Author Topic: N-Channel JFET Vgs(off) really wide variation in datasheet.. how to deal with?  (Read 2189 times)

0 Members and 1 Guest are viewing this topic.

Offline K3mHtHTopic starter

  • Regular Contributor
  • *
  • Posts: 98
  • Country: ca
So I'm trying to spec a JFET for low quantity production (20-100) and while my prototype works fine and simulations work fine.. I'm noticing that in the datasheet for any JFET that I'm considering, the range of Vgs(off) is crazy wide
J105: -4.5 to -10V
J106: -2.0 to -6V
J107: -0.5 to -4.5V

My circuit works if the Vgs(off) isn't any lower than -3.5V but as you can see, none of the parts listed are guaranteed to be in that spec.

Is this a worry that I'm making up in my head? Should I find a different part (any suggestions.. hand-solderable SMT please)? Or should I add some circuitry to "condition" the input voltage to guarantee that it goes low enough to turn the JFET off - I kinda hate adding parts..


 

Offline unitedatoms

  • Frequent Contributor
  • **
  • !
  • Posts: 324
  • Country: us
Consider that part number is very ancient. That may mean that manufacturing story is not very precise.
Whatever cut off values parts got after being manufactured were binned into different part numbers.

Yes, you me want to choose a different part. With may be 4-5 digits in part number spelling. If you need just a switch, then may be mosfet is easier choice than jfet. Like 2n7000
Interested in all design related projects no matter how simple, or complicated, slow going or fast, failures or successes
 

Offline Calvin

  • Regular Contributor
  • *
  • Posts: 179
  • Country: de
    • CalvinĀ“s audio page
Hi,

if the circuit allows for You could choose a JFET with a higher Idss and use a source resistor to reduce to the required drain current value.
The tolerance spread will be reduced accordingly.
The Jxxx JFETs are almost all intended for ccs application.
They are already grouped in Idss batches ... differently to say the 2SKxxx whose Idss-group is indicated by a suffix -L, -G etc.

regards
Calvin
..... it builds character!
 

Offline Gerhard_dk4xp

  • Frequent Contributor
  • **
  • Posts: 353
  • Country: de
Yes, that is a badly controlled parameter in FETs.

https://www.flickr.com/photos/137684711@N07/37321004540/in/album-72157662535945536/    >
<   https://www.flickr.com/photos/137684711@N07/36909661793/in/album-72157662535945536/lightbox/     >

Some are more consistent than others.
<   https://www.flickr.com/photos/137684711@N07/41639547720/in/album-72157662535945536/lightbox/    >

At Infineon, you can get them pre-sorted. But you get only the same group on a reel; you cannot specify
which group. I have made these plots because I was looking for an easy Cascode transistor that leaves
consistent Vds for the CS stage.

<   https://www.flickr.com/photos/137684711@N07/42542542745/in/album-72157662535945536/lightbox/   >

regards, Gerhard
 
The following users thanked this post: exe

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
To not mince words, the circuit was designed wrong -- it doesn't allow for the full variation in parameters of the components chosen, and now you must, in effect, create your own part, or locate a part with tighter specs that's otherwise equivalent.

JFETs are largely distinguished by junction size, and by process variation (which gives the equivalent trio of Vpo, Rds(on) and gfs variation together).  (Voltage rating counts too, and polarity, but with >40V and P-ch parts being almost as rare as hen's teeth anymore, this is secondary.)  JFET part numbers are further distinguished by selections for various parameters: Vpo or such; Rds(on); small signal or RF parameters; or noise (at LF or RF).  That's about it.

So, basically, what I'm finally getting at is, by receiving loose parts and sorting them by Vpo, you're effectively making your own internal part. :)  Individual part qualification testing is terribly tedious, but it can certainly get you through your run of a hundred, until circuit changes can be made.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline exe

  • Supporter
  • ****
  • Posts: 2647
  • Country: nl
  • self-educated hobbyist
I have same problem, although I'm going to make only two boards :). I use it to generate bias current and I'll simply add a trimpot.
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 15151
  • Country: de
JFET generally are widely scattering in properties, though usually with quite some correlation, so manly 1 relevant parameter for a given geometry.
One may be able to get closer specked parts under a different number, like the 2SK... or 2N... series.  Sometimes they do offer subgroups.
The parametric search of the suppliers may be your friend to see what is actually available in small quantity (e.g. less than a full reel).

For just a small quantity one could take the gamble and order a few and check if they a re suitable or adjust a resistor value accordingly.
 

Offline edavid

  • Super Contributor
  • ***
  • Posts: 3464
  • Country: us
OP, you left out some important details.  Is this VP limitation due to your supply voltage?  What are your IDSS requirements?

Lacking that information, I would suggest BF861C.
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
JFET switching circuits are not suitable for low supply voltages unless the JFETs are selected for low Vgs(th).  In linear circuits where it matters, pairs of matched JFETs can be used to compensate for the large variation in Vgs(th) or sometimes more complicated methods are used.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf