PNP Current Mirror

Thread Starter


Joined Dec 9, 2013
Hi all,

I'm supposed to get 500uA over I2 however due to low beta value of Q4 transistor the current is limited. Can you suggest any topology in order to enhance the current over pnp ?

Thank you



Joined Sep 3, 2015
You could just try reducing R8 until you get 600uA, but this is not a good way to design a mirror. There is a 100:1 ratio of currents in the 3 PNPs in the mirror, and emitter areas need to be more proportional to the current. Beta rolls off at high low current density, due to current crowding in the collector and high-level injection in the base, when the number of electrons exceeds the number of minority carriers. Beta rolls off at low current, due minority carrier recombination at low carrier velocity. 100:1 is a large enough current ratio that at least one of the PNPs is going to have crappy performance.

The point is that it is better for transistors to have a similar current density within a design, and ideally, that density should be near the BJT's highest beta. Let Q2 be a single PNP. Then Q4 should be 10 PNP of that size in parallel. The resistor on Q2 remains 6.6K, Q4's resistor changed to 660. Let Q1 = Q4 = 10 Q2, and also put 660 ohms on its emitter. That means IREF must change from 6 mA (!) to 600 uA, so change R1 to 14.3K.

With the emitters scaled, you can delete and short the emitter resistors if you want, or leave them in.

That might be enough to make this all work, but I would do more: The NPNs need to be scaled, too. 10 in parallel for Q8 and Q13, because they will still need to run 6 mA, and IREF was just reduced by a factor of 10 in the previous paragraph..

This is not perfect, as Q9 and Q10 will have 1/10 the current density of the other NPNs, but that is probably OK. See if this gives you operation you want.