Breakdown voltage

Are you measuring Vgs breakdown or Vds? If Vds, there is a creepage or drift effect when avalanching:

What is the mosfet gate connected to?

What is the rated voltage of those mosfets?

Interesting. Nice test set!

Various semiconductors have long-term trends. Bipolar damage from b-e zenering. Strange gate effects in GaN fets.

I've found that most mosfet gates crater somewhere around 70 volts, and fail hard shorted. The zener-protected ones tend to limit in the low 40s and of course survive at modest gate currents.

Often datasheets fail to mention that. When they state +/-8V or +/-12V as Vgs max that indicates zener protection but one cannot be sure unless also a max current such as 10mA is stated. I have on occasion asked manufacturers and they did not know!

On a sunny day (Wed, 4 Dec 2019 17:14:47 +0000 (UTC)) it happened snipped-for-privacy@math.uni.wroc.pl wrote in :

No, but what is your gate drive circuit?

A slight drain-gate injected charge can at power up lower the drain source impedance when the charge slowly leaks away you see the real avalanche?

Gate is shorted to source. One is mared IRF540, so nominaly 100V. The second FFQP50N06 so nominaly 60V. Both came from Chinese selers and definitely are fakes. Partial goal of this execise with measuring parameters is to find out what they (and other) can do.

One extra clarification: I have batch of 20 marked IRF540, tested two and they behave essentially the same. Similarly, tested two marked FFQP50N06 with essentially the same result. Mosfets from other batches behave differently, if there is change of results is too fast to notice using mulimeter.

Vds, with gate shorted to source.

Interesting. But what I see is repetitive: if I let mosfet to rest for short time and repeat the test I again see climbing readouts. IIUC Infineon writes about irreversible change.

That is a long wire. Are you sure it isn't oscillating? Can easily be

Maybe it is oscilating. However, gate capacitance is more than 1nF, so assuming wire inductance 0.2uH we get parallel resonant circuit between gate and source with resonant frequency of order 3.6MHz and reactance of order 4.5 Ohm. Wire has resitance of order 0.1 Ohm and gate probably more, so Q is moderate or low. Obvious way to supply energy to this circuit is via gate-drain capaciance, but this is negative feedback. Other ways of coupling are likely to be less efficient, so I would expect this to be stable.

On a sunny day (Thu, 5 Dec 2019 15:45:20 +0000 (UTC)) it happened snipped-for-privacy@math.uni.wroc.pl wrote in :

You can stop RF oscillations in circuits like this by adding a 4.7nF capacitor from drain to ground.

Or ... a wire bridge :-)

Just don't use 20cm of wire. There is no reason for it.

I've seen a few slow semi effects. Transistor b-e damage from zenering, GaN gate changes from high forward bias, strange IC failures after a year or so of operation (MAX9690). All three of those could be fixed, to some extent, by high temperature annealing.

After you characterize the fet phenom, you might put them aside for a while, or bake them at, say, 120C overnight, and measure again.

Some transistors, used as emitter followers, will oscillate fiercely with both the base and collector well bypassed. TO5 cans, with long wire bonds, liked to do that. 2N2219 was a classic.

Yes, I saved a note from your Sept 8th, 2006 post.

I'm doing some measurements on the SAV541 phemt. Maybe I'll see if there is a gate damage effect, like I saw on the EPC GaN parts. Probably not, different physics.

I might make a diff pair from them, and I wouldn't like the gate thresholds to change over time.

Anything you learn about SAV541, we'd like to hear.

A while ago I was given an audio test oscillator (build from an ARRL ham-handbook project description) and was told "It doesn't work right". Twin-T design (with a cap to ground at each base), real metal-can 2N2222 transistors with nice long leads. The oscillation frequency bounded all over the place when I waved my fingers near the circuit.

A scope and spectrum analyzer agreed - it was squealing up in the

A ferrite bead around each base lead fixed the problem.

Here's a whiteboard pic.

Unusually for RF parts, the Mini-Circuits data sheet has DC curves, so I don't need to do that.

I measured 0.4 pF drain-source capacitance on my AADE meter, gate grounded.

I need to find the drain blowout voltage, after some other tests. SAV551 dies at about 11.

The 541 is called a low-noise amplifier, but it's kind of a brute. Nice switch. Two of them can make a really fast NOR gate.