Re: About APCC and ISS tracking and PC Time


ap@CaptivePhotons.com
 

Dale Ghent said:

For NTP, there is the Dimension4. This implements a more aggressive NTP client than the standard one that's built-in to Windows.

For GPS devices that emit standard NMEA time sentences over a serial (COM) port, there is NMEATime2 ($20) or the older, free version, NMEATime.

If you are operating on Linux or other type of *NIX OS, then there are many to accommodate all sorts of time sources.
One thing to realize is that NMEA time is inherently inaccurate, exacerbated by USB serial delays. Most GPS chips also emit a PPS or 1PPS (1 pulse per second) electrical signal that is very precise, and can be used to synchronize the information from NMEA. Essentially the later says "the time is X" and the PPS signal says "right now".

Windows cannot easily do this, and none of the GPS' that attach via USB can do this.

It is unclear that driving a mount, even as fast as ISS, require this. VisualGPS / NMEATime2 (Dale's example but good one) has a lot of math added to try to pull out all the accuracy possible from UPS GPS without PPS. It's a good windows tool to get what is likely "good enough".

It is relatively cheap however to get more accuracy: A raspberry Pi with off the shelf code and a tiny bit of soldering will let you build a very accurate time server for around $45 ($10 GPS, $35 rPi 3B). The rPi will run GPSD to process the time signals, and coordinates with PPS on a GPIO pin. Built in CHRONYC (which is NTP in disguise) will provide a time signal other systems can coordinate with. I did one recently with:

https://www.amazon.com/gp/product/B07P8YMVNT

Note that while it has USB, you have to solder to the other pins to get the PPS signal. Interestingly this device is sensitive enough a get a decent signal inside a dozen feet from a window. The rPi can power it. Add a $5 12v to 5v step down and you can power the rPi from your 12v mount supply and strap the combination to your tripod. (Try to keep the GPS device, at least the antenna, a bit away from any wifi emitters as they can conflict -- that's generally a true statement not specific to this).

Then on Windows run the Meinberg NTP server and connect to the rPi (preferably over ethernet for low delay) and it will drive the Windows clock. It is unclear exactly how accurate this combination is, but certainly sub millisecond and probably a few dozen microseconds. Add internet time servers to the Meinberg list and it will fall back to them when the rPi is not available.

Again, it's not clear that NMEA derived time is inadequate, in fact some back of the envelop math makes me think it is quite good enough. But if you are a tad obsessive about getting accurate time on windows, this is a cheap alternative, and easier than trying to "fix" windows directly to use a GPS properly.

Linwood

Join main@ap-gto.groups.io to automatically receive all group messages.