You need a low Vgs(thres) MOSFET to minimize the total voltage drop across the limiter.

A lot of the small ones (BS170, for instance) are generally about 1.5V (Specified 0.8V to 3V)

A schottky diode in series with R2 will deal with the temperature compensation, giving I = 0.4/R2. The compensation has made the current less exact, but more stable!

 

OK, next experiment. I'll try this at 1V,2V,3V, and 8V and report back w/ the results.

​

 

It would probably work best with NPN transistors, and the resistors are meant to be between the bases and the supply.

 

Schematics are the language of electronics. Would you provide a rough sketch please?

 

Schematics are the language of electronics. Would you provide a rough sketch please?

I'm just referring to the circuit in Post #15.

 

Oops, schematic redrawn, gone wrong. Easy fix, thanks.

​

 

Still needs NPN transistors, or the + and - terminals swapping over.

 

Results:
Not good
2V = 237µa
3V=487µa
6V=1173µa
8V=1173µa

Next I'll try 1ma.

 

gotcha.

 

Still not good.
2V=237µa
3V=489µa
4V=706µa
6V=1172µa
8V=1161µa
9V=14ma
Same numbers, will retry experiment.

 

Much better, may not be good enough.
2V=128µa
3V=154µa
4V=176µa
6V=222µa
8V=276µa
10V=311µa
20V=527µa
What do you think we could do for improvement?

 

What do you think we could do for improvement?

Use my circuit of post #3. Diodes/Zetex ZXRE125 reference, BC847C/857C transistors 13k resistors.

 

​

Understand, I'm using parts I actually have. I am planning on ordering 1N5817 Schottky diodes and a LM334 soon. Is there anything I can substitute for the zeners?

 

Is there anything I can substitute for the zeners?

Try a red LED. They usually have a reasonably sharp knee on the I/V curve. It might *just* get under your target for 4V operation.

 

Old or new red LED (1.5V or 2.5V Vf)?

What voltage for the zeners? I might have 400mw version.

I can always tack it on my mouser list.

 

How about 4 forward biased Schottkys in series?

 

I must be thinking of an old one - I always thought of red LEDs being 1.8V. The absolute minimum voltage at which the circuit will work is twice the reference. Two signal diodes in series might also do.

 

Tried something similar, right on the knee of the diodes curve, doesn't work. So you were thinking of 3.3V zeners?

I tend to be a bit of a bulldog on this sort of problem I will pick this up again when I have more parts.

 

No - I was thinking of bandgap references - something like a LM4041.
The LM4041 will regulate down to 60uA.
Diodes[Zetex] have some that will work down to 8uA (ZXRE125).

 

So what is a bandgap reference? Or should I google it? Learning something here.
3 or 4 Schottkys might work.