Subject: Cause of Circuit Burn-Out at CERN From: Kevan Hashemi Date: Thu, 05 May 2011 09:40:43 -0400 Dear Andrey, Yesterday Ben and I figured out how the Dual Rasnik Heat (A2072A) in the Long Guide Tube (LGT) at CERN came to blow up. First, we shipped the LGT system with an H-Bridge Head (A2060H) with which we had tested the entire system. Later the same week, we shipped another A2060H to CERN with newer firmware. When you arrived at CERN, you replaced the old A2060H with the new one. At that point, connector J1 made contact with the LGT electronics chassis, charging the chassis to 24V. Meanwhile, all 5 of the A2072A printed circuit boards have the same error in the middle copper layer. There are two tracks, one for 3.3 V power and one for the Chassis Ground. They should be 10 mils apart, but something went wrong in the photographic processing at Advanced Circuits in Colorado, and the two tracks are touching. Thus 3.3 V power is connected to the A2072A mounting bracket. At first, the LGT worked fine at CERN, because the A2072A was in the tube itself, and not connected to the chassis. When you connected the tube and the chassis, you connected 24V to the 3.3 V power supply on the A2072A. When you plugged LWDAQ power into the A2072A, this meant that 0V was connected, so we now had 24V across the logic chips. They blew up. So, you see that two errors had to act together to produce this failure. First, the 24V connection had to be made to the chassis, and second, the chassis had to be connected to 3.3V power on the rasnik board. But notice that if we had not discovered this problem with the printed circuit boards two weeks ago, we would have discovered it sooner or later, and possibly at a much less convenient time. We will fix the circuit boards we have, and order new ones. Yours, Kevan -- Kevan Hashemi, Electrical Engineer Physics Department, Brandeis University http://www.bndhep.net/