> The device appears to be a "current regulating diode" analogous to a > zener.... 15mA! They're also known as "constant current diodes", or "field effect diodes". The standard part numbers for them are around 1N5300, depending on the current rating. Very nifty devices: I guess the only reason you don't see them used all over the place is that they seem to cost at least $4 each in small quantities... (I'd be REALLY happy if someone proved me wrong on that, I would like to use some in an upcoming project.) You can find some info on CRDs at: http://www.intlsemiconductor.com/catalogs.htm > You can make a current limiter out of an LM317 with a resistor across > the "Output" and "Reference" terminals Yes, but with a limitation compared to a CRD: the LM317, and all other similar regulators I've found that support this use as a current regulator, require about 5-10 mA of current flow to maintain regulation. When used in the normal voltage regulator mode, it's easy to pick resistor values that will pass that much current to ground if you aren't sure that your load will. However, in current regulator mode there's no connection to ground, so all that current has to go through your load. CRDs, on the other hand, are available with ratings down to under 1 mA. Another advantage of CRDs in certain applications is that they can safely drop up to 100 volts, assuming that this doesn't exceed the power dissipation limit. Few 3-terminal regulators can handle more than about a 30 volt drop. > Related question; Wonder what happens when one parallels these > devices? I'd think the "conductances" would add, 2 paralleled 20mA > units should let about 40mA flow, but I haven't tested it yet. Should > get to that some day According to the reference I mentioned above, you can connect CRDs in parallel for increased current without any special precautions. The individual ratings don't even have to match, they simply add together. Of course, given how much CRDs cost, I can't imagine many people taking advantage of this feature... Jason Harper