Constant-current LEDs?

A JFET is pretty much a current source if you connect gate to source, a few types go to 20mA - if you're using power LEDs, you can gang more JFETs in parallel.

The problem with JFETs is wide parameter spread - if you ground the gate and take source to ground via a resistor, you can calculate the current setting from R and IDss.

That's optimistic; if the voltage range were 4 to 20V, a 3V (white) LED will go over 10:1

So, what you really want is a d'Arsonval movement, a simple dark vane that (when powered) swings over a bright spot. It's fairly easy to get those to indicate over a 100:1 range of input. They work better in sunlight, too.

"Current regulator diodes," like the OnSemi part I linked, are selected JFETs. If they were common, I'd use them. But they aren't, they're not the 5mA I seek, and a 10mA unit would over heat.

There'd be lots more JFETs used if they came with selected Vgs(th) or Idss.

Cheers, James

This was discussed a while back when I was looking for something that would be temperature compensated. Someone mentioned the LM334 which was an interesting choice. One $0.27 part (qty 1k) and one resistor.

My LED's "dropping resistor" voltage swings from 4 to 22V, a little more than 5:1. If I size the dropping resistor for 5mA @ 22V (a very rare circumstance), then the LED will barely get 1mA as most commonly-used. That's the gripe.

I hope they're $.10, waterproof, indestructible, and fit in a 5mm hole!

Grins, James Arthur

I thought of those, but rejected them as only available in SO-8.

Maybe I shouldn't have been so hasty--an SO-8 isn't much bigger (if any) than my kludge. $.31 each by the reel at Digikey.

And if I'm going to use an SO-8, I might as well use the LM317L-- $.12 by the reel, and ubiquitous.

Thanks for prodding me into reconsidering that.

ISTM I'm still forgetting one of the old National parts though, one of the three-terminal regulators or references...

Cheers, James Arthur

TI has TO92, LM334Z/NOPB, Digikey stock >1K, Mouser >5K. Are you looking for smt?

You can do slightly simpler, by building a shunt (rather than pass) type CCS. It's wasteful, because it shunts away up to 4/5ths of the load current (as the name suggests).

Otherwise, an adjustable LDO is pretty much ideal.

And yeah, CRDs have been boutique for a long time. You can make your own with JFETs, if you don't mind a not-quite-as-ugly range on current, like

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website:

Harumpf yourself. What's your voltage range?

Can I sell you 3.2 mA? Two LND250's in parallel at Idss.

Or if you must have 5 mA, one DN2530 and a 499r source resistor.

Would that really be any simpler? If you try to do a one device solution it can only regulate voltage, not current, right? I'm thinking of something like a Zener. To regulate current in a shunt device would take two actives, no? Or is there a special part for this?

Yes. SMT is required.

Cheers, James Arthur

Shunt would be awfully warm. To get 5mA at the low end I'd have to dump

I guess 3-8 mA wouldn't be the end of the world. One advantage of my discrete kludge is that it's pretty tough. Another is that it doesn't add any line-items to the BOM, saving on stocking and P-n-P set-up.

Cheers, James Arthur

The CCS will see ~4-to-22V normally. Withstanding the occasional 60V(?) spike for a uS or three would be a nice bonus.

Those are pretty cute. I s'pose I could dredge through the old JFET books for a suitable JFET, too.

It's only three LEDs, but it does seem wasteful using 4 parts each to limit their current.

Cheers, James Arthur

A current-regulating diode is a special part. I linked one in my OP.

Cheers, James Arthur

The Supertex parts are very repeatable as regards Idss. Jfets are awful.

Can't run them in series? At least parallel them with a bit of ballast resistance, or maybe no resistance at all.

LM317L is available in SOT-89, too.

pike

it

A quad matched NPN transistor - use three as the 5mA constant current sourc es, and feed the fourth with 5mA to set up the common base voltage. Getting the constant 5mA is left as an exercise for the reader, but a PNP, a zener and a couple of resistors would do it, or a 10mA CCD and a 130R parallel r esistor.

Four discrete transistors (or two cheap duals), each with 130R in series wi th the emitter (which could be a resistor array), and a 1.23V voltage refer ence to drive all the bases would work too. Two duals, a quad resistor pack and one voltage reference is four parts.

I was thinking about some switching thing... does anyone make a switch mode current source? (OK it won't come in a sot-23 pac.)

George H.

I've used that, kinda, with a FET for the upper transistor

Can't. Three separate indicators, indicating different things. Besides, I only have 4V headroom, worst case.

Yep.

Cheers, James