Kind soul out there know the Tek 576 Curve Tracer well?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Mbira

Well-known member
Joined
Jun 4, 2004
Messages
2,422
Location
Austin, TX
We've got one at work, and I am trying to match FET's. Yikes, I don't really know what to do with this beast! I put together PRR's FET jig, but it took about an hour to match 7 of them (including building the jig) and it seems like I should use this awesome thing here for such things...

Joel
 
They are a bit formidable. The main issue for me with them is that, being just an oscilloscope output, you have to play games with templates and grease pencils etc. to do comparisons, unless you have a really superb photographic-like memory.

Most of the ones I've encountered have serious screen burns at the origin too, from careless user abuse.

They are also going to be limited in accuracy, just from how well you can read stuff off the screen.

I missed PRR's fet matching fixture---could you point to it?
 
I think we have that type as well - not sure about the type-no. As long as its knobs are still attached firmly enough we can operate it.

But I don't know for which circuit you're matching.
The several jigs presented here are better suited I figure, for reasons mentioned above.

I've only tried this one so far, and it was fine (use a low offset opamp):
http://www.geofex.com/Article_Folders/fetmatch/fetmatch.gif
 
Thanks Joel.

When I was working on the first version of what was to never get finished as the computer-controlled attenuator for the H*rm*n R&D subjective eval facility, I had a jig that servoed the temp of the FET under test, then measured the pinchoff voltage and saturation drain current. At the end of this exercise we did a two-dimensional sort and made up matched pairs, which were then embedded in aluminum blocks with heaters and temp sensors to servo the temp at slightly above ambient. The 2SK364 FETs were all from the same date code to begin with so that made it easier.

I didn't know about the 2SK389 at the time, although having the two parts totally isolated does have its advantages, among them no shared substrate with its gate-to-gate variable coupling capacitance.
 
Huh?
:green:

Well, I guess I may have been using the curve tracer correctly then, it just gave me a bunch of vertical lines that were spaced acording to cutoff voltages (?) things were pretty hard to see the differences with and I think I was definately getting better results with PRR's method-it just took a little time. I was using a trim-resistor, so I'll just switch it out with a pot and that should speed things up considerably.

Joel
[/quote]
 
Joel;

I matched several hundred 2SK170 at a friends house
using a fixture like PRR that had 4 test sockets.
I matched to better than 1% VGS
Piece of cake. Boring though.

That type of matching is by far faster.
I have a Tek 570 (the one for matching vacuum tubes)
and it is a pain to use like that. Great for design and load lines
and so see what the tube can do but once you know that if
you can be happy in your on mind about matching at one data point
then do it at a realisitic operating point that your device will see in the
circuit you will be using it in. Then match to that one data point.

Method
Dump the fets on a table I suggest using a ESD controlled surface.
place them in the fixture and do a rough sort. I.E. sorth them into
10 to 15 seperate bins. Label each bin with a measured value
and if the device falls within that value put it in that bin.

Then go back and retest each bin.
I find that Pink ESD safe polyethlyene foam works great for this.
Tape a paper label to the edge of the foam.
Write on it with a pen.
Lets say you were matching Idss and the large bin above had
6.0 to 8.0 ma parts in it.
Label the paper from 6.0 to 8.0 in 0.1 increments.
Measure the fet and stick it into the foam in the area near the label
to what it measures. When you are done you will have 20 or so fets
arranged from left to right in increasing Idss currents.
then try to use the devices next to each other and do a third round of matching. Use the devices next to each other to get double or quad
matches of your parts. Some people match to 5% but I had no problem
getting to the 1% mark. When you are doing that step it is very good
to have a least 4 sockets with 4 fets. and spend a least 2 minutes
letting the fets warm up and stabilize. It may be helpful to use pliers
so you hands do not warm the fet up. Be sure to do this in a very
temperature stable environment.

When I buy more fets I rough match then and dump them into the big bins. I do not keep the fets in a row on the foam I just repeat the steps 2 and 3 with the older and newer parts and I hope I find a fet in the bins that now match in pairs.

The more parts you start out with the easier it is. If you need a quad
match of parts you need to start out with a least 100.
So a real bad rule of thumb is buy 50 if you need 2 really good matched parts. Using this method will keep the FET manufacturers happy and maybe our beloved parts in production.

It is a mind over matter thing.
Do rough sorts and repeat many times till you get pairs or quads.
 
Curve tracers have uses.

But for sorting JFETs, it is sufficient to know the Ids and Ggs. For common JFETs, especially all from the same process, the curve between these two values is fully knowable from these two points.

And when sorting for a specific project, just mock-up that part of the project. If the supply voltage is higher than the likely Vgs, most FET projects reduce to a supply voltage (9V is often ample) and a source resistor, with gate grounded.
 

Latest posts

Back
Top