We have an old HP-5890 GC with FID that has been working flawlessly for years (decades in fact). All in a sudden the FID signal stayed saturated, and, upon disassembling the FID, we were shocked to find there was A LOT of water and rust inside the FID cavity. We cleaned the gunk out and sonicated/scraped everything clean, only to find out that water and rust returning after a day or so. Considering it is a GC and the column didn't die, there is no way the rust came from the sample; the detector was maintained at 250C so water should not condense. Then where did the water/rust come from and why? The GC runs on commercial compressed gas cylinders of the UHP grade, and we even put an extra activated charcoal filter in the air supply line, and that didn't help.