>>The tube gain is also a nonlinear function of applied voltage, so you may have
>>similar ultimate sensitivity, even if at low gain one PMT needs a higher bias
>>voltage. Therefore, it does not follow that you can reduce the gain by a
>>factor
>>of two with an RSI of 2. The real question is whether your signal to noise
>>ratio improved.
>>
>Ok. So let me turn this into a more biological problem. You're imaging a
>live >cell (which can be damaged by laser irradiation) with a constant
>level of >fluorescence. You need an optical section so you're not going to
>open the >pinhole. So ideally what you want is a detector which allows you
>to pick up a >similar signal (with a better signal-to-noise ratio) but
>with a lower laser >power. In this case, you have to depend on your gain
>(or voltage applied to the >PMT) to adjust the intensity of the signal.
>Shouldn't then the sensitivity of >the PMT be related with the gain
>settings ?

The Gain (either eletronic or increasing the PMT supply voltage) just
allows you to make the picture brighter, not to detect more photons (and
thereby improve the statistics).

>>Concerning accumulate to peak, the number of frames required depends a lot on
>>the kind of noise you have. Are the images obtained after 2 passes with
>>the old
>>PMT are as good as the images obtained after 6 passes with the new PMT?
>
>Well, the test performed is with no sample, just using the brightfield
>lamp of >the scope as the constant light source. So it's difficult to say
>what is good >and bad. Indeed, the histogram standard deviation is higher
>in the image >collected after 2 passes.

In the Accumulate mode you add counts if you sum more frames.  The sqrt of
more counts (the noise?) is more counts.  However, the ratio of the more
counts to the sqrt of more counts (i.e. the S/N) also becomes more.
>
>>The accumulate-to-peak function would be much more useful if >it had
>>smarter >end conditions. For example, one could >require that a minimum
>>percentage of >pixels reach a certain threshold.
>
>In this case I must say that we only followed the Bio-Rad protocol. We did
>>several other tests like imaging different
>samples and 200 nm fluorescent beads. But here was just to test and
>compare >different settings on both new and old PMTs. And in these tests,
>the image >collected by the "enhanced" PMTs was always extremely similar
>to the one we >had.

>Rui Malho'
>__________________________________________
>Dept Biologia Vegetal, FCL, Bloco C2
>Univ de Lisboa, Campo Grande, 1780 Lisboa,
>Portugal. Tel. 01 7500069  Fax. 01 7500048
>Email: [log in to unmask]

It is a bit late to do the "pre" test in photon-counting mode but that is
the easiest may to see how many phtonts you are actually counting.
However, if the old images are stored (and featureless) you could still
ratio the standard deviation to the average brightness (and hope that the
black level was set so that zero counts in the memory was really zero
signal from the "specimen" (i.e. the microscope light source)) and this N/S
ratio should get smaller if more events are really counted using the new
tube. A 2x increase in QE will come out as a 40% reduction in N/S.  Try it
with a red filter (either in the substage condenser or in one of the filter
blocks) as this is where the major improvement should be.

Cheers,

Jim Pawley

                   ****************************************
Prof. James B. Pawley,                                       Ph.  608-263-3147
Room 1235, Engineering Research Building,                    FAX  608-265-5315
1500 Engineering Dr., Madison, WI, 53706
[log in to unmask]