Re: ATM amateur infrared imaging?

[Dominic-Luc Webb molmed <Dominic.Luc-Webb@molmed.ki.se>]

On Thu, 14 Jun 2001, Stan Kaufman wrote:

> Don't the issues Frank raises about cooling system
> requirements pertain to your devices?

Stan refers to PMTs and photodiode based detectors. Answer
is simply a matter of how intelligent the detector is. I
guess all of these detectors have difficulty in
distinguishing between thermal signals and photonic ones.
The measurements you need to make will dictate the degree
of cooling.

Here in my lab (microscopes, BTW), my CCD is cooled
to -32C for measurements of light at about 510 nm. If you
can imagine typical fluorescence blue-green image of a cell
seen through a microscope (plenty to be seen in biology
books, etc), we can use this as a common example. I have
measured the ambient thermally derived signals to generate
about 102 counts per pixel in the dark (don't concern yourself
with the definition of counts for the CCD, just the magnitude
and differences in values). The stray light not from the
optical path directly add another 4-5 counts. I subtract both
of these for measurements of fluorescence from the cells. The
counts from a fluorescence intensity of one of our classical
textbook example images will be about 200 to 300 counts over
the background. We routinely make measurements when the counts
are as low as 20 counts above the background. The recordings
are then very noisy, partly due to random background and
dark current noise. This is accomplished with a Kodak 1400 KAF
(expensive) photometry CCD at 4x4 binning and 200 ms integration
time. The stuff above gives some ideas about the dynamic range
we have to work with under these conditions. The numbers are
different of course for astro applications, but very similar
problems are encountered, problems, that is, mainly if you are
like me from a background of PMT pulse counting....

That same microscopic system above is also fitted with a
PMT pulse counting system, the PMT itself being a typical
one for astrophotometry (the CCD is too, BTW). With very
similar settings as used for the CCD, we get some very
different values for the PMT pulse counting, cooled at -20C.
The dark current is about 800 counts. Noise from stray light
is a few hundred. Sounds terrible, yes? Well, in fact, it
is no problem at all. With the same light source to excite
the fluorescence giving roughly the same photon flux as the
CCD, I get something like 25000 to 100000 counts per second.
Thus, I have a massive dynamic range to work with at what has
thus far proven to yield far greater precision. Precision is
estimated by placing dyes like fluorescein into the optical
path with very small differences in concentration. With
the PMT, I can easily discern 1-2% differences in
concentration between two very dilute samples of dye,
microscopic samples at say 1.00 or 0.98 nM. I have
never been able to resolve this with the CCD.

Cooling would make a difference for both the CCD and the
PMT, but in this example, I hope I have demonstrated that
for some applications the CCD is the clear loser, even
with more cooling than a PMT. Recall, the CCD is cooled
to -32C and the PMT only to -20C. But I am not a CCD
opponent. I do use them rather much. This is not a
CCD bashing email, but my users-point-of-view attempt
to explain clearly observed differences in these systems.

I should not drag this discussion on. I can say that PINs
are not normally recommended to be cooled to very low
temperatures. Burle gives temperature specifications
for PMTs down to -80C, the max limit. Texas Instruments
has similar info for their line of CCDs, which they in
the past have warned to not cool much below about -32C. I
think it was Siemens that offered a spec like this for
some of their PINs. All of these have web pages with
suitable technical links.

Bottom line, all of these give improved performance
with cooling. Some of the newest devices are more
temperature stabile than others. I think the subtle
differences ATMs are trying to detect in the IR
may require rather extreme cooling to resolve them.
S/N is improved in PMTs with cooling, and presumably
this is common to all these devices.


Cheers,


Dominic

North 59 37' 30"
East  17 48' 10"
Sweden