I thought I knew electronics but apparently don't know what a current mirror is!

[crutschow]

............................ Matching transistors - in particular the Vbe - appears to be necessary, or figure out a way to compensate for the differences.

That's what the emitter resistors are for.

 

[Mosaic]

I couldn't fit 7 more thru hole resistors into my circuit dimensions. Matching Vbe was the only way.

 

[Rich D.]

And the winners are: Everybody who recommended emitter resistors!

I put 47 ohms on each emitter - and it worked just fine. With a power input ranging from 12 to 20 volts, I was getting about 20 mA through eight LEDs, and in fact the supply current only changed by 2mA from 12 to 20 volts.

I put up the schematic showing all the voltages read at 12 volts in purple, and 20 volts in red. Works very well - each LED is allowed to "find" it's own voltage between 3.2 and 3.3, and each LED is running 20mA. I even replaced some with green LEDs and everything still held up well. What could be better than that?

I suppose a normal voltage regulator and some calculated resistors would do the same, but at least there is another way to skin this cat - no more resistor calculations and no more worrying about input voltages if I use this circuit.

Thanks for all who helped. There's now one more person in the world that has some experience with current mirrors.

 

[MrAl]

Hi,

A bipolar with emitter resistor is one of the simplest current regulators you can find. the base is help at constant voltage so the drop across the resistor is constant. With a given beta, the collector current will reach a certain value and stay relatively constant. The sensitivity to beta is then possibly the only issue left.

 

[throbscottle]

Here it is in operation....don't laugh too much, its for kids. Play back is more in sync if watched on YouTube natively.

Awsome! I just love the music sync, truly great

 

[ronsimpson]

The sensitivity to beta is then possibly the only issue left.

Look at the right two transistors. Lets declare one hat a gain of 100 while the other one has a gain of 200.
Note the emitter resistor has both the collector and base current added together.
Note, crude math, rounding off, etc
The goal is to get 20mA in both collectors. One transistor needs 200uA while the other base needs 100uA. So the collectors will be 200 or 100uA short.
19.9mA and 19.8mA of collector current to get the right voltage across the emitter resistors. 1% error
Considering the use of 5% resistors, a 2:1 beta difference seems good.
High gain transistors have less (beta difference effect) while low gain transistors have more effect.

 

[kinarfi]

I decided that I would like some experience with current mirrors also, so I made one using several 2N4403s with ~ .682 Vbe using the diode check on my Fluke. The master was set a 4ma using 12v and a 3K resistor, and as I tested each of the followers using a 1K resistor, they would would start out close to 4ma, but climb toward 12ma, NO EMITTER RESISTOR, I will try that today.