I’m bringing home the analysis software for the acoustic microscope today. I have a report to work on at home tomorrow (“day off”), and I need to be able to manipulate the images I have and play with a small amount of data that I took. Let me describe what a “small amount” of data is… words behind cut to spare those who could care less.
Those of you that play with/produce music might be able to appreciate this on some level if you deal with waveforms.
We shoot focused ultrasound into sample, usually a plastic IC or electronic component of some sort, to get an internal image of the part. we do everything from plastic ICs to crazy high tech chip packages to silicon wafers, etc. usually the purpose is to inspect for voids. Anyways, when we focus the ultrasound to whatever area inside the part we’re interested in, we get a waveform readout on an oscilloscope. it’s a bunch of positive and negative peaks going from left to right across the time domain. when you’re a badass pimp like me, you can look at this waveform and know what it is you’re looking at. you set gates to take the data from the peaks you’re interested in, and the scanner goes at it. it collects the information from these waveforms at every location on the part, basically translating the highest amplitude on the waveform into a point with a value of -X to +X (max negative to max positive amplitude). a color map is assigned to the image based on these values, and we get all those pretty pictures you see in the linkys above.
the machine has another mode where instead of just taking absolute intensity values, it saves the actual waveform for each point. the effect of this is that we can go back and use this data to look at the part as if it was still in the scanner. we have a virtual acoustic “solid” of the part. we can perform fourier transforms in on the frequency responses at different locations and do all kinds of other crazy stuff. here’s the part that I have trouble wrapping my brain around…
I took one of these chunks of data of the part I was interested in to play with tomorrow at home. It’s about one square inch and a few millimeters deep, scanned at 50 MHz and 1024×960 resolution. After I was done scanning I went to check the file size. It was 3 gigabytes. I’m used to these files being pretty big, but that’s crazy. I reduced the scan depth to the millimeter or so in the middle of the part that I was interested in and got it down to a modest 500 megabytes to fit on a CD. It blew my mind a little that so much information was neccecary to describe such a small area.
yikes.