EPROM erase times and lifespan

Working with EPROMs, one must have an eraser. I built my own a while back, using a germicidal UV lamp tube, a flourescent fixture, a sealed box and a programmable timer. It works well, and I typically can erase an batch of eproms in like three minutes. At least, when I was working primarily with 2732's and 2764's, I could. Now, I've been needing to use higher capacity chips - and these seem to take longer to erase. Five minutes isn't enough for some. AMD 27C010's seem to take closer to ten minutes to erase. My eraser has a drawer that can accept about a foot of eproms (I used a 12" tube). I could probably cram more on there, but I try to keep them centered under the light. So, if I stick a dozen chips on there, of varying sized and types, and 'nuke' them for five minutes, I'll go and blank-check them, and some will be blank, others will be nearly blank (lots of FF's, but some garbage left), and some don't seem affected at all. I've been trying to come up with a correlation between manufacturer and erase time, but it seems to vary. Last time, I nuked four AMD 27C010's that I pulled off of an old LaserJet font cartridge. Three were unaffected and the other was blanked. I worry about leaving the chips in there too long, since I know too much UV can damage them. But at the same time, I wonder about the lifespan of the UV tube - and how many hours it's good for. (I guess I should have checked the package when I bought it...). So, how much UV exposure does it take to damage an EPROM? Is UV exposure cumulative? For example, if I have a chip that's unaffected by five minutes, and I put it back in for another three - is that the same as as running it for 8 minutes total, or do the charges that the UV is supposed to dissapate simply linger? Are CMOS chips harder to erase, taking a longer time than the older parts? -Ian