jFET matching... how close is "close"?

[Vallhagen]

Hi folks:)

I have browsed around a bit, and i have measured Vgs for a bunch of recently bought 2N5457 jFET (see table below).

Now, when it comes to matching them in pairs (or quads), primarily for phaser builds (i'm just about to build my first Phase 45 / mb Smoothie), how close must the Voltages be? As said, i have browsed around a bit, and i can read all over the places that "they must be close", but i cant really find recommendations about which range that stated "close"  really means... is -0.940 V (enough) close to -0.960 V? etc.

I would really like someone to make a statement here... like "no more than x.xx percentual diff!"

Here is my bunch of 2N5457's ... sorted by voltage:

jFET nr   jFET Vgs
   

3   -0,350
7   -0,503
2   -0,584
5   -0,623
4   -0,647
6   -0,749
1   -0,800
12   -0,808
8   -0,824
31   -0,882
16   -0,910
24   -0,915
38   -0,918
18   -0,921
11   -0,926
21   -0,936
32   -0,936
29   -0,971
34   -0,980
23   -0,983
37   -0,985
15   -1,010
35   -1,013
20   -1,014
14   -1,016
30   -1,036
27   -1,037
19   -1,041
22   -1,042
26   -1,049
25   -1,057
33   -1,066
17   -1,074
40   -1,100
28   -1,147
13   -1,243
39   -1,264
36   -1,285


Cheers and thanx in advance for hints n help:)

[DutchMF]

Hey Vallhagen, first off, I'm no expert on this, but I bought some matched JFETs from Jacob Kokura, who labelled them with the voltages, and there is no more then maybe 5 mV's between them. Looking at the voltages you posted, there are more than enough that close for a Phase 45, and even a quad for a Phase 90 should be possible!

Paul

[jkokura]

Here's how I'd group them.

3   -0,350

7   -0,503

2   -0,584

5   -0,623

4   -0,647

6   -0,749

1   -0,800
12   -0,808

8   -0,824

31   -0,882

16   -0,910
24   -0,915

38   -0,918

18   -0,921
11   -0,926

21   -0,936
32   -0,936
(Best Pair)

29   -0,971

34   -0,980
23   -0,983
37   -0,985
(triple)

15   -1,010
35   -1,013
20   -1,014
14   -1,016
Best quad

30   -1,036
27   -1,037

19   -1,041
22   -1,042

26   -1,049
25   -1,057

33   -1,066
17   -1,074

40   -1,100

28   -1,147

13   -1,243

39   -1,264

36   -1,285


Cheers and thanx in advance for hints n help:)

So, 8 pairs, 1 triple and one quad. Not bad. Could be better, but you usually have to get up over 100 before you can eliminate all those singles. I'd keep the singles, get another batch of 5457's and do some more matching. But you have enough pairs and a quad to get a couple phasers going at least.

jacob

[jkokura]

Think about it this way - you got 8 pairs, which is 16 out of 40 transistors - that's %40. Also, one quad makes up %10 by itself. You managed to get %50 of your transistors to match up with something. That's actually pretty good for a small batch.

Jacob

[Vallhagen]

Thanx guys for replying:). Exactly the kind of answers i was hoping for!

And a special thanx to you Jacob, for suggesting the grouping. I'd keep that as an advice of how to make pairs/groups in the future too.

...and yes, it looks pretty good for just ordering 30 by chance(i had 8 at hand when i placed the order, none even close to make a pair).

Cheers!

[jkokura]

Oh yeah! I forgot one note I should have made.

#38 is actually capable of making a triple with either pair in front of it. I would have grouped it with either. You could potentially make another quad there, either by taking the #24 and grouping it with the higher two, or by taking #18 and grouping it with the lower pair (plus #38 obviously).

Jacob

[croquet hoop]

Wouldn't 30, 27, 19 & 22 make a fine quad too ? (1,036, 1,037, 1,041, 1,042)

[jkokura]

Yep, that's not as close as I aim for, but it's close.

Jacob