In a message dated 18/08/02, Cellectronic@.............. writes:

> Hi All, I have been reading your mails regarding "DRIP". 
> With reference to the cut and polished L.E.D. I was wondering if you are 
> aware that flat L.E.D.s are available, I have found several of these in 
> cellular telephone keypads for illumination purposes, my thought is if 
> 

Hi Steve, 

       Thanks for the input! There is a discussion regarding LEDs and 
filament lamps. You won't get a light source quieter than a well regulated 
filament and you can get high outputs. LEDs tend to be noisier, sometimes a 
lot noisier. The 'quantum well' superbright orange and red ones are fairly 
quiet and so are the IR LEDs. Whereas a filament will not pass on high 
frequency noise from the power supply, LEDs will. The main problem with LEDs 
for seismometer applications is drift due to temperature changes / heating. 
(Lasers are incredibly noisy and should be avoided.) 
     The photo output of LEDs varies exponentially with temperature and you 
may get about a factor of five reduction in light output at a constant 
current if you increase the temperature from 0 C to 100 C. Since you can 
expect ambient changes of >10 C deg and the LEDs themselves can heat up 
considerably, the variations are very significant. While you can balance out 
the zero fairly well using a common source, having the scale uncertain within 
50% is clearly not helpful. With a pair of Si or GaAs photo sensors operating 
in current mode, you can add the outputs to determine the illumination level 
that you want to regulate and subtract the outputs to give the sensor 
dispacement. Superbright orange is well matched to GaAs and IR is matched to 
silicon.
    It is a great advantage to use LEDs which have a metal can backing, since 
this enables you to dissipate the heat / control the temperature. You can get 
metal can LEDs with flat windows, metal backed LEDs with epoxy encapsulation 
and LID ceramic chip leds. 
       There is a relationship between the resolution that you can get, the 
overall range, the response time and the photocurrent. With maybe +/- 1 mm 
range and 10 Hz, it is not too difficult to get a noise floor ~20 nm. This is 
OK for amateur seismometers which do not drift significantly or which have 
force feedback. Reducing the range or the bandwidth enables you to improve 
the resolution.

       Regards,

       Chris Chapman  
In a message dated 18/08/02, Cellectronic@.............. writes:



Hi All
, I have been reading your mails regarding "DRIP". 

With reference to the cut and polished L.E.D. I was wondering if you are aware that flat L.E.D.s are available, I have foun
d several of these in cellular telephone keypads for illumination purposes, my thought is if these are OK for your purpose it m
ight save you some work. 



Hi Steve, 



       Thanks for the input! There is a discussion regarding LEDs and filament lamps. You won
't get a light source quieter than a well regulated filament and you can get high outputs. LEDs tend to be noisier, sometimes a
 lot noisier. The 'quantum well' superbright orange and red ones are fairly quiet and so are the IR LEDs. Whereas a filament wi
ll not pass on high frequency noise from the power supply, LEDs will. The main problem with LEDs for seismometer applications i
s drift due to temperature changes / heating. (Lasers are incredibly noisy and should be avoided.) 

     The photo output of LEDs varies exponentially with temperature and you may get about a factor of f
ive reduction in light output at a constant current if you increase the temperature from 0 C to 100 C. Since you can expect amb
ient changes of >10 C deg and the LEDs themselves can heat up considerably, the variations are very significant. While you c
an balance out the zero fairly well using a common source, having the scale uncertain within 50% is clearly not helpful. With a
 pair of Si or GaAs photo sensors operating in current mode, you can add the outputs to determine the illumination level that y
ou want to regulate and subtract the outputs to give the sensor dispacement. Superbright orange is well matched to GaAs and IR 
is matched to silicon.

    It is a great advantage to use LEDs which have a metal can backing, since this enables you to dissipate 
the heat / control the temperature. You can get metal can LEDs with flat windows, metal backed LEDs with epoxy encapsulation an
d LID ceramic chip leds. 

       There is a relationship between the resolution that you can get, the overall range, th
e response time and the photocurrent. With maybe +/- 1 mm range and 10 Hz, it is not too difficult to get a noise floor ~20 nm.
 This is OK for amateur seismometers which do not drift significantly or which have force feedback. Reducing the range or the b
andwidth enables you to improve the resolution.



       Regards,



       Chris Chapman