Yet another transistor matcher

[Elvee]

Some time ago, I presented a complementary matcher based on a novel principle:

Thread 'An original, high-performance complementary transistors matcher'

2024-12-08 5:26 pm

Transistor matchers are not uncommon: in fact they are all over the place, and generally work satisfactorily. A good example has been presented recently by Mark:

The vast majority however matches identical transistors. Complementary matchers are much rarer, and sometimes have issues:

It is possible to build something much better, whilst keeping things simple.
This...

• Elvee

• 
• 

The same principle applies equally well to identical transistors, and it brings the same advantages: direct comparison between the junctions, with the TUT's themselves performing the comparison and a substantial amplification.

To demonstrate the operation, I have chosen a BC547B and a BC847B, which should be identical but aren't in reality (sim)

This time, no cheat is possible regarding the 0.1% tolerance of R1 and R2.
The calculation is the same as for the previous one:

Suppose that for a given collector current, the delta-Vbe is 1mV.
From the junction equation we can infer that the delta-Ic will be Ic*Exp(1mV/26mV)=1.039-1=3.9% for both transistors.
The difference is split between the two transistors, resulting in a 1.95% change. As the collector current is ~1mA, the variation will be ~20µA.
The voltage appearing across the two 3K3 resistors will thus be 6.6*0.02=132mV.
Thus, a theoretical magnification ratio of ~130; not bad for such a simple jig.

Here, the current is 1mA, but it can be changed to any value by scaling all the resistors values in the same proportion; self-heating remarks apply too.

Here are some pics: