In a message dated 30/05/01, twleiper@........ writes:

> The method I speculated earlier today was off the top of my head. But it 
> does make sense to see if the numbers agree with my instinct. A modest 
> attempt follows:
> 
> It seems reasonable that the "angular amplifier" I described using two 
> opposed mirrors that move slightly off parallel should work and, if you 
> were spinning the slit
> at around 30 revs per second any noise and instability would be effectively 
> averaged out for any but the shortest period instruments. And, depending 
> upon how many times you bounce the path between the mirrors it seems 
> reasonable
> that you should get a amplification of angular displacement of the boom by 
> at least an order of magnitude, meaning that the .01 arc/sec requirement 
> becomes .1 arc/sec. This could also be expressed as about 1/200,000 of a 
> revolution and
> would occur in about 160 nanoseconds in the 30 RPS example. Detecting phase 
> shifts in 160 ns chunks is a piece of cake, in fact you could probably go 
> 

Dear Tom,
       I see the argument, but 1 arc second is 1/60*60*360 of a revolution - 
1/1,296,000, so at 30 RPS, 0.1 arc sec represents ~2.6 nano sec, doesn't it?  
You also need the size of the detector to be roughly comparable, so we need a 
very tiny sharp edged photo detector which is extremely fast - you have to 
physically scan the beam across at least part of the detector. While you can 
get 0.5 sq mm diodes with a 1 n sec response, isn't scanning an edge across 
the device likely to just see a diffraction pattern and hence be limited by 
wavelength considerations, even if you could do the timing? 
       Regards,
       Chris Chapman
In a message dated 30/05/01, twleiper@........ writes:



The me
thod I speculated earlier today was off the top of my head. But it 

does make sense to see if the numbers agree with my instinct. A modest 

attempt follows:



It seems reasonable that the "angular amplifier" I described using two 

opposed mirrors that move slightly off parallel should work and, if you 

were spinning the slit

at around 30 revs per second any noise and instability would be effectively 

averaged out for any but the shortest period instruments. And, depending 

upon how many times you bounce the path between the mirrors it seems 

reasonable

that you should get a amplification of angular displacement of the boom by 

at least an order of magnitude, meaning that the .01 arc/sec requirement 

becomes .1 arc/sec. This could also be expressed as about 1/200,000 of a 

revolution and

would occur in about 160 nanoseconds in the 30 RPS example. Detecting phase 

shifts in 160 ns chunks is a piece of cake, in fact you could probably go 

down another order of magnitude.



Dear Tom,

       I see the argument, but 1 arc second is 1/60*60*360 of a revolution - 

1/1,296,000, so at 30 RPS, 0.1 arc sec represents ~2.6 nano sec, doesn't it?  

You also need the size of the detector to be roughly comparable, so we need a 

very tiny sharp edged photo detector which is extremely fast - you have to 

physically scan the beam across at least part of the detector. While you can 

get 0.5 sq mm diodes with a 1 n sec response, isn't scanning an edge across 

the device likely to just see a diffraction pattern and hence be limited by 

wavelength considerations, even if you could do the timing? 

       Regards,

       Chris Chapman