Dear Mark,

> It certainly looks like there is an offset in the system that has not been
> nulled. From the 'graphs' it looks like the offset is independent of gain (?)
Yes. It looks like it's an electronics (pre-amplifier) offset that has to
be corrected. People form Carl Zeiss UK that I spoke to in principle
agree with that.

>  In any case, you should always subtract backgrounds before ratioing which
> should remove the problem of an uncontrolled offset...
Yes again. I really don't mind whether I use a hardware or software
solution. However, this is valid only in theory. If you look at
my "plots" you'll see that they extrapolate to a negative value (in
reflection mode it's around -20). But as the system only displays positive
integers, in practice that offset can never be accounted for.

> > Intens|
> >       |
> >       |         /        /
> >       |        /       /
> >       |       /      /
> >       |      /     /
> >       |     /    /
> >       |    /   /
> >       |   /  /
> >     0 |--/-/--------------------------
> >       | //            Gain
> >       |/
> >       |
> >

When working at a low gain, the displayed background will be 0 when
actually it should have a "negative" value and it is this unknown value
which I should subtract for ratioing.
Obviously the plots indicate that if I collect images at 3 or 4 different
gains I can extrapolate the background more or less accurately. But that's
not a practical solution. A hardware correction (some pre-set thing?)
will be better.
The beauty of the system is that if my plots would extrapolate to 0,
I could collect images at different gains (PMT voltage constant) and
then normalise by the ratio of gains, since they're extremely linear.
In practice my dynamic range would go far beyond 0 - 255.

Regards,

Carlos