I recently ordered several E153 CRDs from Semitec (data sheet here) to maintain a steady current to an LED (data sheet here). Mistakenly, I didn't read the "Dynamic Characteristics" portion of the data sheet before ordering which seems to suggest the E153 doesn't line regulate very well, with current peaking at 10Vin and dropping off thereafter. The "Basic characteristics" graph however gave me the impression that any Vin above 10V would maintain a steady pinch-off current.
My questions is, is this expected behavior of a CRD or is the E153 just bad? Can I still use them as a constant current source for my LED?
For my purposes I need to use a 9V power source which droops by at most 0.5V under the heaviest load we require. Temperature will range from 25C to 30C worst case. The consistency of LED brightness is crucial for its purpose; we need the LED to maintain +-0.1mA of current. We bin our LEDs to ensure they all output the same brightness at a specific current.
Sorry for the basic question, and thank you!
Best Answer
It's a single JFET, and probably a resistor, packaged as a diode. This won't be very accurate. The "voltage reference" used is the JFET's threshold which depends on temperature, which depends on voltage and dissipation. This, plus other JFET properties, mean it will vary quite a lot. It's still decent if you want a somewhat regulated current, with one part, and it is much better than a resistor if the voltage will vary a lot. It is not adequate for your application.
If you need something a lot more accurate and stable, try this:
TL431 is cheap, readily available, low noise, accurate and stable with temperature. It will adjust the current drawn from pin "A" to keep pin "REF" at 2.5V above pin "K", via feedback through the internal error amp. This adjusts Q1's base voltage to control LED current.
I wired the BJT this way to make sure the base current (which depends on temperature and BJT) is not included in the regulated current.
For a 10mA current in the LED, set R2 = 250 ohms. Adjust accordingly.
This will cost maybe 20c and result in excellent stability. Main source of LED brightness variations should be air currents changing the LED temperature...