[time-nuts] GPS->audio interface

Magnus Danielson magnus at rubidium.dyndns.org
Mon May 11 08:20:45 UTC 2009


Lux, James P skrev:
> 
> 
> On 5/10/09 4:26 PM, "Magnus Danielson" <magnus at rubidium.dyndns.org> wrote:
> 
>> Lux, James P skrev:
>>>
>>> On 5/10/09 12:15 PM, "Brooke Clarke" <brooke at pacific.net> wrote:
>>>
>>>> Hi James:
>>>>
>>>> You might want to get a KIWI-OSD, Video overlay of GPS precision time
>>>> stamp.  It adds HH:MM:SS EEEE OOOO FFFFFF at the bottom of the image and
>>>> so can be seen in every field.  They have also developed a way to
>>>> calibrate the camera shutter in relation to the frame time by using a
>>>> number of LEDs.  The main use if for star occulation timing.
>>>> http://www.geocities.com/kiwi_36_nz/kiwi_osd/kiwi_osd.htm
>>>>
>>>> Have Fun,
>>>>
>>>> Brooke Clarke
>>> Yes, but that requires getting access to the video stream, which isn't easy
>>> in a "camcorder" environment.  Separate video recorders actually cost more
>>> than camcorders.
>> What precision of anything do you need?
>>
>> Expect 50-100 ppm oscillators in there.
>>
> 
> Yes, but probably fairly good in the short run, and if you were recording
> almost any sync signal (e.g. The 1kHz sine) you could calibrate that out.

Certainly, this is what I expect.

A very simple approach is to pull the audio-track down and post-process 
it with say Matlab/Octave to pull out the frequency, if it is 1 ppm low 
then the sampling rate was 1 ppm high. That's about all the things you 
need to know from that I think.

>> Recorded material is compressed. Consider how that affects your readout.
>> Recorded material is said progressive/interlaced. This may not reflect
>> how the CCD is snapshot and readout. You may not have the expected 1125
>> or 750 lines per frame, and exactly what frame rate do you have?
> 
> I think that's something they'll need to experiment with...

Indeed. I think a pair of (fast) diodes blinking at some adjustable rate 
and some motor rotating a pointer or something at a rate of 1 or a few 
turns a second should be interesting objects to test on. They should 
both be able to illustrate interlaced/progressive scanning and rate of 
line-scanning/snapshot. The rotating object is easier to understand for 
students while the blinking diodes can require a little more involved 
discussions. Maybe three or four in a row would even more illustrative 
since you can make the rate such that every other should be dark if you 
adjust the frequency just right.

I think grey-scale contrast should be at about the maximum for best 
result, since both vertical and horizontal sub-rate sampling can exist. 
Keeping the picture free of other distractions can also allow the 
bit-budget to be directed to the object of interest.

For your purpose I think progressive scan would be the best, if available.

Cheers,
Magnus



More information about the time-nuts mailing list