[Laser] variable field of view for noise elimination

[Glenn Thomas]

Hi James et al

We seem to be in violent agreement.

The mechanism certainly can be aimed either by moving the entire 
contraption or by moving the sensor & mask itself within the focal 
plane or both. Or even put something like a CCD imager in the 
f-plane, determine the pixel you like and then ignore the others. All 
pretty much the same thing.

Uhhh... I believe that the pin-hole camera lens works because of edge 
diffraction. Perhaps I really mean refraction? - though I'm not sure 
that Snell's Law would apply if the index of refraction doesn't 
really change anywhere.

The diffraction effects of the mask might be significant if it were 
small enough and if it were located far enough away from the sensor. 
That small hole is in fact a lens, just as if it were part of a 
pin-hole camera. If the mask is close enough to the sensor, the 
pin-hole lens effect is minimized because all of the energy still 
falls somewhere on the sensor. I think you said this - just not sure. ;-)

You did say that:

>I  think that a sensor that is about the size of the signal blob at the focal
>plane  would be a bad design.  As I describe in the original post, that
>represents  nearly the minimum field of view for the instrument, so 
>it would be
>degraded by  small vibrations, not to mention that it would be very 
>difficult to
>acquire the  signal.  If you make the signal blob larger than the sensor,
>perhaps by  moving the sensor off the focal plane, you can expand 
>the field of
>view, but you  lose signal gain.  If you have the margins to do 
>that, it will
>work.   At the very least, sensors that small should be part of 
>an  array with a
>much larger area.

I don't think this is quite correct. The ideal case is when the 
sensor size and shape is exactly matched by the size and shape of the 
blob. This provides for maximum signal power recovery and minimum 
noise generation.

If the sensor happens to be smaller than the blob, you will not 
recover all of the available signal energy. If the signal is weak, 
this can make the difference between acquiring the signal and not 
acquiring the signal. Thus the ideal sensor will be at least as big 
as the blob.

If the sensor is larger than the blob, all of the signal energy will 
be utilized. However many sensors (CCD for sure and possibly others) 
produce internal noise in proportion to their area. When the sensor 
is larger than the blob, even if we use a mask to block external 
noise sources as discussed, the internal noise will be larger than it 
would've been had the sensor area matched the size of the blob. This 
degrades the SNR when compared with the matched sensor. Again, if the 
signal is very weak, this can make the difference between acquiring 
the signal and not acquiring the signal. Thus the ideal sensor will 
be at least no bigger than the blob.

The usual sensor I read about on this list is a photodiode feeding an 
analog amplifier. I've not heard much about anyone doing any 
integrate and sample work.
However, as you suggest, it seems that a CCD array might be used for 
communications if anyone is willing to make the effort. That rules me 
out, but here's an approach that ought to work.

A good sensor to use is a CCD imaging array. Put it in the focal 
plane and aim the whole contraption so that the signal blob falls 
somewhere on the CCD. Exactly where doesn't matter much.  Ideally, 
the size of the signal blob will be exactly one pixel. Processing the 
CCD image with a PC will allow the specific pixel with the blob to be 
identified that the others ignored. In this way we can synthesize 
both the mask and the mechanism to move the sensor (a little ways anyway).

Always having the blob & pixel sizes be the same sounds too good to 
be true. Besides, even if they are, the blob may well choose to fall 
at the intersection of four pixels, worst case. It may be better to 
have a CCD array with pixels that are smaller than the blob. The 
software mechanism here is very similar to the same size case above 
except that the PC needs to identify all of the pixels that contain 
the blob instead of just one. The received signal is then the sum of 
the signals from all of the selected pixels. In this way we can 
approximate a sensor that is the same size & shape as the blob 
(subject to pixelation of course) and at least approach an ideal 
sensor. Smaller pixels yield a better approximation at the cost of 
the need for increased computational bandwidth.

While we don't have to deal with the mechanical issues of the 
mechanical mask & sensor, possibly moving, we do have to deal with 
the fact that the CCD array is more fragile than a photodiode and 
more complicated to control. Also, the bandwidth of the received 
signal is limited by the available computational bandwidth, but at 
least you have the option of easily trading signal bandwidth for 
sensitivity by simply changing the integration time.

73 de Glenn WB6W


WAR IS PEACE!
FREEDOM IS SLAVERY!
IGNORANCE IS STRENGTH!
(be seeing you!)