Semiconductor stress isn't always electrical. No matter their size, semiconductor devices don't like mechanical stress either. Sometimes, one does need to take both into account.
Many years ago I was using 2N918 transistors in an RF application. That transistor comes in a size TO-72 package, which has four leads, one each for the collector, the emitter, the base and the transistor’s can. Design and drafting was having layout problems and wanted to cut off the lead for the can connection. I contacted the vendor to inquire about doing that.
“No! Do not do that,” came the reply. Cutting away that connection could induce a mechanical impulse into the device from which the semiconductor chip could be damaged.
A mechanical impulse leading to chip damage. Hold that thought.
Someone came to me just recently to discuss a power supply problem in which one particular diode would fail shorted following a mechanical drop test of the unit. Over and over again this would happen. Thoughts of that 2N918 issue came rolling back.
Since mechanical impact seemed to be the failure trigger, I suggested taking a look at how rigid or how flexible the connection was to the diode’s wire terminal. Please see Figure 1.
Figure 1 Alternative connection methods for a mechanical drop test.
If that wire were stiff enough to support one wing of the United Nations Building, it could be carrying mechanical impulses to the diode’s package during the drop test. Those impulses could then be transmitted to the semiconductor chip itself which would not likely withstand the insult.
I recommended that the diode connections be kept flexible and further suggested that using a flexible braid instead of a rigid wire would accomplish that.
Sometimes it’s the simplest things. But I never heard anything more about failed diodes.
This article was originally published on EDN.
John Dunn is an electronics consultant, and a graduate of The Polytechnic Institute of Brooklyn (BSEE) and of New York University (MSEE).