51
\$\begingroup\$

As far as I understand, the die of a DIP package is located at the center and the rest is just the lead frame. Given that I have unused pins, can I cut the top part of this microcontroller (ATmega16/32)? Will it still function afterwards?enter image description here

Edit: thank you for all the answers. I have realized that cutting an IC is a delicate process and there is a high risk of damaging the chip. But I've done it anyway, shear cutters worked a treat. I decided to go for the 3 lower pins instead of the upper ones since they are further away from the ISP connector. Here is a photo of the final result (My new DIP-34 package works just fine):

enter image description here

\$\endgroup\$
  • 12
    \$\begingroup\$ This is so silly, I can't even count the ways. \$\endgroup\$ – Olin Lathrop Aug 2 '15 at 21:21
  • 5
    \$\begingroup\$ Very cool - though I'd have used a dremel rather than shear cutters for fear of shattering the package. Props for the adventuring! \$\endgroup\$ – Dewi Morgan Aug 2 '15 at 22:46
  • 2
    \$\begingroup\$ @DewiMorgan - using a Dremel was my initial intent, but I started to worry about static. Veroboard also helped to reduce the stress a bit. \$\endgroup\$ – v.m. Aug 3 '15 at 1:48
  • 3
    \$\begingroup\$ Hrm. To prevent static, a very slow, grounded hacksaw could also have worked, perhaps. But looks like the snips did a very decent job! :) \$\endgroup\$ – Dewi Morgan Aug 3 '15 at 1:51
  • 3
    \$\begingroup\$ It didn't fit in the project box. \$\endgroup\$ – v.m. Aug 3 '15 at 14:09
21
\$\begingroup\$

How do you intend to do this butchery?

Unless you have very specialized tools, a Dremel cutting disk or something like that could generate a lot of static charges. Well enough to kill the chip!

Moreover, mechanical stresses could damage the internal bond wires, or even the die. Let alone you would have the bond wires to the cut-off pins protruding flush from the cut side (8 of them), maybe shorted together due to harsh mechanical stresses during the butchering action.

People needing to reverse-engineer a chip do this kind of things, but they use much more "delicate" measures. Moreover, they want to expose the die, so they "cut" off the upper part of the package.

In particular see this link about decapsulation of the chip or this video showing LASER decapsulation.

Google for "chip decapsulation" and you'll find tons of references and you'll understand why it is a costly process if you want your die to survive! People pay big bucks to reverse engineer chips (both for legitimate and criminal purposes). Legitimate purposes comprise failure analysis ("Why our top notch IC failed unexpectedly?!? Let's crack it open and see what happened!") or retrieving lost designs ("OK, we acquired this little IC design house with these excellent parts. But, wait! Where are the design sheets of the groundbreaking HQC954888PXQ processor?!? Who fired the design engineers who knew?!?" - Yes these things happen!).

BTW, Did I mention all these methods are delicate?!? Side cutters are not what I could call delicate. When I was a boy I remember cutting a (dead) IC to see the die using a big side cutter: it splintered wildly!

YMMV!

\$\endgroup\$
  • 1
    \$\begingroup\$ Let's keep the comments civil. Off-topic comments will be deleted. \$\endgroup\$ – W5VO Aug 2 '15 at 14:04
22
\$\begingroup\$

I've never heard of anyone trying to cut an IC package like that, but it seems very risky to me. In addition to the potential for shorting bond wires and crushing the die that Lorenzo mentioned, I would also be worried about the performance of any analog subsystems like internal oscillators and flash memory. Package stress can shift the performance of analog circuits -- even the normal deposition of the mold compound tends to shift currents and voltages.

I have worked with decapsulated ICs and wafers, and I can tell you that IC dies and bond wires are thin and delicate, and are not designed to handle trauma. I'm curious to hear whether your chip still works, but I wouldn't recommend doing this with any chip you actually care about.

EDIT: I got curious and looked up pictures of DIP lead frames. Here's a picture of some DIP-36 lead frames on a reel that I found at a wholesaler's site:

Three DIP-36 lead frames, still joined from the factory

And here's a labeled close-up of the cut side of your package:

Labeled close-up of cut DIP-40

\$\endgroup\$
  • 3
    \$\begingroup\$ Yeap, chip seems to work just fine. Shear cutters did a good job - I've added a photo of the final result. \$\endgroup\$ – v.m. Jul 31 '15 at 23:09
  • \$\begingroup\$ Fascinating! I got curious and looked up some pictures of DIP lead frames. My answer now includes pictures showing what we're probably seeing in the cut. \$\endgroup\$ – Adam Haun Aug 1 '15 at 3:25
  • 2
    \$\begingroup\$ The center lead is connected to pin 11 (ground). Interesting observation: the resistance between pins 11 and 31 is about 2 ohms, although they should both be connected to ground. \$\endgroup\$ – v.m. Aug 1 '15 at 9:47
  • \$\begingroup\$ @v.m. 2 Ohms? Is this what your multimeter displays? Hint: Connect the two tips of your multimeter probes. If it still displays 2 ohms, it's your multimeter, not the lead you're measuring ;-) It'll just never go lower than 2 Ohms... \$\endgroup\$ – zebonaut Aug 5 '15 at 7:27
  • \$\begingroup\$ Of course I took that into account. There is a short between PIN11 and the center lead and about 2Ohms between pins 11 and 31. My guesses are: a) PIN31 is not connected to the center lead of the frame, but is routed through the die itself b) It was connected to the center lead but through the lower part of the frame (which I snapped off) \$\endgroup\$ – v.m. Aug 5 '15 at 16:55
13
\$\begingroup\$

Such things have been done at various times for various reasons. If one doesn't get too close to the chip cavity such techniques would be likely to short out an unknown number of adjacent pins but otherwise could work if one uses a cutting device which doesn't generate excessive voltages. There is a substantial likelihood of breaking the hermetic seal on the package exposing the chip to air and moisture, which could greatly hasten failure; I know of no means of inspecting a chip to determine if damage has occurred.

If one can accept a high likelihood of rendering a chip immediately unusable and uncertain reliability after that, such a trick might be usable if there is some particular reason you need to use a particular package which is too large for your requirements (e.g. if a part had 15 consecutive I/O pins, it might be possible to solder the legs of the chip directly to the legs of an LCD without using a PC board). Such designs tend to be rather hokey and unreliable, so cutting the chip might not make things too much worse.

\$\endgroup\$
6
\$\begingroup\$

There is no hermetic sealing involved like metal and ceramic packages as the plastic moulding is homogeneous and does not contain an open space that is brazed, soldered or welded closed. The thickness of plastic required for protection is very little, less than 1mm on modern thin flat packs, on a 40 pin DIP it is generous in all directions.

The biggest danger is cracking the casing close to the bonding wires and this is most likely to occur if a pincer or shear method is attempted.

Using low speeds with a coarse abrasive cutter to reduce vibration or using a fine abrasive or abrasive water jet to cut slowly there should be little chance of mechanical or thermal damage.

Using water or alcohol spray, flood or immersion would solve cooling issues in high speed cutting and mitigate possible static build up though a humid work environment might be enough.

A typical 40 pin DIP IC might tolerate 7 or 8 pin pairs removed safely from each end if the internal die size is not unusually large.

In general the plastic ICs are very robust and most static protection on mature components is quite robust.

\$\endgroup\$
  • 1
    \$\begingroup\$ It's also worth noting that sealing a chip is not necessary for functionality - they'll work just fine exposed to normal, relatively dry atmosphere. \$\endgroup\$ – W5VO Aug 2 '15 at 18:53
0
\$\begingroup\$

How barbarous! Even though it was to be expected the chip would still work (provided the surgery is done properly) why haven't you opted for a non-destructive method like a socket adapter or build yourself one using a flat-cable between 2 sockets of different sizes, for instance? While your method obviously works it irremediably damages the IC and makes replacement [almost] impossible.

\$\endgroup\$
  • \$\begingroup\$ I thought the OP was pretty clear, he had the chip already soldered into circuit and wanted to reduce one dimension. Drilling holes for LEDs into a IC is something I am going to try for sure. Building a complete circuit into a machined cavity would be very cool, a button cell or two may be the limiting factor. \$\endgroup\$ – KalleMP Aug 3 '15 at 14:31
  • \$\begingroup\$ Even then, de-soldering is still a valid option in this case. IMHO at least. \$\endgroup\$ – user59864 Aug 3 '15 at 17:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.