a glass-ball collimator for SQM

[Jan Hollan <jhollan@amper....muni.cz>]

Measuring small parts of sky and landscape with SQM at last...

---------- Forwarded message ---------- (adapted and supplemented)
Date: Fri, 10 Nov 2006 22:13:25 +0100 (CET)

after getting the idea how to reduce the cone of sensitivity of SQM, I made
it true yesterday. I attached a glass ball (22.2 mm diameter, which I got
from a local producer) to the SQM. The sensor itself is somewhere between
the focus and the ball itself. This means illuminating the sensor
(TSL237S) from almost full cone in which it is sensitive. The new input
cone (from which the rays can get through the lens to the sensor)  is,
however, below one radian, with most light coming from some 10 degrees
cone only (apart from some stray light from very strong sources).

Some bad images showing the initial easy adaptation are within
  http://amper.ped.muni.cz/light/luminance/sqm/glass_ball

Loss of light is some 0.21 mag +- 0.01 mag, close to the theoretical
value. How did I found? Measuring a cloudlit coarse facade of my office.
Using SQM with ball, at first I checked that its luminances are not much
dependent on the angle at which I point the instrument to the facade. Then
I measured, close to the facade, without the ball. In both cases I hold
the SQM well above my head, avoiding casting shadow as much as possible.

Subtracting this new constant from the reading done with adapter, I get
the original faintness of one square second, as before. With the advantage
that I know what cone (space angle) I have measured really.  I can measure
sky luminance avoiding Milky Way, zenith luminance staying anywhere among
trees or houses, even near to streetlamps. When near streetlamps, an
additional baffle is needed, broad and long, to limit stray light to
negligible (should be then calibrated once more, as some more
centimagnitudes may be lost).

I can also measure road luminances, billboard luminances etc. Having a
uniform screen (your monitor in a large dark room, a white
transilluminated billboard outdoors), you can check what the cone of most
sensitivity of your adapted SQM is (the cone of sensitivity of original
SQM can be checked similarly).

For adapting the current SQM, a glass ball is the best choice. Of
course, a simple hole in a paper box (black inside, the hole being at the
optical axis of the TSL sensor) does a similar job, just the amount of
light is so much reduced that measuring in dark places might not be
possible.  And you have to remember some much larger constant to get an
estimate in your ``old'' scale.

To get repeatable results, the ball is to be centred very accurately over
the sensor. I glued a rubber ring onto the front plate of the SQM to ease
this, but also the black baffle holding the ball should fit the SQM
properly, to bring the ball always onto the same axis. Unfortunately, the
sensor is shifted a bit to one side of the box, making the centring
difficult. I attached a slice of a wine cork to one side of the box, now
the baffle (a cut film can) censers the ball well, and the can holds on
the box a bit (this adaptation is not on the images, I did it just today).

A small warning: unlike the plain SQM, which has very uniform sensitivity
close to the optical axis (so you can easily get a brightness of a distant
lamp, by taking another measurement in which your finger casts a shadow
onto the sensor), the SQM with a glass ball has an unknown course of
sensitivity near its optical axis. A point source can be magnified a lot,
or give just very faint signal. The glass ball version is just for scenes
with no prominent lights. Even Vega might affect the reading, when you
point just at it by chance. No problem, take four readings, not more than
one will be affected.


Another ways of calibration:

An even quicker method would be to keep the SQM very close to the
translucent white screen (of the monitor you are sitting by...). Once with
adapter on, once with adapter off. However, the luminance of the screen is
notably lower at larger angles, even for the classical monitors (the more
for LCD ones). SQM without the adapter is sensitive at these large angles,
and gets lower signal than it would get from a hemisphere with the same
luminance as you see looking at the screen. The difference of readings was
therefore lower, mere 0.14 mag. This would underestimate the loss of
sensitivity when using the glass ball. (I got this idea for easy
calibration from SQM fathers, Doug and Anthony, but unfortunately I found
no transilluminated object with luminances being constant at least up to
some 60 degrees -- they have made such one in their laboratory.

To get a most accurate calibration (even a direct photopic one, using a
luxmeter) I started to make an integrating sphere. After much thought, I
took and old plastic globe, meaning to cut it in halves.  After I washed
it, it fell from a table to the concrete floor, cracking... but nicely, at
the original glued line, at the equator! So it remains just to paint it
very white inside and make a side opening for a detector (SQM, luxmeter,
camera lens). The small hole at the polar axis will be enough to put a LED
into it, as a light source (it is to be baffled to cast no direct light
onto the detector). When I will finish the sphere, I'll write again.


Another possible collimator:

For new, analogous instruments, an alternative exists to the glass ball. I
found it searching for LED lighting. The name is Luxeon Collimator.
For technical info see

   www.lumileds.com/pdfs/DS26.PDF

 and the other search result, leading to a html containing a price list,
reads:

   Collimator lens specifically designed to fit Luxeon LEDs. Up to 90%
   efficiency; 10° viewing angle; Fits all Luxeon LEDs (Except Star/O);
   Made from optical ...
   www.luxeonstar.com/item.php?id=396& link_str=88&partno=LXHL-NX05

I'd recommend to put the collimator into a good black baffle, cutting the
edges of collimated beam off, so that there no light would be registered,
if it comes from an angle larger than some 10 degrees from the axis. Some
0.3 mag might be additionally lost this way, so perhaps 0.4 mag for the
whole assembly, compared to a plain TSL237S sensor.

Any filter, Hoya or other, can be then put onto the collimator, with an
advantage of working at exactly defined thickness, due to the light going
perpendicularly through it (just this way its spectral properties are the
tabulated or plotted ones).


with best regards,
 jenik hollan