[sdiy] Transistor matching circuit

[Ian Fritz]

At 02:59 PM 3/15/2007, Tim Parkhurst wrote:
>Does this look to be of any value for transistor matching? Can be used
>to match two NPNs, but requires a triangle wave source and a
>two-channel o-scope...
>
>http://www.elecdesign.com/Articles/Index.cfm?AD=1&ArticleID=6251

Tim --

It looks like it might be OK for rough matching.  But for an application 
such as matching devices to make a discrete OTA it has some problems.  For 
an accurate matching measurement,  the applied Vbe's need to be the same 
within better than 10 uV, which is about an order of magnitude smaller than 
the offset errors of the opamps.  Then, if the Vbe's are identical, the 
collector current difference needs to be measured to within less than a 
percent, which may be difficult to read from the scope traces.  Also, the 
offsets of the other opamps will contribute to measurement errors, as will 
any resistor mismatches.

For accurate matching -- but at only one operating point -- I have used 
passive bridge techniques.  This requires accurate matching of a pair of 
current sourcing resistors, and additionally it is necessary to swap the 
transistors and combine the measurements to remove any residual 
mismatch.  The transistors also need to be well coupled thermally and 
shielded from air currents.  But with care it can be done at home.

   Ian