Clock synchronization to the midday sun
Synchronization to the midday sun is turning out to be an excellent exercise in engineering design.
Here are the three steps:
1) Begin with a concept, theory and initial design.
2) Collect data to support and verify the theory.
3) Test, test and test again.
As I intimated in the previous post, the objective is to re-calibrate the clock against the midday sun. Three difficulties were anticipated:
1) What if there isn't a south facing window to locate the clock?
2) How to account for cloudy days or clouds obscuring the sun at the critical moment?
3) What happens if someone moves the clock?
To make the design simple for this first iteration of the prototype it would be best to eliminate problems #1 and #3 by insisting the clock is permanently located in a position that faces geographic south exactly if you are in the northern hemisphere.
Leading and trailing edges
The sun sensor responds to the infra-red rays of the sun as it passes overhead at noon. Readings from the IR photo-diode are taken once every minute. The leading and trailing edges of the signal are very sharp and the rise and fall times occur in less than a minute. The total duration (width) of the sun signal is about 10 minutes.
Registering both leading and trailing edges of the sun signal provides two opportunities per day to pin-point the exact midday time. Detecting the first rising edge of the signal is straight forward. Registering the final trailing edge of the signal is more problematic in the event that there are intermittent clouds in the way. For this reason I have chosen to ignore the trailing edge and use only the leading edge.
I collected data over eight consecutive sunny days this past week. As I do not have a south facing window, the sensor was aimed to trigger at 11:30 DST (Daylight Savings Time) which is actually 10:30 standard local time.
Over the eight day period, consecutive daily readings were within 1-minute in agreement. What is interesting is that there was a progressive shift of 5 minutes earlier over the 8-day period. My conclusion is that this is a result of my location, latitude 43° N, and time of 10:30 and date, Aug 24 - Sept 01, 2014.
So now my next quest is to find a south facing window and use an exact midday sighting of the sun.
PREVIOUS
MSP430 Tutorial - Index
Synchronization to the midday sun is turning out to be an excellent exercise in engineering design.
Here are the three steps:
1) Begin with a concept, theory and initial design.
2) Collect data to support and verify the theory.
3) Test, test and test again.
As I intimated in the previous post, the objective is to re-calibrate the clock against the midday sun. Three difficulties were anticipated:
1) What if there isn't a south facing window to locate the clock?
2) How to account for cloudy days or clouds obscuring the sun at the critical moment?
3) What happens if someone moves the clock?
To make the design simple for this first iteration of the prototype it would be best to eliminate problems #1 and #3 by insisting the clock is permanently located in a position that faces geographic south exactly if you are in the northern hemisphere.
Leading and trailing edges
The sun sensor responds to the infra-red rays of the sun as it passes overhead at noon. Readings from the IR photo-diode are taken once every minute. The leading and trailing edges of the signal are very sharp and the rise and fall times occur in less than a minute. The total duration (width) of the sun signal is about 10 minutes.
Registering both leading and trailing edges of the sun signal provides two opportunities per day to pin-point the exact midday time. Detecting the first rising edge of the signal is straight forward. Registering the final trailing edge of the signal is more problematic in the event that there are intermittent clouds in the way. For this reason I have chosen to ignore the trailing edge and use only the leading edge.
I collected data over eight consecutive sunny days this past week. As I do not have a south facing window, the sensor was aimed to trigger at 11:30 DST (Daylight Savings Time) which is actually 10:30 standard local time.
Over the eight day period, consecutive daily readings were within 1-minute in agreement. What is interesting is that there was a progressive shift of 5 minutes earlier over the 8-day period. My conclusion is that this is a result of my location, latitude 43° N, and time of 10:30 and date, Aug 24 - Sept 01, 2014.
So now my next quest is to find a south facing window and use an exact midday sighting of the sun.
PREVIOUS
MSP430 Tutorial - Index
