I don't have a CCD camera, but I am casually attempting to build a IR illuminator for gen-1 NV scopes.
In a 2-D-cell mag-lite, so far I have tried:
1) using a visible red LED "bulb" with resistor (mA==?I don't remember, 35?),
2) using a 940 nm LED "bulb" with resistor (about 45 mA),
3) using a 840 nm LED "bulb" with resistor (about 45 mA), and
4) using a perfboard with nineteen 840 nm LED's+resistors, mA=23 each, as it turned out. I wanted more current, but gave up on tweaking it as I didn't see it making a lot of difference. The beam was too broad to be of much help, and I didn't want to carry around a telescope to focus it.
-------
None of these worked real well out past about ten or fifteen feet, especially outdoors.
In fact of all of them, the visible-red single LED worked the best, it was useable out to perhaps 25 feet, and "visible" out to maybe 50 feet. And I found that being able to focus the LED "bulb" beams (using the original mag-lite reflector) was a tremendous advantage, but the single IR LED's just didn't put out that much light. The 19-LED setup is very smooth and bright, but not focussed at all.
--I have ordered some low-cost plastic IR filters, and will try them on the regular-standard mag-lite setup. After that, there are expensive glass filters, for $80-$100....
~~~~
And by the by, the 940 nm LED's are totally invisible to the eye. The 840's die glows a dim red when powered, visible in the dark if you are within 10 or 15 feet, and it is pointed straight at your eyes. Gen-1 NV peak response is supposed to be between 740 and 800 nm, but I didn't really see much difference between the 940 and the 840 LEDs.