Uh, don't leave us hanging... we're gonna need the full story on that one.
So we made an Ooops.. Uwe and I - mostly Uwe - developed a power supply for large industrial electrical vacuum furnaces. This one was large! Could hold a 60 ton payload that was 24 feet long and up to 10 feet in diameter. The entire furnace took 1.3 Megawatts. We're playing in the big leagues here.
So, we developed a power factor corrected, fully digital power supply that used phase fired SCRs to control power. We switched power between taps on a transformer using an averaging algorithm. Changing taps on the fly on large inductive device is not trivial as our hearing and eyesight would attest to on the several prototype mishaps that occurred. <BANG and Flash Before magical gray smoke!> And these were on 10Kw or less.
Well, our prototype units used what are known as easy firing SCRs. Give the device a nice electrical kick and they start! For the mother of all furnaces, we needed to use power devices that could handle continuous 800 amps at 460v ratings. These are massive devices and known as hard firing SCRs. We thought we had a big kick to start the scrs. We learned it was not as big as we thought.
Because the scrs were not conducting equal half cycles, the transformers had a DC bias in them. What does that do? If large enough, the core of the transformer becomes "saturated" - the magnetic flux can no longer change with the change in current in the primary winding. A saturated core presents to the primary connection, only the winding resistance as the reactance of the coil drops to near zero. In effect, it becomes a dead short across the line!
Now, because the DC bias was just enough, it only saturated for the peak of the line voltage - and that for just a few microseconds - maybe as much as 100 microseconds. In that time, enough energy was dumped into that short to cause a tremendous thump in the transformer with the huge inrush of current. When you have energy of that magnitude, it has to go somewhere. The force of that short was enough to lift the 10,000 pound cabinet off the floor the one inch I mentioned! Talk about needing to change one's shorts! (Pun intended!)
Now, when this first happened, I was on vacation. Uwe had left the business and was thinking things VAG-COM. The shop manager calls me on my cell in Raleigh, North Carolina, and asks, "What do I do with dancing transformers?" Now this guy was always pulling practical jokes - so I'm like - "Yeah right Bob." He says, "I'm serious! Here, listen." He holds the phone out and I can hear the cabinet slamming to the floor. And I turned pure white! I nearly passed out as he added, "And we have 30 tons of titanium valued at over $2M in there right now on a extended cycle that heats for 72 hours and then cools for 48 hours. We can't stop the furnace or we ruin all of the material." "How far into the cycle are we," I asked. "8 hours!" Oh .......
I gulped... and I told him, "I need to sit and think. I'll call back in about 10 minutes." Uwe will have to comment as I do not remember if I called him at that point. I don't think I did. I decided to have them unbalance the heat distribution just a little - very small amount so we could back the power down on that 600Kw section of the furnace. We did and the dance terminated successfully. The heat run finished without further issues and I arrived at their plant to begin studying the problem. Uwe and I made a correction a week later and we never had dancing big guys again until 2 years ago when I performed a song and dance at Automechanika - but then that is another story!