N(o) channel JFET

[Alex Eisenhut]

Quite sad, I blew up a dual N-channel JFET in my 1A4 scope plugin. I took it apart to see what a dual transistor looks like, it's just two die.

I guess that's what they mean by two monolithic JFETs. Are the two transistors from adjacent areas on the wafer?

Apart from the thermal matching, what's the point of putting two separate transistors in one case? Can I just glue two super-small SMT JFETs together to do the same job? The only difference I see is that the original part (2N5546) seems to have the two gates somehow tied together although I don't see how. The case?

It just occurred to me that the mounting post for the die is the gate pin, is there some hidden junk inside that post? Why the two diodes drawn in the datasheet?

[SeanB]

As the 2 dies come from the same wafer, and are made in the same small area using the same dies they will be very well matched ( typically to better than 0.1% ) in all parameters, so they will track very well in a circuit.

The way a Jfet is made is that the gate is the substrate, so will have it's own pin, and in these the gate and the whole fet is in a P channel well buried in the silicon dice, so that there is a parasitic diode created between the gate connection of the Jfet and the bulk silicon, which reduces capacitance of the device to the substrate and the case. In this case you leave the case floating, and it will sit somewhere around the most positive gate level. It does give good matching, and even more critical good matching with temperature, exactly what you want in a differential amplifier, and the voltage range of 50V is pretty good as well.

[Alex Eisenhut]

I'm having a hard time understanding the connection shown between the two gates in the datasheet. I don't see where it comes from physically, or what it does electrically.

Anyhow. The substitute part (ECG461) doesn't work in the 1A4, but two MPF102s does. Picks up some stray high frequency 60ishMHz junk though.

I can't see the path for that feedback either, should I wrap the TO-92 cases in copper foil? That's all the case of the TO-71 does AFAICT.

[Paul Moir]

You know Sphere has a used one for sale?
151-1009-00 right?
http://sphere.bc.ca/test/tek-parts/tekparts2.html

[Alex Eisenhut]

Thanks but I ended up needing 3 and anyhow, looks like 2N3819s glued together do the trick. Maybe I can find some tiny ferrite beads to try to suppress that oscillation.
For my next trick, I also managed to blow up the dual diode input protection circuit...

[Paul Moir]

Cripes, what are you plugging that thing into? 

Good news then.  I wonder if you could dunk them in some conductive epoxy or something to help shield them.  Or maybe just tape them together with copper tape (which would help with the thermal matching too I suppose?)

I've got a wonky matched NPN transistor from an old HP power supply to deal with myself.  Fortunately it's much less critical a part.

[Alex Eisenhut]

Cripes, what are you plugging that thing into? 

Good news then.  I wonder if you could dunk them in some conductive epoxy or something to help shield them.  Or maybe just tape them together with copper tape (which would help with the thermal matching too I suppose?)

I've got a wonky matched NPN transistor from an old HP power supply to deal with myself.  Fortunately it's much less critical a part.

It was my fault. As I dug in to get to channel 3, I left the ch4 leads dangling and when I applied power I zapped the +12/-15V supplies together.

Not my finest hour 

At least I have signals again!