Moshe, I also have used an number of TO-220 transistor packages. In the past to get the isolation I used thin ceramic substrate. Several things were done to provide the 1.5 MM required. 1) a shoulder washer was used on the screw, 2) there was a counter-bore in the heat sink to provide a little extra creepage as the substrate was only 1mm thick. On occasion we have tried kapton film but European standards require either multiple layers or 0.4mm thickness. Also the kapton is not puncture resistant to the edges of the transistor.
My favorite solution is the use of a substrate with no holes drilled, and a plastic clip is shaped like a "U" bracket with one leg shorter and the other with an angled foot. The clip legs were about 8 mm square. The leg with the angled foot receives a screw and as the screw is tightened pressure is transferred to the transistor body. The transistor side of the bracket is shaped to hold the transistor captive and prevents a compromise in creepage and clearance. I have also used this successfully with thermal switches mounted to copper bus bars. I know the group does not like graphics to be embedded in email so if you would like a sketch please let me know which graphics format you prefer. One last choice was more difficult to achieve. We used TO-220 style resistors and by using a diamond saw were able to section several and show that the 0.4MM of solid dielectric was provided to the metal tab. However my inspector requires an impulse test on a component in every shipment we receive. IEC 664-1 recommends 2.5 kV for our application. Some parts had to be removed from our authorized vendor list for the part as they did not withstand the test (i.e. no breakdown, punch-thru, flash-over, etc.). Doug Powell Compliance Engineer Advanced Energy Industries, Inc. Ft. Collins, Colorado USA ---------- From: [email protected] To: [email protected] Subject: isolation of TO-220 package List-Post: [email protected] Date: Monday, June 08, 1998 4:32AM Hello everyone, The following problem has probably several solutions, some of which I've seen implemented in power supplies. PROBLEM: In many applications the metal part of the TO220 package (that's the square with metal backing and 3 legs in a row) is (or should be considered) under "live" voltage. At the same time, for thermal reasons, it is required that this metal be coupled to the best heat sink around = chassis. This obviously creates a problem of isolation, which seems to be solved by using a thermally conductive isolator sheet between the transistor and the chassis heatsink. However, if a screw, which passes through the hole in the TO 220 package is used (even if it is a plastic screw) then creepage requirements cannot be maintained (because the insulator is very thin). Solutions I saw to this problem are based on not using the hole and pressing the transistor to the heat sink by other means. I would be grateful for any suggestion on how to solve the problem: cool the TO 220 through the chassis while maintaining at least 1.5 mm (possibly more for reinforced insulation) creepage between TO 220 and chassis. Thanks Moshe Valdman
