Reverse-Engineering

of an X-ray Imaging Device


Continued from part 1.


With some trepidation, but no other avenue to follow, I opened the cover of the actual imaging module (Figure 4). Underneath the fiberglass cover, I found three layers of foamy plastic sheet. Underneath those there was a blue foil that was not glued to anything. Lifting the foil revealed the actual image sensor array (Figure 5).

Figure 4: Removing the brown fiberglass cover from the imaging module reveals three layers of plastic foam, with the blue scintillator foil underneath. The white side of the scintillator foil is in direct contact with the image sensor array, but not glued to it.

Figure 5: Close-up of the image sensor array. My initial guess that the array is composed of eight independent panels is confirmed.

With an Internet search for x-ray image sensors, I hit gold: Teledyne Rad-Eye image sensors. These are large-size CMOS sensors with a 512 by 1024 matrix of silicon photodiodes with 48 micron pixel size. The 2x4 array therefore features 2048x2048 pixels on an approximate 10x10 cm footprint.

Even more interesting: The web site offers full data sheets and two application notes. Each sensor module has a 15-pin ribbon connector, and the significance of the pins is:

1 VDD 5V main power supply
3 OUTS Positive analog pixel signal
4 OUTR Negative analog pixel signal
6 VREF Reference voltage (3.8V)
8 SCAN Scan mode control
9 START Frame start (input to the module)
10 CLOCK Pixel clock (input to the module)
11 BIN Binning (scan mode control)
12 NDR Nondestructive readout (input to the module)
13 FSYNC Frame sync (output from the module)
14 LSYNC Line sync (output from the module)
2,5,7,15 GND Ground

The datasheet even comes with scintillator recommendations. Apparently, the blue foil is a Kodak Lanex GdOS scintillator. A search for the part numbers (RE1004-01, RE1004-02, or RE1059) comes up blank. Probably discontinued.

Two challenges exist: First the output voltage swing is +0.7V on the positive analog output (OUTS), going from 2V to 2.7V from dark to saturation, and -0.7V on the negative analog output (OUTR), from 2V to 1.3V. The data rate is 2.5 MHz, and the differential signal requires an analog amplifier with a 5MHz bandwidth. A correspondingly fast ADC is required. Rad-Icon suggests a discrete instrumentation amplifier built around the OPA2353 in combination with the AD9221 ADC for 12 bits per pixel.

An intereting question emerges... can any of the on-board components be subjugated and put under the control of a different microcontroller? After all, the high-speed amplifiers and ADC are exactly what we need.


Let's get to the nitty-gritty... Click to continue.