Transistor matching - my results

[Magnus Danielson]

> I just thought I would share this with the rest of you.
> 
> I downloaded a transistor matching schematic from someone's
> web site (info was posted to Synth-DIY earlier).  The schematic
> originally came from a Moog service manual.  I don't have a 10V 
> power supply hooked up yet (I have a +/- 15V linear supply, but
> it's not hooked up.)  So I used 4 x 9 volt batteries in pairs as a
> +/- 9V supply.

According to the transistor matching description that I have in the Minimoog
service manual you can use either +/- 10V or the +/- 15V power. Actually will
+/- 12V work great too... cause it is a current source involved.

>  I tested and found that all of the transistors 
> (or at least the first 7) matched perfectly!  They would still have to 
> be epoxied together to keep them at the same temperature when 
> they are on the board.  It looks like transistors from the same 
> production lot are very likely to match.  Here is the test data for 
> my first matching session:
> 
> Transistor #1	4.574 V, retest after #7	4.571 V
> Transistor #2	4.574 V
> Transistor #3	4.573 V
> Transistor #4	4.573 V
> Transistor #5	4.572 V
> Transistor #6	4.571 V
> Transistor #7	4.571 V

I quest weither these numbers are correct since the matching curcuit will
measure the Vbe for an Ie of 100 uA. Note that the circuit will set the Ie and
not the Ic as normally, however most transistors have a hfe of about 100 or 
more
so it is approximatly Ic controlled anyway.

The example voltages from the Moog manual is 0.600 V and 0.598 V which is much
more in line than 4.573 V. An high Vbe is 0.75 for 100 uA for a normal
transistor.. recall also that Ic = Is * exp (Vbe * const) (or there abouts) so
the Ic will increese much for sligt increments on Vbe... so this indicates that
something is out of hand.

Could you possibly have been measuring the Vce? This is likely since the 51k
resistor will nock of about 5V (or rather, about half the powerrail rating... 
).

The Moog matching curcuit is pretty good than having nothing or doing just
resistive measurements... I perticularly like their temperature handling.

> You'll notice that the voltage was dropping as I went.  I suspected
> it was just battery drain, thus I retested several transistors and 
> they all matched.  I showed Transistor #1's retested value to
> illustrate this.  These values don't mean much in themselves, as
> they depend on battery strength, the exact resistors and Op-Amp
> I had in the circuit, etc.  BTW, I was using a Fluke 8060A digital
> meter.

I'd check the schematics and curcuitry again if I where you since I think the
values are out of line and possibly have a little to strong dependence on the
powersupply. I'd run the curcuitry of that +/- 15V you do have. Best thing 
would
be to give the curcuit their own stab's so that the stab's have a known 
constant
load. Personally I'd also put some effort into improving the curcuit itself.

Your DVM is probably not a source of problem, it's a fairly good one I'd say :)

Cheers,
Magnus