[Laser] Beacon to test a photodetector sensitivity limit

[Art]

At 03:24 PM 5/20/2007, you wrote:
>
>Thank you all  for the remarks about my beacon experiment.
>I add some  notes to precise the experiment conditions.
>The receiver  is the photo electrometer from K3PGP with a S2386-18K Hamamatsu
>photodiode (NEP  6,8.10^-16 W / HZ^1/2) and a 2SK170 JFET.
>The test is  done in a wine cellar and there is absolutely no visible light
>during the  experiment.
>(I do not  test wines during experiments!)

I used to sample alcohol based beverages during optical test sessions 
sometime. Somehow the data always looks better after the 4th drink::> 
But, it looks different the next morning.


>The   beacon is powered by a separated battery to avoid spurious coupling
>with the RX  and a manual shutter shows the signal loss when I 
>interpose it in
>the  beam.
>I use no lens  in the RX to be able to compare the signal with different RX
>types versus their  sensitive area without optical gain.

I understand this! But, you need to beware of the large scattering 
area that the sides of the test range present. If 2 photons per 
square inch bounce off the wall, ceiling and floor area and enter the 
receiver, the sensitivity will seem much better than it really is.

So, I use pinholes on the emitter and on the detector to minimize the 
error. Two pinholes (on both rx and tx) separated by an inch really 
narrow down the stray response of the system. Despite the pinholes, 
all the surrounding surfaces need to be flat black or flocking paper 
lined....so putting the test range in a small box makes this easier 
and less expensive.

I used 1/16 inch drill and punched a hole in the flocking 
paper-buying high precision micron diameter pinholes is expensive and 
not needed for 'relative' results.

In my test range, the absorbers were glass from an arc welding hood. 
It definitely filters out IR and attenuates all the visible 
wavelengths quite a bit. I bought a single piece of glass that was 
about 5 inches by 3 inches and cut it into 1 inch squares with my 
Dremel Tool and diamond cutting wheel. You can buy ceramic cutting 
tools at your local hardware store, they cut glass just fine (but not fast).

>The LED used  is rather flat and with red coloured bulb.
>It has a very  low directivity that avoids the RX beacon alignment changing
>effects.
> >From your  informations I  notice that:
>- Red LEDs  have a spurious spectrum in infrared.
>- They have a  very poor light / current linearity.
>- Pin holes  and absorbing layers are very best ways to calibrate a LED
>biased at nominal  current.
>- A one axis  shielded frame is better to avoid indirect reflected light.
>It seems not  so easy to built a low cost calibrated light source...
>I will try to  use directly a very poor quantum yield 1N914 diode with a
>dielectric IR  filter.
>It could be  more stable than a very low current biased LED.

I have seen all kinds of light emitted from the surface of digital 
IC's under test (before packaging). If you connect a device to power 
and run some signals through it, it lights up all over the place 
although I don't know how much of the light is IR and near IR and/or 
UV. But, under a microscope, it's a pretty impressive light show.

You might try using a digital IC with applied power and some input 
signals to produce low intensity light. If you pasted a bandpass 
filter between the emitter and the PD, you would be relatively sure 
that UV or long wavelength IR wasn't being measured.

Again, this would be a relative source and probably not reproducible.

Regards,

Art