how to match devices -help-

[Stefanoo]

I'm searching on the forum how to match transistor-jfet-mosfet.

I would like to know...if it is possible do it on your own.
If so, is there any circuit test that i can build and that work fine for the purpose?


hope to receive ...as everytime...your good advice!

best regards,
Stefano.

[taj]

I seem to remember Prof. Leach showed a simple circuit (or two) you can use to measure and match BJTs with. But after a quick look, I can't find it on his website.

http://users.ece.gatech.edu/~mleach/lowtim/

[EDIT] Found it! It's here...

http://users.ece.gatech.edu/~mleach/lowtim/part2.html

..Todd

[AndrewT]

Hi,
Pass and ESP both show methods for matching Vbe and/or hFE for BJT and Vgs for FETs

[taj]

Now I have a question about that Leach circuit. And I am a total newbie, so go easy on me.

I thought we were trying to measure/match current gain, but we're measuring the current through the base segment. Wouldn't it be better to provide a constant current to the base and measure the current gain on the collector/emitter segment? I'm sure there must be a simple reason for not doing that.

..Todd

[AndrewT]

Hi,
most amps using BJTs have a fixed collector current controlled by a fixed base emitter voltage (don't always believe those that say a BJT is a current to current amplifying device - but that's a different argument).

If you fix the collector current at your chosen value and test all your devices at this Ic, then you can measure the base current (indirectly by measuring the base resistor voltage).

Alternatively, fix the base current and measure the various collector currents. The disdavantage here is that different Ic result in different power dissipation and that in turn results in heating of the junction at different rates making it VERY difficult to repeat test at constant temperature. The gain will vary significantly as junction temperature changes.

You need a set up that let's you test quite quickly and at a fixed repeatable time interval, note your readings.
If you're a whiz kid with computers you could devise a computer controlled test procedure that takes your measurements at say 1 second intervals and watch what happens as the device heats up. Then select your matching parameters.

[taj]

Thanks for the explanation Andrew. I'll have to read it a few more times to digest it properly. Much appreciated.

..Todd

[TomWaits]

Is matching devices from measuring HFE, a good method overall?

Shawn.

[AndrewT]

Hi Tom,

BJTs have a fixed collector current controlled by a fixed base emitter voltage

from this statement you can gather that I favour Vbe matching.
Many others go for hFE matching, I think to minimise distortion.

Maybe they will elaborate.

[taj]

Originally posted by AndrewT
I favour Vbe matching. Many others go for hFE matching, I think to minimise distortion.

Andrew,

Does Vbe then affect current gain such that matching Vbe somewhat matches hFE as a result? Or by matching Vbe can the hFE still be wildly different between devices?

Sorry if that's pretty elementary, but I'm only in class 101.

Secondly, which of these parameters affect DC offset most?

..Todd

[CBS240]

Hi Stefano

Jfets

I have a better cad drawn circuit of this but I can't find the file. This is how I match Jfets and my circuits work. It is quite simple, measure gate voltage to GND with DC voltmeter and select R value for whatever Ids you are going to operate them.
Just change R to +14.6V and the drain pin to -14.6V for p-channels.

You will find a much greater degree of difference in the Jfet Vgs than variables in BJT's. This may have something to do as to why people pay extra for duel Jfets. I use SOT 23 Jfets, and stick them to the tabs to keep them sorted in more general Vgs groups. Then re-measure from each group to find more exact pairs. It is really not as bad as it sounds, but you must have a good supply of transistors. Some types are more various than others.

I know this has been discussed on this forum elsewere, but this is what I did and it seems to work. I could buy 10 or more of these Jfets for the cost of a duel.