Decapping Chips

I wanted to get into decapping back in 2004 when my friend took this die shot of a Channel F ROM through a biological microscope, but was put off by the difficulty. Then I got all fired up about the work done at Visual 6502, and really got inspired by John McMaster's blog. I picked up a relatively inexpensive microscope and a camera adapter, and quickly found out that you need to automate the process of taking all those pictures. So I built an X-Y table from some aluminum stock and two PIC Design linear slides (unlabeled, maybe PB08-10?) that uses two linear actuator stepper motors (Haydon 26363-12) to position the dice under the microscope, then wrote code to control it using a PIC 16F88 microcontroller, using some Sainsmart MD430 stepper motor controllers. I added a 3rd stepper to control the focus; you pan to 3 of the 4 corners and manually focus at each, then the microcontroller calculates the focus while you pan around and moves the stage up and down accordingly. I wired the PIC to the shutter release of a digital camera, and began producing hundreds of pics. When I attempted to stich them into a large panorama, I discovered that that's even more work! Many of the tools seem to have steep learning curves, and the ones that are easier to use seem to have problems with this type of panorama. Microsoft's ICE, for example, looked like it was going to work well, but it keeps doing some kind of color/exposure compensation that pretty much ruins the resulting panorama. [Fix: export as layered Photoshop] And its preview of the panorama looks very good, but the resulting image has geometry issues. I wrote some Python scripts for GIMP to crop images, adjust for uneven lighting, and position the tiles on a large canvas, but they require a lot of manual work. I think my next step will be to use GIMP to crop and adjust lighting, have a custom program position the tiles, then create a project file for ICE with the position info and let it create the final composite.

Update 9/8/2016

Here's my latest setup: I was given a very nice motorized XY table and wanted to control it with trapezoidal velocity profiles, so I bought a KAE PIC-STEP IC. It has direction and step outputs for one stepper motor, as well as limit inputs and general purpose outputs. Since I don't need to drive all 3 axes at the same time, I multiplex the step output to the 3 stepper controllers. I wrote a new control program in C and talk to the chip via an old laptop's serial port. I have a manual tip/tilt stage and a manual rotation stage to level and square up the die before taking pics. I now use Fiji to composite the pics.

I bought a die inspection microscope off ebay. It's an Olympus BH2 on a heavy-duty base, with a special platform to bring in wafers one-by-one for inspection.