JFETS in a strange land...

Attached is an excerpt from Linear Technology's application note #118

I draw your attention to "D5" and "D6" --- Here a 'brace' of JFETs are employed in a so-called 'clamp' arrangement (I tend to favor the term 'clipper' inasmuch as level translation is not the point -- but to continue...) --- I must confess that I fail to see the advantage of such a scheme over a pair of 'fast' diodes?

Granted, the pictured arrangement would feature 'near infinitesimal' leakage current, still I can't 'shake' the feeling that I'm afflicted with a 'mental scotoma' as regards some essential property of JFETs and am, hence, missing 'the big picture'???

Any insight, ideas or idle musing will be greatly appreciated!

Many advance thanks!
HP

 

Diode-connected JFETs are used for D5 and D6 simply because the 2N4393 has extremely low gate leakage; switching speed isn't a factor.

 

As far as I can tell, Confirmed. Diodes with less than 100 pa reverse leakage cost about $5 each in quantities of 500 or more, and I can't seem to find out if they were even available when that Apnote was written.

http://www.mouser.com/Semiconductor...Switching/_/N-ax1mp?P=1ypxw8gZ1ypxw81Z1ypxw8i

 

A Jfet in this configuration simply has a more abrupt transition than a diode. Think of it as an amplified diode...sort of.

 

Clarification.....by "abrupt" I mean with respect to voltage, not to time.

 

I don't see why it would exhibit such a characteristic. What is your justification for that?

 

Well hey! Many thanks for the informative replies! Being, as I am, 'less than au fait' with 'off-label' applications, I've learned something today!

Thanks! Between the two of you, you've brought the most bewildering aspect of this application to light!

Very interesting! -- Looks like a matter for the 'Keysight' -- I'll keep you posted!

Again, many, many thanks!
Best regards
HP

 

Just to shine a little more light...

A common 1N4148 silicon diode (http://www.vishay.com/docs/81857/1n4148.pdf), the one we all use for "jellybean" applications, has a maximum reverse current of 5 μA at 25°C.

I've often used a 1N459 diode (https://www.fairchildsemi.com/datasheets/1N/1N459A.pdf) in applications that required low leakage current; it is spec'd at 25 nA maximum, 200 times lower than the 1N4148.

A 2N4393 N-channel JFET (http://www.centralsemi.com/leadedpdf/2n4391-4393.pdf) has a maximum Igss of 0.1 nA at 25°C, which is 1/50000 of the leakage of a 1N4148.

Here's a Burr-Brown application note that discusses the use of diode-connected JFETs: http://www.ti.com/lit/an/sboa058/sboa058.pdf

 

This doesn't sound right; I've never observed this myself, nor can I find any mention of it thru Googling around in the literature. Where did this come from?

 

Brilliant tidbit of information! Fifty ma continuous forward gate current compared to femtoamps of leakage.
(Referring to the datasheet posted by OBWO549)

 

I used a diode similar to the FJH1100 in the 80's: 10 pA leakage at 15 volts and 25 degC.

 

Now, those diodes cost $11 each in 500 quantity!

 

I think the parts we were using may have been FDH11oo but I can't find an info on the web... I will have to check my old Fairchild data book to be sure.
We never paid anywhere never that much for the ones we were using. They cost more than 1N4148's but not _that_ much. I know that has to be the case because the were used 64 at a time.

 

My 2 best sources say there ain't no such animal.

 

I am beginning to suspect that my brain cells storing the part number have failed or been re-assigned to other valuable information.
I am petty sure that the part -- maybe by a different part number -- is in my 1980's vintage Fairchild diode data book. I am not near my data books right now so the answer will have to wait.

 

I ended looking at an old PCB assembly. The diodes are actually FDH300's.

To keep from hijacking HP's thread any more, I have started this one:
http://forum.allaboutcircuits.com/threads/low-leakage-diodes-some-stories.113663/