About two months ago, I added a heated bed to my custom 3D printer in order to print larger ABS parts for my research project. The heated bed (the PCB kind) was not new, but taken from an old printer I had built, but took apart. The bed worked well for a few weeks, but after one print finished, the glass bed above the heater PCB had shattered into several pieces (represented by bed 1 in the image below) and the nozzle was below the level of the bed (I believed it had lowered into the glass causing the breakage. I haven't determined what caused this motion, but it hasn't happened since). Notably, this print was using the heated bed at 90 °C. I chalked this up to a freak accident, and since it did not happen again, just replaced the glass and kept printing.
However, as soon as the heated bed was activated after the replacement, a small crack appeared on the glass and continued to lengthen as time progressed. I took off the glass as soon as possible and prevented it from fully breaking (see bed 2 in the image below. This bed was smaller as I didn't have access to a large enough piece of glass at the time).
At this point, I figured something more than an impact caused the glass to shatter. Since both cracks occurred when the bed was heating or cooling, I figured that thermal shock could potentially be the source of the cracking, and a quick google reinforced this idea. Due to the nature of both cracks (not being straight shards but meandering around the build plate and propagating slowly), they both appeared to have been caused, or at least propagated, by thermal effects.
To try to avoid future cracking, I took care in assembling the third bed. The heater PCB was attached tightly to the glass with Kapton tape and a thin layer of thermal paste was added as an interface layer to try to get an even contact and heat distribution throughout the glass plate. I made sure that the cardboard shims (which press the glass into the clips) were not too compressed, thinking that pressure in the middle of the glass plate from the shims may have accentuated the cracking by putting the top of the glass under tension.
But after a few cycles with this new bed, the same problem appeared (bed 3 below). This time, the cracking was as severe as the first case, but no impact occurred and I was not touching the bed. The bed was heating up to temperature (90 °C) when the cracking occurred. The strangest part is, the file set to print was one I had already printed successfully on the newest bed.
At this point I am at a loss and don't know what to do next. I don't want to make another bed just to have it crack in a few prints, but I need the bed in the near future. Any suggestions to mitigate this problem would be greatly appreciated.