cm^3, so the potting goo will weigh 1.9Kg (4.2lbs) plus the weight of board. That's about the weight of a common red brick. I suggest you hand the customer a red brick and ask them if they want to ship this much additional weight.

The tray is a problem depending on what flavor of epoxy you select. Forget about one-part epoxy. The stuff is highly exothermic and will melt you components. I know, I've melted plastic caps with one-part. Two part is better, but usually takes 12-24 hrs (or more) to cure properly. During curing, the epoxy moves around, expanding and contracting with the reaction and with changes in temperature. I've successfully ripped the leads out of components by trying to accelerate the process with heating. In other words, with a big board, forget about epoxies. Look at wax or silicone encapsulating compounds.

If your PCB has any devices that require cooling, you'll find that the potting process blocks all air flow. If you pot a heat sink, you might was well remove the heat sink because it won't work. Of course, I had to try it to see what would happen. The difference in thermal coefficients caused the epoxy to crack and eventually fall off in flakes. This was aggravated by heat buildup in the aluminum heat sink.

Put a soda straw over the IR LED's and be done with this diversion.

Actually you can if the potting compound is fairly soft. However, that offers little protection against reverse engineering. I've used a thin wax coating for waterproofing and reparability on marine radios.

See comments on one and two part epoxies. The company I worked for during the 1970's had a division that made function modules (A/D, D/A, amps, etc) in potting shells. This wasn't for IP protection or waterproofing. It was for shock resistance and to eliminate thermal gradients. Epoxy was used on the smaller modules, but when the epoxy started tearing boards apart in the larger modules, we switched to silica sand followed by a plug of epoxy to seal the shell. As long as there were no air gaps or voids, it worked quite well. Something 1 meter long wouldn't have a chance. The PCB is going to flex, cracking the encapsulant and there goes your waterproofing.

Yeah, epoxies are evil. That's another reason you should switch to silicone, acrylic, and wax encapsulants and coatings.

You left out plaster of Paris, concrete mix, and whatever tar Apple uses to embalm their new MacBooks and iPads. You might also consider using lard, which much of the world considers to be a religious abomination and won't touch it.

I'm not sure what to suggest. It's beginning to look like a mix of conformal coating and potting, which are rather different. Therefore, I suggest you talk to the usual vendors. etc.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

y

Permatex Ultra Black is interesting stuff. It's supposed to be safe for O2 sensors (i.e., cures without releasing acetic acid), it's cheaper than mos t electronic-grade RTVs, it's visually opaque (which may or may not be a de al-breaker for your IR application), and it's hard to remove without destro ying whatever it covers.

I agree with the other comments. Nobody who is capable of posing a competi tive threat will be deterred by potting compound, but good luck explaining that to the customer...

-- john, KE5FX

On Fri, 19 Jun 2015 21:59:43 -0700 (PDT), "John Miles, KE5FX" Gave us:

Nobody is grinding chip tops off to steal the guts any more. Time to market elements mean that their IP invasion will cost them far more than the goal will net them. They are simply too dense to copy in any viable time frame. It is a truly rare "method" these days.

Maybe idiots like the Russians are still *attempting* it.... Usually unsuccessfully.

Please don't try to talk me out of something the customer is dead set on. I've tried talking him out of it -- I need solutions to the problem he's presented.

--
www.wescottdesign.com

In the early 1960's, I had all kinds of strange problems with PCB's that used potted transistors. They had a white base and a round dome like top, which was some form of epoxy. The problem was that the potting compound was transparent to IR. I spent quite a bit of time trying to remove the 60 Hz hummmmm from a circuit. The problem was obvious when I found that turning off the nearby incandescent lights would make the hum go away. Amazingly, I found an old board in the closet with these transistors (the two white/black devices in the foreground): The problem is that I don't know what type of epoxy was used and have no clue as to its optical properties. My guess(tm) is that it's similar to the dark potting compound used for IR LED's, such as:

Epoxy dissolver is fairly common. For example: I've used something similar for post mortem analysis and reverse engineering. It will dissolve the epoxy in the typical G10/FR4 PCB's, but will leave the traces intact. If handled properly, that's usually good enough to extract the surviving components. Depending on the formulation, these solvents will either soften IC packages, or leave them mostly intact. I haven't done this for several decades, so I'm way behind the times. I used acetone and xylene for softening silicon potting compounds. Search for "epoxy stripping", "de-potting", and "unpotting".

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

I'm not trying to "talk you out of it" -- just explaining the realities that you will eventually face. I've worked in markets where counterfeiting was a rampant problem -- even for incredibly small sales volumes! If that's a *real* (vs. "imagined") issue, then your efforts will amount to naught.

Once *one* copy of your design (or parts thereof) is "out", you've lost the game; the fact that the next 5,000 units that you produce are STILL "obfuscated" means you're *still* carrying that cost -- even though the genie's out of the bottle!

The solution I offered was an opaque conformal coat. It will prevent people from "casually" inspecting (and probing) the circuit. It will do little to nothing to prevent folks from seeing what you've done from actually doing so if motivated as such!

[The original Votrax (classic speech synthesizer) modules were potted COTS digital-analog circuits. Would you like me to send you a *schematic* of those contents? :> StorageTech made a product line out of IBM clones... and you *know* IBM wasn't publishing "manufacturing details" of those products!]

Give your client a token solution to his problem; expect it to be INsufficient; then, you'll have invested very little time and money (that could be invested in actually improving the product!) in something that turns out to be, effectively, "of little value" (i.e., a step above "useless"). He'll either learn a LESS expensive lesson -- or not!

[DataI/O goes to great lengths to make their products (e.g., Unisite) difficult to copy -- even encrypting the "firmware". But, pull the COTS CPU out and insert an ICE, click on "record trace" and you don't even have to bother disassembling the opcodes!] [A "fan" *hand* disassembled the binaries to Robotron2084 (an arcade piece). Then, went on to isolate and remove the run-time protection mechanisms (what we called "Jap-zappers" -- as the primary counterfeiters were Japanese garage-shops), recognizing them as offering no functional value to the design.]

Ask them if an incendiary or explosive anti-tampering device is acceptable. It probably won't be acceptable but it might get them thinking along those lines. Very small destruct devices are available, if you know where to look. For example, a tiny charge that destroys a single component. A pyrophoric coating that incinerates the PCB when exposed to air. Vials of caustic liquids that break when tampered. Tiny UV flash bulb that erases EEPROMs. Firmware in battery backed up RAM that clears when the battery is disconnected. An internal battery to fry sensitive components when triggered. Use your imagination and I'm sure that the various acronym infested agencies have tried it. With proper outside packaging (for safety), such anti-tampering schemes are viable.

Meanwhile, back to coatings versus potting compounds. The big difference is that coatings are made to be applied in thin layers, while potting compounds are designed to cure in "bulk" volumes. If you try to use one-part conformal coating compounds for potting, the outer surface layer will harden, but the underlying compound will remain soft. You can solve this problem with heating, catalysts, and applying the stuff in layers, but it's easier just to use a made for the purpose potting compound. An 8 mm thick layer might be problematic. It may need to be applied in layers. I don't know if it will cure properly when applied that thick. Ask the manufactory.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

On Sat, 20 Jun 2015 09:56:03 -0500, Tim Wescott Gave us:

The question is... did you tell him that anything other than a thin conformal coating will add ten pounds of weight to the finished "product"?

Hi Tim, I know nothing about coatings. But sometimes people get all excited about having one "thing" solve two problems... but when you look at it you find it's perhaps better to separate the tasks. So I'll just ask if it possible to talk about separating the water proof and conceal tasks.

George H.