are you really, really, really, REALLY sure you didn't power it backwards? An electrolytic with -20V on it will crap out FAST.
Yeah, I am really really really sure of that. I checked that many times and the circuit has been running for a while at around 10volt. Each time I rerun, I check again so I was really sure. What I am not sure is which capacitor actually did the venting.
Since I was trying to do a test as voltage decrease, I ran a bunch of caps in parallel. The 4700uF was the second one in a chain of 6 (470, 1000, 1000, 220, 4700, 330), the smoke came from the 4700 end and it did looks like it came from the 4700. Careful inspection later (after the runs I was going to do) shown the 330 is the one that looks a little bulge out on the top whereas the 4700 looks absolutely normal - the 330 is the only "reused" cap whereas all the other ones are new.
In any event, my need for that is done, I ran it attended and finished the runs I wanted to do. I plan to order a replacement 4700 should I want to run that test again. I was actually fairly confident about it. The tests started with the caps fully charged and starts discharge as I run the test. I lowered the desire voltage to around 22. I know that when it didn't vent again at the start, once started, as the cap discharges and the voltage will continue to decrease and as the voltage decreases the less likely it will vent. While it made logical sense to me (that the high risk of vent/spark is at the start), I was interested in thoughts from those more experienced.
My concern was not really life or death. I know a 4700uF@30V doesn't really have enough energy to burn down my desk. I hate the smell, and I did not want the smoke alarm going off at night or sparking - and I was hanging on the hope that I could do over night run unattended. Rather than waiting another week for parts, I decided to go with it and just babysit the silly thing. It worked out.
Thanks
Rick