Matching a large number of jfets - best howto

Hi all,
I'm considering binning several thousand jfets with a view to selling matched sets. Given the scale of the task, some thought around best practice seemed wise.
At one end I could buy a curve tracer and individually mark every part, in a climate controlled environment... at the other end, I could sit at the kitchen table, measure Idss and bin in particular ranges, say to 0.1 mA.

What would be the 'pragmatic best practice' for this sort of task? In a home environment...

Cheers,
A.
 
Cheers ZM, in the back of my mind I'm wondering if I test at a notional current, or build a jig that makes it easy to do say Vp and idss and then bin in a 2 dimensional way.

I'd hate to get to the end of it all and have someone suggest a minor time addition that would make a big difference to saleability.:ROFL:
 
Last edited:
more time you spend on them, greater story you can write , greater price you can ask

:clown:

simple full Idss test is good enough for me , thinking as buyer
9 to 12Vdc as PSU , just take care to get a reading when numbers are steady (temp. equilibrium)
 

Attachments

  • JFETMatching.gif
    JFETMatching.gif
    2.6 KB · Views: 272
Hahahaha. Indeed :)
Idss matching has been good enough for me, and certainly is fastest in the end.

I suspect after 2k or 3k of them I'll be heartily sick of any additional steps...

I'm recently unemployed and I thought this might be a way to make a bit of pocket money whilst interviewing. Going to the beach also seems pretty compelling... :)
 

ezavalla

Member
2008-12-08 11:53 pm
Hahahaha. Indeed :)
Idss matching has been good enough for me, and certainly is fastest in the end.

I suspect after 2k or 3k of them I'll be heartily sick of any additional steps...

I'm recently unemployed and I thought this might be a way to make a bit of pocket money whilst interviewing. Going to the beach also seems pretty compelling... :)

As soon as you have measured ten of them you will need some kind of software to tell you which of them you'll need to pack together.
 
As soon as you have measured ten of them you will need some kind of software to tell you which of them you'll need to pack together.

Not really. I'd thought to bin then by some arbitrary range, say 0.1mA and anything in that bin is within 0.1mA of each other. After 2K of them if there's some outliers, then so be it.
Of course I could bin by the 0.01, and have a truly enormous collection of bins (for 4mA variance that would be 400 bins vs 40 for 0.1), but then I'd have to decide on the time/additional price charged math.
 
Test them with a 12v wall wart or power supply so voltage stays consistent, with a DVM in series. Let them heat up until the measurement stabilizes to somewhat simulate in circuit conditions. May take a minute or so but results are pretty accurate. They will change over this period and some more than others. What you measure initially and after they warm up can be significant. Consistent room temp is important too.

TJ
 

Nelson Pass

The one and only
Paid Member
2001-03-29 12:38 am
BTW, perhaps I should share a couple observations about testing large
quantities.

Obviously test conditions should be relatively constant. Room temperature
is a factor also.

My test rig settles in about a second, and I take the voltage then. The
value will drift as the seconds go by, but it works if you are consistent.

Obviously high Idss values will heat more, so at values above 15 mA
or so consider using the Vp pinch-off test instead.

If you are doing .1mA resolution, it's best to sort them into 1 mA bins
and then retest those bins at .1mA. Gets the human and other error
factor way down. Once you are confronted with accidentally tossing
a fet into the wrong bin which has a couple hundred parts, you will see
that it's worth the extra effort.

:cheers: