Laser-ing the IC markings off on a budget.

[Arek_R]

Hi I'm looking to into making the product difficult to reverse engineer.
I believe it can be done using some generic hobby grade 5W laser engraver, right?
Has anyone tried doing this?
Does anyone know what's the required power of the laser module to quickly do it?
Thanks.

[MuhScopeBroke]

just blast a few junk components with the laser and see what is too much/too little power. You can scrape those markings off with an exacto knife so I bet its doable with a laser

[janoc]

I guess you are not doing this in volume (hobby grade engraver ...) so even a Dremel and a sanding disc will do it.

On the other hand, this is a pretty futile thing to do if you are hoping that it will make the product difficult to reverse engineer. People routinely decap ICs these days to identify them, especially anyone who could be sufficiently motivated to copy your product.

However, what you certainly will achieve by doing this is earning the unmitigated hate of anyone who will have to repair your product later (I do hope it is not meant to be disposable ...). I.e. typically your own clients.

[Arek_R]

I know everything can be reverse engineered.
It's mostly to make the directors and managers sleep better.
It is for a product that will sell around 10k units a year, and is not intended to be repairable.
And hobby grade so that it proves how easy it is to do.
You also don't need laser precision, +/-1mm is acceptable.
Later on might buy prosumer machine.

[daqq]

On using a Dremel as was advised... beware. Those are some nasty vibrations applied onto the pins and part and it could result in issues. I've actually had pins that got broken off of the solder when I used a Dremel to "adjust" the board shape.

As to removing markings in general: Avoid. It's a trivial inconvenience to even a slightly dedicated reverse engineer, adds processing and equipment cost.

[Uksa007]

It is for a product that will sell around 10k units a year, and is not intended to be repairable.

With that volume you maybe able to order from the manufacturer with your own part number printed on them.

[Arek_R]

We manufacture to order and in smaller batches, buying 10000 components at once isn't convenient.
So anyone knows what power is required?

[SeanB]

Black solder mask, along with a pour of epoxy around the critical parts, after cleaning, will do more than just removing numbers. At least the black will make reverse engineering harder, especially on a multilayer board, with tracks routed in the inner layers, and a less than visible amount on top and bottom, and the epoxy pour will keep them from moisture.

All that will mean the cloner will just have to sit there and do extra steps to get a full circuit design out, so best, if you are using programmable parts, is to make sure to set the lock bits on them, and make it harder to just copy, plus if you can add an extra encryption to the firmware updates, so they are unique to each serial number, making it harder to simply reverse engineer an update to get the firmware. 10k units per year doable with only a few extra gray hairs a month. And yes for 10k units most suppliers of non jellybean chips will happily give you a custom part number for you to use in house, though you might be doing a bulk buy for a 3 year supply initially. Will just have a manufacturer mark and lot number on them, in addition to your custom part number. Otherwise invest in some paintable epoxy and do your own blacktopping of the parts after assembly, which is what the epoxy will be doing.

[Arek_R]

Thanks for advice but I didn't ask for alternative countermeasures or if it can still be reverse engineered or if it makes sense.
I asked if someone used laser and if yes what power of the laser is recommended.

[Benta]

10k is a reasonable volume, but not for a semiconductor manufacturer.

On the other hand, the big semiconductor distributors (Arrow, Avnet) will do laser marking as a service. They prefer to supply the parts themselves, mainly due to quality and MSL control, but will also mark what you give them (probably with a disclaimer).

They can mark anything, you can even upload a .BMP if you want your own logo or whatever on the chips.
Call your local sales office.

[Sal Ammoniac]

and is not intended to be repairable.

Why not? If you're using commonly available parts (or why else would you want to remove the IC markings?) then why shouldn't the "thing" be repairable? We already have too many disposable, non-repairable things in this world. 

[Arek_R]

It will be potted and enclosure riveted, also as a industrial product none of the customers would even think of fixing it.
I guess I will just have to buy the engraver and find out myself.

[Benta]

Did you read my post #9?

https://www.avnet.com/wps/portal/emea/solutions/logistics/product-modification-center/marking/

I can't find aything at Arrow, though.

There's usually a set up charge, after that it's a few cents per device.

[Arek_R]

I don't want to be rude but I didn't ask for this stuff.
I know there are better options but I don't look for best options.
I've just asked for some advices regarding lasering it off yourself, why do you guys have such a hard time understanding this.
Issue is that there were many qty of the ICs already bought and making it a top secret product came afterwards from the directors/managers, I don't have much control over this.
Also those 10k/year is nothing guaranteed, it would be nothing new if the customer would just cancel it out of nowhere.
Before they wanted to used secondary type of hard black potting as a separate process but it makes the entire manufacturing process a nightmare.
Once it's a established product in a full swing then yeah I would organise the component suppliers or 3rd parties to either manufacture ICs without markings or get it removed professionally, but where the project is now it's not a viable option.

[Arek_R]

Your response I so dumb I'm going to assume you're just a troll.

[KE5FX]

I don't want to be rude but I didn't ask for this stuff.

Welcome to EEVblog.  New here? 

I've just asked for some advices regarding lasering it off yourself, why do you guys have such a hard time understanding this.

15 seconds with a microfocus X-ray machine, and yeah, good luck with your potting and lasering and Dremeling and black solder mask and...

It's like the modern trend of failing to publish schematics "to slow down the Chinese."  It might slow them down by a day or two.  Then they will clone your product and the store it's sold in, if they are inclined to do so.

[nctnico]

It will be potted and enclosure riveted, also as a industrial product none of the customers would even think of fixing it.
I guess I will just have to buy the engraver and find out myself.

If you pot it then it will be next to impossible to get the chips out in one piece anyway. In this case engraving the chips doesn't add any protection.

So the answer to your question is: don't bother with engraving at all. If you are going to do engraving anyway you'll also need to think about filtering the toxic fumes from the engraver burning epoxy resin (nasty stuff when you burn it).

If you are serious about your product then make sure it is cheap (for which potting and rivetting the enclosure doesn't help) and add new features so your customers keep upgrading. Whatever gets cloned is outdated that way.

[Arek_R]

But if you leave the hard potting for the night in acetone it will soften to the point where you can peel it off with your hands.
Also entire point is to not have to use the secondary hard potting because it complicates manufacturing as it was not intended to be partially potted like that.
As I said it's a industrial product and price doesn't matter that much.
Still not a single answer to my question.

[bdunham7]

If someone knew, they'd tell you.  Probably very few here would bother with that sort of thing.  I haven't done it, but I'm sure one of those engraving lasers would do it, its just a matter of trial and error as to how to set it up and how quickly it can do it.

[tom66]

I know everything can be reverse engineered.
It's mostly to make the directors and managers sleep better.
It is for a product that will sell around 10k units a year, and is not intended to be repairable.
And hobby grade so that it proves how easy it is to do.
You also don't need laser precision, +/-1mm is acceptable.
Later on might buy prosumer machine.

Does the product have a microcontroller in it, or could it accommodate a very small one?
Make the MCU do something essential, blow the fuse protection bits, and your product is not duplicatable except by replacing that software+MCU, which will take a fair bit more effort than just copying the hardware design.

[JohnnyMalaria]

Hi I'm looking to into making the product difficult to reverse engineer.
I believe it can be done using some generic hobby grade 5W laser engraver, right?
Has anyone tried doing this?
Does anyone know what's the required power of the laser module to quickly do it?
Thanks.

This guy did it with a "40W" laser at half power

[JustMeHere]

Possible? Yeah, metal can be etched with a laser.

Practical,  well, you will need a 3 axis CNC machine to do it reliably. 

I would start out with 100 grit sandpaper.

[Arek_R]

Another troll?
Since when are ICs made out of metal?

[beanflying]

Another troll?
Since when are ICs made out of metal?

Practical Test tomorrow on my 40W CO2 Laser. Gut feel it will need something like 20-30% of power at near full scan speed (300mm/sec) to get it done.

I also do have a 0.5W Diode Laser I can have a play with but it isn't mounted in anything. The thing to bear in mind if you are tempted to look at a Diode Laser option is they are nearly all  spec and over driven to hell. Also the scan speeds achievable by these will be painful if you need to run a decent number of IC's

I have some random SMD chips I have salvaged from boards as rework practice and some DIP stuff I will sacrifice to the cause of knowledge.

[beanflying]

Ok so I got bored/intrigued and the Bottle of Single Malt Painkiller is in the shack 

Tried sitting some Loose SMD and IC's under the Laser but the Air Assist (Lens Cleaning) was not going to cooperate. Blu-Tack is an option but I decided to quickly lump some bits together for a little mass. Even during the test this moved a little.

Quick Fusion Mask to avoid needing to ablate the entire bit 40% at 300mm/sec and it could certainly be done at less power than this with a bit of experimenting. Also in a production setting you would drop the DPI in the vertical to save more time again but this gives a nice smooth result.

Very rough test but the junk end of the K40 rubbish would get it done or a nicer something for a few $ more like mine is a better option. Obviously better jigging and measuring than the 10 minute option too