repeating the experiment
M0NTV raises a good point on DVM diode test, I checked my Metex DVM and also a Harbor Freight type DVM - both had a current of a little over 1 mA.
source: https://www.youtube.com/watch?v=EkMpLKwNl2U
I tried the LM317 circuit using the DVM diode test and the LM317 current source with fixed resistors of 120 and 240 ohms to measure 9 BAT-42 diodes and got the following results:
number 1.1 mA 5.2 mA 10.5 ma
----------------------------------------
1 284 325.0 353.0
2 278 326.0 350.5
3 281 327.0 352.7
4 281 325.5 352.0
5 281 324.4 352.5
6 281 325.3 352.4
7 281 325.0 351.8
8 281 324.0 352.2
9 281 325.3 354.3
all measurements in mV using a Metex DVM
It appears that the choice of diode makes a big difference. The BAT-42 has a much lower forward voltage at 10 mA than the 1n5711. In fact the 1n5711 measures similarly to the 1n4148 switching diode which is a PN (non-Schottky diode).
I also noticed that the readings never quite calmed down, at first they decreased and then they started climbing, not quite sure what the problem is.
One final observation, the LM317 is a good candidate for this circuit, for one thing the 1.25 V nominal output voltage means that not that much power is wasted across the 120 ohm resistor which isn't a problem in this case but might be more significant if the current being sourced was larger. But also one other interesting thing with the LM317 is that the quiescent current doesn't present a problem. If a 78L05 regulator was used in this same configuration the quiescent current of the regulator (over 3 mA) would add to the calculated current making the design slightly messier :)
I do have some actual (supposedly) HP 5082-2800 diodes around here somewhere which are comparable to the 1n5711. I'll try those tomorrow.
Best Regards,
Chuck, WB9KZY
http://wb9kzy.com/ham.htm