Distribution of Idss in a bunch of random 2SJ74?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I ordered 40 2SK170GR. I was able to get 16 matched pairs but Idss values range within the valid GR range. See overlaid graph of curves for all 40 Jfets. Second picture is 2 pairs matched, 4 jfets. I did not paid much attention to the specific Idss (vertical not fine calibrated) but dividing the vertical by 3 gives the approximate Id values. On the pairs shown, there is roughly 1ma difference between the pairs.
 

Attachments

  • 2sk170 all overlayed.jpg
    2sk170 all overlayed.jpg
    483.9 KB · Views: 370
  • 2sk170  2 pairs matched.jpg
    2sk170 2 pairs matched.jpg
    373.3 KB · Views: 366
I ordered 40 2SK170GR. I was able to get 16 matched pairs but Idss values range within the valid GR range. See overlaid graph of curves for all 40 Jfets. Second picture is 2 pairs matched, 4 jfets. I did not paid much attention to the specific Idss (vertical not fine calibrated) but dividing the vertical by 3 gives the approximate Id values. On the pairs shown, there is roughly 1ma difference between the pairs.


Those curves don't say much because they are overlaid so much.

Maybe you could post just the "2SJ74s" like one typical, and one typical of the 170s?

Same settings...

_-_-bear
 
Those curves don't say much because they are overlaid so much.

Maybe you could post just the "2SJ74s" like one typical, and one typical of the 170s?

Same settings...

_-_-bear

That was the point, to show that the curves are all over the place (forming almost a solid forground) but still within spec for Idss. I was able to ger 16 pairs, but i doubt i can get more than a couple of quads.

I dont have 2sj74s, i am building a simetrical b1 buffers.
 
I would expect the curves to not coincide.

What is most important is to see if they are genuine parts or not.
Based on the physical similarity between the currently available "170" and the "74" parts I am skeptical of the K170 parts. However, they appear to be much closer to the expected curves when I measured the ones I purchased.

Super matched (virtually identical curves) may not be terribly important in most buffers due to the feedback that is usually applied to get it to unity gain...

_-_-bear
 
Why don't you simply check for transconductance?

I'm not aware of any part that offers the IDSS and the transconductance of the K170/J74, which is exactly the reason there is no 1:1 replacment. There are some parts that have the same IDSS (that could be relabeled), but much lower transconductance.

If the part meets IDSS and transconductance, then the only thing in question that remains is low noise and that is truly not easy to check with simple means. The fakes I've seen so far are very cheaply made with not much of an effort to hide it.

By the way, now and then it happens that one jfet is slightly out of spec, but that's more like 13-14mA at most. I haven't seen a 17mA BL part. Even big semiconductor factories are not perfect. Any case as long as transconductance is ok, use it!
 
@bear: ;)

do you have a tip for us how to determine the transconductance "on the kitchen table"?

Transconductance is change in drain current for a given change in gate-source voltage; means measure IDSS (VGS=0) first, then insert a source resistor (VGS!=0) and measure ID. Divide the differences and you have an approximative transconductance value.

It's not so straight forward as transconductance depends a lot on the drain current, but this way should be sufficient to distinguish a genuine high transconductance jfet from a low transconductance part with both having the same IDSS.

Alternatively (but less practical), one could measure pinch-off voltage (when the jfet stops conducting), because that is very different also; for K170 it is about -1-2V VGS or so, for long channel low transconductance parts about -3-4 VGS or so.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.