Corrections vs Constellations

I have a local RTK correctionn provider who provides GPS/GLONASS corrections over NTRIP.

My question is…

This gives me (for example) 13 satellites with corrections. If I turn on SBAS, Galileo, QZSS, I get a further 7 I corrected satellites.

Which configuration provides the
… quickest time to FIX
… Most robust solution.

In the RTKlib that you use is there ANY use made of signals withou correction?


I have done some testing, and have come up with one insight from my ,omitted sample set.
With only gps & GLONASS correction available, but with gal,qzss, and sbas enabled the unit appears to hold fix better once achieved. Especially on first FIX


Fix will always be more sustainable with a bigger constellation, but accuracy can suffer if one strays. This is probably the number one reason in modern surveying to use a total station or robot. You locate the control on GPS and GLONASS and you workmate comes behind you on GPS and SBAS and you don’t tie in. Who’s correct?

Hi @Simon_Allen,

The minimal subset that is required for RTK to function is 1002 message for 1Hz with GPS observations and 1006 message for 0.1Hz with base station antenna position.
As @michaelL said, it’s better to use mode GNSS systems to get a quicker fix.
Also, GNSS choice can depend on your country.

I’d recommend you to look through this docs section to clarify some points.


No that documentation does not provide guidance on my specific question.

Let me try again…

With corrections only available for GPS and GLONASS, is there any advantage in having Galileo, QZSS and or SBAS turned on? Is any use made within the RTKLIB engine of the uncorrected signals?

I have noticed that 5hz output seems to achieve FIX quicker than 1Hz is there any reason this should be so, or am I imagining it?

Why do you only have available corrections for GPS and GLONASS? Are you using a service?
As for frequencies, 5hz is faster so the theory is that it tracks movement much better. That doesn’t mean that is more accurate. It may actually introduce more garbage data than doing good. I have occasions where I use 5hz because I have a mount for my vehicle and do rough topographic surveys while driving. It maintains the fix better and is more accurate in this scenario.

Yes, I am using a networked service. This delivers either nearest, or networked VRS, but only for GPS and GLONASS at the moment.

Hi @Simon_Allen,

Yes, more satellites - the better for RTK.

Yes, it can be so. 5Hz on the rover is the recommended updated rate.

Agree, more satellite is better, with corrections for them all, but the question is how are uncorrected satellites used and do they help or hinder.

Hi @Simon_Allen,

Reach will not use that kind of satellites to determine its position.

So if using CORS for base, it makes a lot of sense to pick one where you can get the same constellations and with your rover (so, + GAL, SBAS, etc) ?

Thanks Andrew.
This is proving frustrating, getting a single considered answer.

Clearly prior to getting even a float solution the unit could use all the satellites, and probably does. Therefore in theory having satellites turned on that I know I do not have corrections for could help with that initial localisation. this would be especially true in a less than optimal location, say for example, urban canyon environment. Especially when using a CORS where you need a single solution to get a position to send to the network.

I accept that once you get corrections then the FLOAT and FIX solutions only used the corrected pseudoranges.

This is my view, it frustrates me when the one liner responses do not look at the whole picture and provide a considered response that looks at the whole question.

Please can you confirm the above?

With the EMLID units providing a solid fix, If the unit was a rover getting corrections from GPS and GLONASS it would be possible to calculate the corrections that should be applied to the uncorrected satellites. ( you do this when you send out base corrections) therefore it would be possible to either
A) calculate the corrections for the pseudoranges without CORS data and apply them with a lower weight in the next calculation
B) calculate the position with and without the additional satellites and their pseudocorrections and use this info when AR drops for the CORS corrected solution.

It is clear you are not doing this, which was the basis of my question.

With your current units it is not too much of an issue as I cannot enable Beidou and GLONASS, but the next gen of GPS chips outstrip the CORS networks in their available constellations.

Hi @Simon_Allen,

Yes, I can confirm it. As I mentioned in my previous message, Reach uses all the enabled satellites only for a single. As for float/fix, a rover will use the same satellites as a base even if it has more visible.


This topic was automatically closed 100 days after the last reply. New replies are no longer allowed.