[Aavso-photometry] Linearity test puzzle

[Vello Tabur]

Radu Corlan wrote:

>On Mon, Jan 16, 2006 at 12:57:14PM +1100, Vello Tabur wrote:
>  
>
>>I've been testing the linearity of my camera (SBIG ST8XE) using a LED + 
>>voltage regulator, as per the circuit in AIP v1 (pg 201).
>>Ten exposures (30sec each) were taken either side of a linearity test to 
>>assess LED stability during the test.  The plot below shows a gradual 
>>increase in LED brightness of 0.24% over a period of one hour. A curious 
>>systematic feature appears on the plot: the first exposure of each group 
>>shows an rapid increase in flux relative to the other exposures. It 
>>occurs in each group and the same effect occurs in tests conducted on 
>>other nights.  Any suggestions as to what could be causing this?
>>    
>>
>
>i'm not sure how that circuit looks like, but please keep in mind that 
>the light output of a led is greatly affected by temperature. a red 
>led's intensity can decrease by 0.5% for each degree of increase in 
>temperature (at constant current). green and blue leds are a little 
>better (they have a variation about 1/2 of that), but still quite 
>unstable. a regular led has pretty bad cooling (mainly through the 
>leads), and will self-heat with a time constant of seconds. 
>
>a second effect appears when the source driving the led has significant 
>output conductance (so it doesn't look like a "perfect" current source). 
>the led voltage decreases with increasing temperature, which may 
>increase the current through it and affect the light output.
>
>when you first power a led, the combination of the above will give you 
>a transient that lasts several seconds (and maybe more). 
>
>so keep your environemnt stable, and you use the led at a 
>constant, not too large, intensity (5ma or something).
>
>radu
>  
>

Thanks Radu. I believe this explains the drift in the LED rather well. 
Smith (R.M. 1998, Experimental Astronomy, 8, 59) quotes -1%/C drift, 
rather more than your figure above, but in the same ballpark. FYI, I am 
using a green LED.

Actually, the drift is not a problem because it is so smooth and 
predictable. Using a best fit linear relationship, I corrected my three 
linerarity tests to account for the increase in lamp brightness during 
the test, and folded the three tests into one plot:
    http://users.bigpond.net.au/vtabur/lintest2.jpg
As you can see, I have a very tight, repeatable relationship, 
particularly between 9 and 70 seconds. The 5 and 6 sec exposures have a 
larger scatter due to shutter effects.

Forgetting the lamp stability issue, do you have any idea why the first 
exposure of each group might show a lesser flux compared the rest of the 
group? Here's the plot again for reference:
    http://users.bigpond.net.au/vtabur/lintest.jpg
BTW, I tried taking groups of 30 exposures to bracket the linearity test 
and found that still only the first exposure in the group was affected, 
so the effect appears to be  independent of the number of exposures in 
the group. In all cases, the camera was idle (and shutter closed) for 
30-60 seconds before a test was started.

Thanks,
Vello