DATE-OBS not exact

Nineplanets

Hey all! As I'm doing astrometry I'm heavily dependent on having a very accurate timestamp in the DATE-OBS field in the FITS-file. I did a simple experiment to verify that the timestamp is correct, and for this I downloaded the atomic clock app for android, which simply shows the NTP-time for the device (in UTC). I then connected to the ASIAIR (Plus) and took a series of pictures of the phone simply displaying the time. I took a picture with just 0.001s exposure time. I also left the phone to properly synchronize to the NTP server.
Attached you will see a picture that shows the DATE-OBS field of this picture, that is showing the time on my phone. Note that I was running the latest firmware (2.1).

A few questions?

Why is there a time-difference of 850ms between the DATE-obs-field and the time (NTP-synchronized) on the phone?
Does anyone know what the source-data is for the ASIAIR for filling in the DATE-OBS field?
How does the ASIAIR even get its time-synchronization at all?

w7ay

Nineplanets Why is there a time-difference of 850ms between the DATE-obs-field and the time (NTP-synchronized) on the phone?

The ASIAIR marks the time when the exposure is requested, not the time the exposure is being taken. The 850 ms is the typical latency time of the firmware in ZWO cameras to actually start an exposure (see a different thread on this forum -- Camera discussion group, I believe -- in the past week or two about this). ZWO camera firmware does not report back the actual time the exposure finally happens after receiving the exposure request -- what is needed in the SDK of a serious camera.

The 850 ms number is not fixed either, and you can expect a jitter around that number for different exposures, and it will depend on the camera model and firmware version and USB bus activity.

Use some other software and camera model if you need accurate numbers in the FITS file header. There are some camera models with built-in GPS (from other manufacturers) that you can use for your purpose.

Chen

Nineplanets

w7ay
Thanks for the clarification, this is very useful information.
Do you happen to know where the ASIAIR gets its timesource from? Is it from the phone? Any idea of how accurate this is?

w7ay

Nineplanets Do you happen to know where the ASIAIR gets its timesource from? Is it from the phone? Any idea of how accurate this is?

I don't know for certain, but it almost for sure is getting the time services from the tablet's OS API (so there is a latency from getting the time from the tablet to the ASIAIR device), unlike INDIGO, which can get the time directly from a GPS dongle (in the case of INDIGO Sky, USB GPS dongle plugged into the Raspberry Pi).

If it is an Apple tablet, the tablet is most likely getting the time from an NTP server at time.apple.com (a bunch of stratum 1 servers, apparently).

This is what the UI looks like on macOS to select any other NTP server:

macOS connects to NTP once every 20 minutes or so (perhaps less often for iOS), and it depends on the crystal clock on the computer in between those times. A good TCXO Is good for 10ppm, but I don't know if the phones and tablets have that good a clock. So that should give you an idea of how (in)accurate the OS time server is.

This is a nice stratum 1 NTP server using GPS as the source, but it will be a cold day in hell before ASIAIR lets you connect to it.

https://www.leobodnar.com/shop/index.php?main_page=product_info&cPath=120&products_id=365&zenid=31ae819ca8d660a915988ac5f4ae7951

I have a couple of their GPS frequency references and they work like a charm.

Chen

Nineplanets

w7ay
Chen,
thank you for this.

Yes, I suspect it too getting it from the device (an Android phone in my case, not much of an apple-guy here). Only question is how accurate this synchronization is (NTP for example, requires the device to be connected for a while to improve accuracy).
I will go a different route when I need to take photographs with a higher accuracy for timing.
Given all of this, I think that sub-second accuracy is probably more than enough for most people.

Thanks!