Saci,

?

You posted a chart from a program called “NTP Plotter”.? This looks like a measure of the offset between the system clock and a reference clock.? The chart doesn’t provide a lot of detail on the reference clock.? ?But let’s assume the reference clock is a PPS signal from a GPS receiver and additional assume that there is no significant latency in measure the difference between the PPS signal and the system clock.? In this case, the chart demonstrates that the system clock with accurately synchronized with UTC.? I have no trouble believing the system clock can be accuracy synchronized with UTC (as I think I said in an earlier post).? However, this is not good enough to generate reliably accurate timestamps for the incoming image data from a camera connected to the system via USB. ?As I tried to describe earlier, the more significant latency issue is the timing of the data: the delay between the start of the exposure in the camera and the time the image data arrives in the code that requests the system time (for the timestamp).? ?This “image communication delay” is highly variable and dependent on a variety of factors which are not in our control.

?

Steve

Hi all

?

Thanks for the interesting discussion.

?

Another source of 1PPS flashes are GPS dongles such as the Ublox7 et al. Cheap off the self solutions.

?

These should work with software such as NMEATime2 to bring the PC clock well within the <1sec accuracy required to know which seconds the flashes correspond to while using capture software that uses the PC clock for timestamps. All you need is a USB extension lead to use the dongle as both a flasher and a ?time source. No need for Internet for NTP.

?

Steve P, I hope to find the time to test this out prior to the next TTSO.

?

Cheers,

Steve R.

?

“To Steve, there is a delay between the GPS time and the recording of the video, this is easily adjusted with a PPS LED flash. ”

?

You can use the PPS signal of a GPS receiver to measure the difference (delay) between GPS time and the System timestamps.? But this GPS/SystemTime delay can vary significantly based on the factors I mentioned earlier (e.g. Windows process starting/stopping, other device drivers, Windows updates, etc).? If you measure the GPS/SystemTime delay now, it may be different five minutes later and different again an hour later.? As a result, you cannot generate reliably accurate times by adjusting system timestamps.? For reliable timestamps, you must reference the exposure times directly to the PPS signal via a methodology with known accuracy.

?

For example:

Various “flasher” devices (e.g. TimeBox, Aart’s flasher, etc) provide a direct reference to the PPS signal by marking (PPS LED flashes) exposures at the beginning and ending of an event recording session.

?

Steve

?

The short answer:? Use a GPS flasher.? NTP is not a good idea.

Although it is possible to accurately synchronize the PC clock with UTC, you cannot easily guarantee enough accuracy with the timestamps for the exposures.? The major challenge is measuring the delay between the start of exposure at the camera and the time the capture software requests the system time (to generate the actual timestamp).? This delay varies based on several factors which are not in your control:? the camera firmware, the camera configuration, the camera device driver, all other windows device drivers, and all windows subsystems.? The greatest risk is probably Windows.? Windows is constantly firing off various automated processes which may, or may not, impact the delay factor mentioned above.? In addition, the overall behavior and timing profile of Windows can change with every Windows update.? Even if I took the time to measure this delay prior to every recording session, I would not trust the times due to the lack of control over other drivers and processes in Windows.? Window, Linux, Mac are all about the same – none of these operating systems are true real time operating systems where the system can enforce timing requirements.

?

GPS Flasher:

If you are comfortable with small electronics projects, you could build a flasher based on one of Aart Olsen’s Arduino designs (posted to the files section of the main IOTA mailing list).? Otherwise …. a group of people is currently working on a “productized” GPS flasher.

?

Steve