EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: mojoe on February 17, 2024, 11:04:08 pm
-
For GPSDO use, I know that it is recommended to use just one GNSS system, like GPS. I've also read that you want SBAS/QZSS off. I wonder, has anyone actually done a comparison of the various combinations, and the effect on timing? I'm thinking that since GPS and Galileo both use the same L1/E1 frequency, that the use of those two may not be as problematic as using GLONASS.
What effect would leaving SBAS/QZSS on after a survey-in and setting a static position have on timing?
If anyone could point to a study where these various combinations have actually been tested, that would be very useful.
-
In comparison with my rubidium, using a TinyPFA and TimeLab, I could see no difference in short term stability with my GPSDO, between using GPS, Galileo, or GPS+Galileo. I have left it set on GPS only.
If you cannot mount an outdoor antenna with a clear view, using a "hockey puck" magnet mount vehicle antenna on a window ledge with a disc of steel underneath it, will likely work OK. If you are not seeing a good number of GPS satellites, consider adding Galileo as well, but not the Russian or Chinese systems as they are not as accurately controlled as the USA or European ones. My GNSS module is a Ublox M8T with a roof mounted "mushroom" antenna from Leo Bodnar.
Having read various manufacturer publications on the subject, the recommendation is just as you found; use a single satellite service (GPS), do not use position correction input (there is no point as you are not navigating), use an outdoor antenna with a good view of the sky, "survey in" the location of the receiver over a period of around 24 hours or more, and use a timing GNSS receiver in preference to a navigation one.
For what it's worth, my own measurements between an M8T (timing) and a 6M (basic GPS only navigation) is that with the M8T you can improve the short term stability to about 50% less jitter, it sounds a lot, but we are splitting hairs and spending around 10 times as much for the basic GNSS module. I can obtain around +/-2 parts in 10^11 with an M8T and around +/-4 or 5 parts in 10^11 with a $10 eBay board with an Ublox 6m (or likely fitted with a Chinese fake version at that price). I also have a Leo Bodnar GPSDO and that is only +/-3 or 4 parts in 10^10. All these measurements are short term "per pixel" frequency difference measurements against a rubidium over a 30 minute period using an outdoor antenna. I am only interested in the short term accuracy for holding amateur UHF radio equipment on frequency, or for use with my counter and signal generator.
Using a high quality oven is perhaps more important than GPS modules/settings. I use Toyocom (NEC/Epson) TCO6703N ovens which have an SC cut crystal, I bought them several years ago on the surplus market.
Hope that helps a little.
SJ