Can I use an oscilloscope to match JFETs?

[alparent]

Can I use an oscilloscope to match JFETs?

[PRR]

Yes.

But a voltmeter is easier.

[alparent]

A man of few words.

So using RGs technic with a voltmeter is easeyer?

But how would I proceed with an oscilloscope?
I couldn't find anything on google.

[R.G.]

A man of few words.

So using RGs technic with a voltmeter is easeyer?

But how would I proceed with an oscilloscope?
I couldn't find anything on google.

A lot depends on what you are trying to do. "Matching JFETs" is not very specific a goal.

The voltmeter method derives a voltage which is a good indicator for how a JFET will perform in a phaser. It was specifically aimed at being simple to set up and to give one and only one number - that voltage.

You can, of course, use an oscilloscope for measuring voltages. But that is much like using a television to measure the speed of gazelles. It can be done, but is difficult, needs a lot more setup, and then a lot of interpretation of the results.

I suppose more to the point, the voltage measurement was intended to be a "one-parameter" match. It is by no means a perfect matching scheme, as at least one critic expounded on at great length here. It matches only one thing, not several. However, each additional thing you add to the list of what is necessary to call a "match" makes it both more complicated to do, and makes many, many more JFETs tossed out because they don't match.

One way I've done matching on an "oscilloscope" is to use the scope as a curve tracer. This gives you a two-dimensional picture of the response of the device to changes in voltage and control/drive. This is great if you want to match two and only two devices for use in a low distortion, low-offset amplifier, but since each device's measurements are a picture, not a single number, the data to be kept to evaluate for matching explodes. And even better characterization would use this two-dimensional plot and extend it to a set of three-dimensional surfaces to include more parameters for getting a more identical match.

In my opinion, this leads not to better matches, but madness. If the match of devices has to be that perfect, one should be using monolithic devices - several devices on the same piece of silicon - not discrete devices.

And that leads us back to: what things about them do you NEED to match?

One could add a few measurements - Idss is a good one, pinchoff voltage is another, Rds on, forward transconductance, etc.. Each one requires a new setup - or that curve tracer, which is a very specialized scope, or an adapter to a normal scope. But these are somewhat interrelated, and produce reams of data to sort through.

And I think that's one interpretation of what Paul meant... 

[toneman]

you could use a transistor curve tracer attached to a scope to look at the "curves" of the transistor.

matching the "curves" is matching the linear characteristic gains of the transistors.

I have a Heathkit IT-1121 Curve Tracer.  It measures all types of transistors.

a jfet is voltage-based where a conventional bi-polar transistor is current-based.

google "transistor curve tracer" and you will find many simple circuits.

but, as the man of few words said, R.G.'s technique is quick, cheap and dirty.

T

[alparent]

Thanks for the lesson guys.