Here are some results of first Fibers/LED tests.
    The simple picture describing test setup is presented here - 
Setup and LED image  (t_s.ps) . I've made two 5 meter long
fibers with some tips on the ends. The fiber ends were polished
up to 3 micron quality. LED was driven by instrumental pulse
generator with rectangular shape pulses 50 ns in duration with
amplitude from 3 to 5 volts and constant offset from 1 to 2 volts
added. The LED end of coaxial cable was terminated by 50 Ohms.
Generator frequency was 2 KHz. The first thing what I should
mention is an interesting shape of LED's light image. I've put
a sheet of paper as a screen and have observed the image shown on
this picture  (Setup and LED image ) (t_s.ps) with a O-ring and two
hot spots inside it. The LED used was a small one, 3 mm in diameter.

    Typical amplitude distributions are shown on following pictures: 

    For this pictures - HV=1700V, LED pulse = 4.5V + 1.0V offset.

    Dependencies of Amplitude and Sigma from LED pulse voltage for
different conditions are shown on following plots:

     I've also made a bending/heating test. The fiber part which is outside
of the black box was bent by making a several loops on the approximately
1.5 meter length. The dependence of light amplitude vs bending radius shows
that there is no such dependence at all up to 1 cm bending radius. This
picture is in  Bending test (bend.ps). After this bent part of the fiber (1 cm
bending radius) was heated approximately to 70-80 degrees of C.
No difference was observed.