At LinuxCon last week we had a free event at the National Museum of Scotland. As well as the free whisky tasting and the Black Knight, there was something a little bit more personal on display: an early wire X-ray detector (pictured above).
My first ever job was making these. That’s an exaggeration: I was in no way smart enough to match the people who really made these, but I did do a bit of the software and also designed and built the test hardware.
The one on display dates from 1970 and is very crude compared to the ones we built (it’s made of wood!) But in principle all the bits are the same.
The large square (of wood) holds a grid of wires, horizontal and vertical, separated by a small gap. When running, the wires carry a high voltage. The frame is filled with argon gas (hence the clear plastic pipe). The grid is placed in front of the source of X-rays, which pass through from back to front. When an X-ray hits an argon atom, it ionizes it (knocks off an electron), and a small amount of current flows between probably two adjacent horizontal wires and two adjacent vertical wires.
The pulse this produces is amplified by the circuitry at the top and right, which also timestamps it and digitizes it. It then passes through the black wires (now rudely cut off) to some accumulator memory (RAM) which records the location, strength, length and timestamp of the pulse. Ours had 1 GB of RAM contained on 24 separate boards plugged into a VME crate.
The “resolution” of the wires is quite low, roughly 1mm, but by looking at the relative strength of the pulse on adjacent wires you could get around 10x the resolution.
The whole shebang was used at Daresbury for protein crystallography, but similar hardware is used on the LHC for detecting Higgs Bosons and other less glamorous science.
A final thing: We “competed” against a team using CCDs for X-ray detection. The CCDs had a much higher resolution, but were incredibly slow to read out (as in: hours), whereas the gas-filled wire detectors had a low resolution but the results could be viewed essentially instantaneously. Of course CCDs are now much, much better. You probably have a very very high resolution CCD array in your phone, with a sub-second read-out time.