File this under, "You learn something everyday".

I didn't know these existed but ran across it today:
http://www.luxdrive.com/products/dynaohm/

They're something like $1 apiece, so there are certainly cheaper solutions but on the other hand you can't beat the simplicity.

 

Hello,

Mouser also has a lot of "constant current diodes" :
https://eu.mouser.com/Semiconductor...rrent-Regulator-Diodes/_/N-ax1ml?Ns=Pricing|0
Prices from about 1 Euro upto ca. 10 euro.

Bertus

 

Also called current regulator diode, current zener, or field effect diode. I first became aware of them in the 70's.

The schematic symbol looks like a diode with the "arrow" replaced with a circle.

 

How efficient is it?

 

How efficient is it?

How could you answer that? I mean, it takes the role of a resistor which has zero efficiency.

 

How could you answer that? I mean, it takes the role of a resistor which has zero efficiency.

I'd be interested in knowing the dropout voltage. I'd wondered if they could be used as short circuit protection, current-limiting outputs that normally need much less than the rated current.

For example, I have a circuit with an 80mA limit for a pair of wires that run a couple hundred feet with marginal protection to some SSRs. When things are working properly, the SSR inputs draw much less than 80mA, but if the wires get damaged, I want the output current limited to safe levels to protect the rest of the circuit. If this could be done with one piece, that would be cool!

 

Ordinary current limiting/regulator diodes are now a scarce commodity because there are only a couple of manufacturers still making them. They are expensive - dollars apiece. They are actually JFETs with the gate shorted to the source and designed/selected for specific IDSS. They do produce quite a good current source with substantial voltage compliance and nicely high output impedance, which makes them good for phantom powering things.

The thing that Wayne found is probably a BJT with some added bits.

Any of these devices has exactly the same efficiency as a resistor that produces the same current from the same voltage.

 

I haven't analyzed it, so I may be missing something, but I suspect the circuit in the thing Wayne found is this (sorry, no picture, but very simple circuit):

NPN transistor
2 PN silicon diodes in series, anode to base, cathode to negative terminal
resistor between emitter & negative terminal
resistor between base & collector, collector is positive terminal
- or PNP if you prefer, just the other way up

Current is mostly set by the emitter resistor which has approximately one diode forward voltage across it. The current through the base-collector resistor is an error term, but for general purpose LED driving the error should be tolerable. The collector current will vary with temperature, but again this should be tolerable for the intended purpose.

The required minimum input voltage for the commercial device is spec'd at 2.6 V, which is consistent with a circuit like this. It implies about 1.2 V across the base-collector resistor which further implies the need for reasonably high beta so that resistor can be high enough in value that the current through it at the maximum allowable voltage isn't too gross. The commercial device also spec's 0.6 V as the minimum for conduction, which is consistent with the required Vbe for a BJT (before the diodes begin to conduct).

In typical circuits like this, the resistor to the base is tied to the positive supply rail, the load is placed in the collector circuit and the terminal I described as "negative" would go to to the negative rail. But I think the circuit I described will work sufficiently well for the intended purpose.

You could probably replace the two diodes with another BJT configured as a Vbe multiplier. That would allow using a dual transistor (common in SMT), but require two more resistors. I can't see any big advantage (could perhaps use a 6-pin pkg & 2 0201 resistors & save a little space, but power dissipation means the thing needs to be big enough not to melt itself anyway).

 

Alternative is much like the current limiting for a voltage regulator where one transistor B is turned on by voltage across emitter resistor of transistor A, robbing A, the main-path transistor, of its base drive. This could be done with a dual transistor and 2 resistors

 

File this under, "You learn something everyday".

I didn't know these existed but ran across it today:
http://www.luxdrive.com/products/dynaohm/

They're something like $1 apiece, so there are certainly cheaper solutions but on the other hand you can't beat the simplicity.

Ha! I was just going to give you that link for that ceiling LED you wanted to make dimable!