How to check the condition of a VFET (SIT) Transistor

[DirkSpec]

Hello Everybody,

This is my first post here and here it goes!

I try to restore unicorns as a hobby and I have picked up a VFET amplifier as my next long term educational restoration project and from the get go I encountered a road block.

I have the PEAK DCA75 Pro transistor checker and a very limited curve tracer and it started to claim some of the 2SJ18s that came off the amplifier were bad. Some helpful folks at the DiyAudio forum suggested an alternative method to check these VFETs and according to that method, they seemed fine. However, while I was checking them I have noticed some differences between the ones that are marked good or bad by DCA75 Pro. You can see what is going on from the YouTube video that I have prepared while doing this: https://youtu.be/nwpo50oJB98

I was wondering if anybody can shed a light on how to determine if a VFET is good or bad in a definitive fashion? Is buying a transistor curve tracer the only way? Anyway, thanks in advance.

/Dirk

[David Hess]

I knew power JFETs existed but not that they were ever used in audio amplifiers.

The ones with high gate leakage which shows up as faster recovery in your test may be damaged.  There should be a gate leakage specification but I could not find any complete datasheets for these devices.

With the gate shorted to the source, the resistance between the drain and source for the 2SJ18 should be lower than 3.3 ohms but I am not clear about the test conditions.  The gate to drain/source should test as a diode with a forward voltage drop of about 0.42 volts and a reverse breakdown voltage of at least 25 volts.

I would just replace them with power MOSFETs which seems to be the common solution.  Unfortunately there are no suitable p-channel depletion mode MOSFETs.

[rfeecs]

These are apparently "Static Induction Transistors".  Significantly different than regular JFETs.  They have a vertical structure with a buried gate and short channel.  They behave like vacuum tube triodes.

A possible (Japanese) datasheet:
http://www.amplimos.it/images/2SJ18-2SK60.pdf

You could manually test some of the DC parameters with power supplies and meters.

[David Hess]

These are apparently "Static Induction Transistors".  Significantly different than regular JFETs.  They have a vertical structure with a buried gate and short channel.  They behave like vacuum tube triodes.

The structure is different but they are still JFETs with all of their limitations that contribute to high channel resistance.  They are very area inefficient making them expensive which is why vertical MOSFETs replaced them so quickly although that is not a disadvantage in applications where power is limit by die size anyway.  That is why you find power amplifiers which use MOSFETs with much higher voltage and current ratings than required.

Teledyne made "FETRON" solid state replacements for vacuum tubes in the early 1970s which used high voltage JFETs and hybrid construction.  NEC, Sanyo, and Toshiba were suppliers of these parts but there many more American companies which catered to aerospace applications.

[DirkSpec]

These are apparently "Static Induction Transistors". 

Yes, that is exactly right. That is why it is throwing some people off when comparing them to xxxFETs of the modern era.

A possible (Japanese) datasheet:
http://www.amplimos.it/images/2SJ18-2SK60.pdf
 

This is super helpful, thank you very much for digging this gem out! I am seriously considering buying a Transistor Curve Tracer, but my main problem is space. The moment I find a Leader or a Heathkit, I will pull the trigger on it. This document will be super helpful as a reference.

You could manually test some of the DC parameters with power supplies and meters.

I am keeping my old, single channel Hameg HM 307 because all the Hamegs of this era has a basic "curve tracer" option. Although the sensitivity/voltage is fixed, I can check the diode operation and observe the characteristic curve. Is this something you meant by manual testing?

[rfeecs]

By "manually test", I meant attach a power supply and ammeter from drain to source and another power supply and maybe voltmeter from gate to source.  Then you could manually point by point trace out the IV curves and measure pinch off voltage.

The other parameter that might be of interest is breakdown voltage or leakage current.  Again you could apply a voltage from gate to source and measure gate current.

I suspect your transistor tester sometimes shows bad because it doesn't have enough drive voltage to fully pinch these off and you have some variation in pinch off voltage from device to device.

[rfeecs]

By "manually test", I meant attach a power supply and ammeter from drain to source and another power supply and maybe voltmeter from gate to source.  Then you could manually point by point trace out the IV curves and measure pinch off voltage.

The other parameter that might be of interest is breakdown voltage or leakage current.  Again you could apply a voltage from gate to source and measure gate current.

I thought maybe I should mention that you can easily blow up your transistor by doing this if you're not careful.  Too much voltage and current and you cook the transistor.  Or if it starts oscillating, combined with long test leads and you might zap it with high voltage spikes.

Just something to keep in mind.

[DirkSpec]

I have finally picked my courage and built the following circuit for the 2SJ18s. I have a Hameg HM 307 with very limited curve-tacing functionality. I have connected it between D-S. I connected a current limited DC power supply between G-S and controlled the voltage from the Current fine tuning adjustment. For all the VFETs that I have, I can observe the characteristic curves in the first quadrant as I vary the DC voltage across G-S. Depending on the "rank" of these VFETs, I can observe that the characteristic curves are shifting in the horizontal and vertical axes. I am assuming that these are working in one way or another, however, to be absolutely sure about the performance, one needs to have a proper transistor curve tracer, or construct a circuit that has many current and voltage measurement points.

[DirkSpec]

I suspect your transistor tester sometimes shows bad because it doesn't have enough drive voltage to fully pinch these off and you have some variation in pinch off voltage from device to device.

I think you are absolutely right. I am observing that, the curves change quite a bit from one "rank" to another and the ones that would require higher voltage potential are deemed as "bad" by the DCA75 Pro. For a ultra inexperienced newbie like me, a tool like DCA75 Pro calling a transistor "Bad" is the end of the world . Now, I think I know a little bit more than I have started about the VFETs, and understand that they are totally unique and require much in depth knowledge.