An overlay transistor cautionary tale

-October 24, 2017

My first encounter with an overlay transistor was with the 2N3866, a very widely used device, especially in television and video work. The parts I was using were made by RCA, who, if memory serves, was the company that invented the overlay design sometime around 1964. Although historians might correct me about the exact date of introduction, that was the general time frame.

I was doing my stuff with those transistors in 1969. The key to an overlay design was a transistor structure with a very large number of emitters that would all be brought together at one single lead of the transistor's package, the (ahem) "emitter" lead. I don't know the underlying semiconductor physics, but from what I read, that emitter structure lent itself very well to RF power amplification.

I proved that point to my personal satisfaction one day when I tried to use a 2N2219 in a +7 dBm output level RF power amplifier for 10 kHz to 40 MHz service. No way!! The 2N3866 worked just fine in that particular circuit but the 2N2219 was useless.

There was a bigger cousin overlay transistor that I got involved with as well, the 2N3375. I had four of them arranged in a push-pull, transformer-coupled amplifier delivering +30 dBm from 10 kHz to 40 MHz and running in class-A for linear service. This was in the US Navy's CVA VAST signal generator, Building Block 21.

Using RCA devices, the amplifier worked very nicely but, when I tried to use some devices from another company who shall here remain nameless, my amplifier was a-huffin' and a-puffin' to deliver my required power level and bandwidth. To be blunt, it was not workable.

Somehow, I found out that the RCA version of the 2N3375 had 100 emitters in its overlay structure, but the other company's version had only 70. Both of those companies are long gone from the overlay scene so far as I know, but the 2N3375 is still being offered by Microsemi who, although they do not use the word "overlay," describe their part as follows: ".... employs a multi-emitter electrode design."


A 2N3375 data sheet by NJ Semi-Conductors

This whole cautionary tale is a reminder to always check out the actual transistors you will be using in your real-world products and not to just rely on SPICE simulations.

For example, the venerable 2N3055, which dates from many, many years ago, was once written up as being improved right out of one fellow's power supply design as the transistor's ft was increased over the years as semiconductor production "improvements" were made so that a once-stable power supply design began to oscillate at RF frequencies.

Similarly, I had 2N6578 problems when changes to semiconductor manufacturing procedures adversely affected device speed in the high voltage power supplies I was involved with at the time. Please see: Semiconductor Evolution.

It's almost half a century since my overlay adventures, but eternal vigilance is mandatory.

John Dunn is an electronics consultant, and a graduate of The Polytechnic Institute of Brooklyn (BSEE) and of New York University (MSEE).

Also see:

Loading comments...

Write a Comment

To comment please Log In