A simple CCS

Yeah- let's see it.

Beta generally increases with temperature, unless you are in or near saturation, then it decreases.

Show me ;-)

...Jim Thompson

Are you talking about testing a spice circuit, with spice models of unknown validity, or are you talking about real transistors in an environmental chamber? How many transistors did you try; can the transistor be changed without tweaking ANY part values?

Yes, me too.

Probably a Spice simulation with no self-heating ;-)

...Jim Thompson

It has the self-heating included, and that is why it works so well.

We choose Re that Vdrive/Vbase= 2.5 Then we choose the the base resistor according to the hFE in this case minimum is 60. output impedance around 130R, short proof Pt=1.8W Isc=367mA I normal 245mA max at 60deg. With less components and cost than a switched resistor limiter. accepts 2.85 and 3.45V Lumileds.

I think you just got lucky. I ran the equations... it CAN be balanced, but what a mess... certainly NOT a production method.

...Jim Thompson

I_normal was a typo 345mA

Well I wrote it was meant for those tinkerers that want to experiment with some LEDs. Everyday we see Watsun and others asking and those commercially available 350mA driver cost *a lot* more, but I doubt they are of better performance. Of course you have to choose the right transistor with not too high max current, but in a nice casing, which happen to be those cheap ones. I'm also checking some other more sophisticated bias methods, seem to work quite well even without emitter degeneration. And in China they would immediatly use it for mass production, I'm sure. :-) THX for evaluating Jim.

Just put a pot from PIC to base, ground the emitter, and tweak for desired brightness. Even simpler.

John

John! You didn't put much thought into that response did you ?:-)

...Jim Thompson

As much as it deserved. If you're going to pay for a pot, you may as well get your money's worth.

And pots are always shipped set to midpoint.

John

What about the temperature effects?

...Jim Thompson

ones.

We're lighting an LED, for pete's sake. A dB or so won't be missed. Besides, beta increasing with temp will partly cancel led brightness declining.

But - brilliant idea - why not put a resistor in series with the LED to set its current?

John

Ban wrote: [ snip ]

That's a stock standard and very elementary IC Q-point stabilization circuit, stabilization against the three biggies HFE, ICBO, and VBE. It is not the "self-bias" current mode feedback warned against in AOE. It is also from the days when 9V was considered low voltage- don't really have the luxury of all that overhead these days.

Old days? 9V? Hell man, that's 30 Vbe's for an "old days" transistor! What are 'ya, wastin' voltage here!?

;-)

Tim

-- Deep Fryer: a very philosophical monk. Website:

Since it is a current source it will only ruin the compliance. And a 350mA LED is not "just a LED", but costing more than 7$ and very sensitive to overheating. You didn't get it. Maybe because it looks simple, but there is more involved.

I disagree with you. It works with a single resistor only for *very* low voltages, otherwise you have to use the thevenin equvalent voltage source.