Ode to Lateral MOSFETs

[megajocke]

One interesting property of the laterals is that not only do they have the negative temperature coefficient, there also seems to be some kind of gate clamping when the transistors overtemperature. At about 160 degrees some kind of thyristor fires and shorts the gate to the source, protecting the transistor. I tried to destroy one on a big 50V power supply but didn't succeed.

I did a little experiment some time ago:
Lateral MOSFET torture...

OTOH, as MJ21195/MJ21196 are $2/pc and FJL4215/4315 (Fairchild equivalent to 2SC5200/2SA1943) are $1/pc it's cheaper to build massive overkill output stages instead.

[Mooly]

How about a current dumping MOSFET amp. Laterals for the low power section and HEXFET's for the dumpers.
Now that would be a challenge.

[peranders]

MiljonJocke, the zener is actually a real zener of +-14 V. Other mosfets have not this.

[suzyj]

Originally posted by megajocke
OTOH, as MJ21195/MJ21196 are $2/pc and FJL4215/4315 (Fairchild equivalent to 2SC5200/2SA1943) are $1/pc it's cheaper to build massive overkill output stages instead.

But it's not, really, because to get the same sort of slew rates with lots of vertical fets, you've got to drive the gates really hard, meaning your driver circuit is significantly more complicated, and burns a lot more power. Then there's issues with current sharing due to the positive temperature coefficient, etc.

The cost of the output pair is peanuts compared to the cost of everything else, anyway. It's PCBs and heatsinks and transformers that really cost money.

[megajocke]

Peranders: You mean the internal zeners? I don't believe those will be enough to protect the transistors from blowing up except protecting the gate oxide. I've heard of the bond wires failing and this I guess is due to high currents when no external zeners are used. 10V seems like a value commonly used. Do you have any idea where this thyristor action on the gate on overtemperature comes from? It's quite interesting.

suzyj: Those are bipolars :P Vertical fets seem just like a PITA with their temperature coefficients and gate voltage spread. Some manufacturers (QSC for example) use verticals with each transistor having it's own feedback circuit with an opamp to equalize currents. Bipolars and laterals seem much easier... And some high current FETs actually have second-breakdown-like characteristics!

If you only have one pair I agree it might not be much of the total price but if the amp needs 8 or 10 pairs per channel it starts to get expensive with the laterals! The self-protection might not be as big of an advantage if they are this expensive considering the cost and performance of modern BJTs. Still, maybe one likes the sound of the lateral mosfet amps - then that's a good reason to use them.

[Lumba Ogir]

suzyj,

I can't see a good reason for the 0.5A Id(max) rating.

Have you checked the Rdson values?!

[suzyj]

Originally posted by Lumba Ogir
suzyj,

Have you checked the Rdson values?!

They're not listed in the datasheet for either device, nor are they particularly interesting, I think. These aren't designed as switches, after all.

The Id curves only go to Vgs=3.5V, so the people writing the datasheet are clearly of the same opinion

[jacco vermeulen]

Rds(on) = 10 Ohm for the TO-220.

(Mr Peter Walker would turn around in his grave if he heard talks about fast current dump devices)

[Mooly]

Originally posted by peranders
MiljonJocke, the zener is actually a real zener of +-14 V. Other mosfets have not this.

The reason they don't have this is to stop the device from experiencing a latching action if ever they were triggered.

[megajocke]

Why would zeners latch? Zeners clamp. (Something is latching though in the laterals)