If a power MOSFET begins to oscillate, would heat develop and be trapped inside the polysilicon channel?
There was this fellow, Rudy Severns, who many years ago went on a speaking tour for International Rectifier to discuss "Hexfets," their power MOSFETs, when that product line was still new. If memory serves, this would have been somewhere around 1981. Rudy was not an IR employee; he was serving them as an outside consultant and spokesman.
Rudy described how Hexfets could do wonderful things for switch-mode power supplies, but he quite pointedly suggested that they might be a troublesome choice for making a linear amplifier, or for use in any other non-saturating service.
There were two issues:
- The Hexfet devices had a series resistive path running between the gate pin of the device's package to the gate itself. That resistive path was located within the semiconductor body and consisted of a buried polysilicon channel.
- Hexfets themselves were/are very high-speed devices which, if put into their linear operating region, could rather easily burst into parasitic RF oscillation.
His concern was that if a power MOSFET began to oscillate, there would be heat developed inside that device's polysilicon channel and that there was pretty much no place for that heat to go. The channel could get thermally hot enough to burst through the gate's oxide layer and thereby destroy the MOSFET.
Some time later, Motorola introduced their TMOS power MOSFET devices, so I contacted them and had a little talk. The fellow I spoke with confirmed everything Rudy had said and also told me that Motorola was at the time trying to find a way to bypass the TMOS polysilicon channel with metallization to avoid that trapped heat issue. The problem was that they couldn't find a way to do it without causing vulnerability to ESD (electrostatic discharge) damage to the TMOS gates.
Some more years later, I happened to visit a company that made an AC power supply product that, to my genuine astonishment, used power MOSFETs from Hitachi as linear power amplifiers in push-pull, class-B service.
I saw then that it could indeed be done despite Rudy's concerns, but if you decide to try it yourself, do be careful to guard against parasitic oscillations.
John Dunn is an electronics consultant, and a graduate of The Polytechnic Institute of Brooklyn (BSEE) and of New York University (MSEE).
- Power MOSFET tutorial
- Per-quadrant linear amplifier distinguishes input polarity
- A simple guide to selecting power MOSFETs
- Series-connected MOSFETs increase voltage & power handling
- An introduction to semiconductor technologies for power management, Part 3
- HEXFET-Based Medium Power Amplifier with 65W Output