Who doesn’t need to take pictures of the microscopic bits inside of an integrated circuit? [Mojobojo] made an end-run around the expensive equipment by building a microscopic lens from an old camcorder. He’s using a regular digital camera with the lens set to its largest zoom level. The camera is pointed into the salvaged camcorder lens where the fine tuning is done. His first iteration was just taped to the desk with a small hand flashlight illuminating the subject. He upgraded that setup by building a LEGO enclosure and changing to a much brighter light source. The images he’s getting are quite surprising and this will be very useful during those extreme hacks when you need to tap into an IC’s internal data rails.
21 thoughts on “Microscope Lens Hack”
Awesome hack. Anyone know how he managed to remove the chip packaging?
Paintstripper gun works miracles on chips that are already broken. Heat for a while and when the epoxi is soft enough just clip it with some wirecutting pliers.
Pics take also with an ordinary digital camera with plastic lens from a toy microscope eyepiece attached to the lens.
You can get those lenses from SurplusShed.com for around $20. I have one that I’m in the middle of adapting to a cheap ($30) CMOS video imager board. Works way better than those dinky plastic lenses and you get zoom and iris control (once I figure out the drive electronics), too!
The ones I got seem to come from old Sony camcorders.
OMG. Legos for the +1.
@decap: amazing pics you’ve got there! Have you ever tried removing metal layers from the dies?
needs more magnification
@svofski: Nope. Don’t have any tools to do so.
There is more magnification, not from digital camera tho…
You can massively increase the close-focusing distance of a lens with a simple 2x magnifying lens (such as from a jeweler’s loupe). This is the same concept put into use in that cell phone macro hack that was featured here a few weeks ago.
As described on that page, a 2x lens reduced close-focusing from four feet to an inch or two. This was just holding the loupe up to the lens, with a constructed rig and a faster lens with a longer focal length (that shot was taken with a 36-72mm f/3.5, probably at the tightest setting) you could almost certainly do something of this sort.
Clever use of existing equipment though.
@decap: I would love to hear more about the stuff you have there. Like MOS6581R3_reconstruct.jpg, is there a story to go with it?
@svofski: Figured out how to crack open a chip and later found a guy who had access to a computer controlled microscope.
After that, one line led to another in Photoshop and two months later it had been redrawn. Piece of history preserved.
@decap: awesome! Although a little bit of a letdown because I suspected that you’re actually planning on running a new batch of the chip ;)
I’ve done this before, bu just placing a 16mm eyepiece of a telescope in front of a digicam. The subject needed to be well inside the enclosure of the eyepiece for good focus, so I used a white SMD LED to illuminate it. This incidentally gave a cool look of depth.
@svofski: I’ve been asked that quite a few times yes. No resources, so if someone is willing to give it a go then I’m not going to object.
@decap: i don’t have a fab, so unfortunately.. Do you have a blog? It appears that many of those pictures are illustrations that went with some text, maybe on a forum somewhere? I’d just like to read more about the details that you found out about the SID. It appears that you did a really thorough job at dissecting it in detail.
I was under the impression that the MOS SID chip’s circuit design was no secret. I’ve read many interviews with the chip’s designer, and he doesn’t appear to try to keep any secrets to its design, at all. I also(perhaps mistakenly) thought that the SID chip had been completely emulated in software to the component level…
Don’t under estimate what a flat bed scanner can do.
Here is the guts of a inkjet print head scanned on my flatbed.
Not as small as an normal IC I admit. I’m not knocking the macro project at all but people often go to a lot of trouble using a camera where a scanner would be easier and do a better job.
Many off the shelf cameras can focus for 1cm but getting the subject lit is a problem.
@svofski: No blog. All that is, is in there, except the huge Photoshop files.
@eddie: That looks really interresting. I’d still like to see a real closeup macro of it.
@Hitek146: Then I’d like to know those details, because the last time I checked, the maker of the SID was very mysterious about the design in his interviews. Afaik, making an emulator at the component level is ridiculous and all emulators are based on the data sheet information and/or experimentation (if you know how it responds to certain commands, just replicate the behavior)
Jeri did it at least once in C64DTV, but she didn’t replicate the filters. I remember SIDPlay (or what was that SID emulator..) sources, they use filter model very similar to what I’ve found in decap’s research. For an FPGA implementation the problem is that finding digital filter coefficients requires algorithmic calculations, which is relatively hard to do in pure hardware. On a second thought, with this analog circuit at hand one could replicate the analog section of SID in discrete analog components, why not?
Getting out of topic of the original hack…
The available datasheets of the SID are only preliminary and some things got changed during the chip manufacturing.
And Jeri took quite some shortcuts while designing the SID for the DTV, missing filter being only one of them. Tho currently all the faults of the DTV-SID have been emulated quite accurately. OTOH rumours say it is possible to build a separate analog filter into the DTV. Also the DTV-ASIC is said to have been designed such that, when fully wired (unlike what was done with the current DTV units), a real SID can be interfaced into it. But these are just “rumours”.
Well I’m still longing for a full C64 project that I could load into my DE1 board. SID is the major and perhaps the only obstacle at this point — the rest of C64 is more or less faithfully implemented in teh FPGA64 project.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)