I'm blanking out on a name for something about current source circuit B in this image:



Somewhere in some forum thread in the last year there was a discussion about errors in current source circuits or current mirrors. IIRC there is a mod to circuit B that adds a resistor in series with one of the bases.

1. Is this in fact a known fix to reduce the effects of xxx (base current, temperature drift, whatever)?

2. Does the error effect or this correction have a name, like "adding the resistor reduces the Wolfgang effect".

Thanks.

ak

 

ohh i ran into this... it's W for sure... hang on I'll find it. You might be referring to Wilson mirror... but I know what you're talking about with the resistor. It's in some of the opamp designs. I wrote something about this last year... I believe let me look.

 

Widlar mirror, or Widlar current source. Works unless the current is extremely small.

 

Don't know of a circuit with a resistor (unless it is degeneration resistors in a current mirror), but Bob Pease had an interesting version of circuit B with a Schottky diode in series with Q2 base.
It cancelled out the temperature coefficient of the transistor Vbe, and reduced the voltage drop in the current sense resistor to 0.4V.

 

Did you mean this ( Figure 11.6 )?
https://wiki.analog.com/university/courses/electronics/text/chapter-11#output_resistance

 

Hello there Mr. analog kid.Sir
That topic was one of the first subjects that I researched here excellent 20 years ago I mean it's still excellent but it happened back then figuratively speaking it feels that long ago
https://www.allaboutcircuits.com/textbook/semiconductors/chpt-4/current-mirrors/

 

You posted constant-current circuits, but the answers referred to a current-mirror circuit.
Which is it that you want?

 

Constant-current. I might be confounding two dimly-remembered threads The base resistor in post #6 smells like what I'm after, but not quite it. This particular detail had a name associated with it, possibly beginning with an E. Pretty sure the guy had a Wikipedia page.

It might be that the TS wanted a constant current in his circuit, tried a standard current mirror, and was asking why it didn't work with real-world discrete parts the way they do in IC's and simulations, and the thread suggested moving to a true constant-current circuit instead of a mirror. The discussion went from the 1-transistor circuit to the 2-transistor circuit to that one plus a resistor in series with one of the bases...

Too many threads, too little brain ...

ak

 

If you change Q1 in circuit (b) to an N-MOSFET, than you eliminate any error due to the base current of Q1.

Substituting a programmable voltage reference, such as the TLV431, for Q2 will minimize the current change with temperature and voltage, but at the expense of increasing the voltage drop across R1 to 1.24V.