Thanks Tridge.
I have a question. Is the data table exported by MISSION PLANER?
@Mirko_Zhou
I got a message back from the vendor that the mounting holes on this GPS
https://www.csgshop.com/product.php?id_product=266
are 65mm apart. Thought I would share that with you in case your thinking of getting one.
Finally got to play with my U-blox C099-F9P eval boards. As was the case when I set up the C94-M8P, it took a while to figure it all out. The documentation was a little better this time and I had some familiarity with Ucenter but there were still some unclear issues.
I have the base and rover boards talking over WiFi (the WiFi setup was very easy). I will not use this configuration with on my vehicle. I will use Lora (that’s for another day right now).
I am impressed with the “reported” accuracy so far. This plot is the deviation map from Ucenter. The rover has been stationary for over an hour and the deviation is as advertised about 1 cm.
I also received an Ardusimple simpleRTK2B today.
Interesting! My results were not that good. Have you tried looking at a second fix?
I just wonder if the wifi connection had anything to do with it? I just used mavlink.
Did your vertical stay stable as well?
I am working on using LoRa for the RTCM3 as I do now with my C94-M8P base and rover comm. Hopefully I will have that going today, although it will be using my C94-M8P as the base, so not a perfect comparison with the above on wifi.
I did not check the elevation. I forget some of you care about that! 
Why not comparing F9T vs M8T ?
The main fact is there is a time mark to report precise geotag which is not possible with “P” series. P are more for precise orbit so it means it’s better for a base station than a rover. High clock rate is important for rover and that’s why it’s better …
T series : Additional UBX messages that provide timing and signal alignment data, as well as raw measurements.
Other “timing” features are T-RAIM ----timing receiver autonomous integrity monitoring — continuously checks that error contribution from a satellite is not excessive, or it removes that satllite from the tminig computationSingle satellite mode: after 24 hours or so of self-survey to refine and lock-in
1 Like
Thanks, that is interesting. I get geotag images on my camera with a m8p currently.
Would the f9t make the data more accurate?
If that is the case I wonder why we do not see the f9t being used much?
m8p doesn’t have time mark pin so you will not have message in Ublox logs … Not very usefull and if you geotag with ardupilot logs, you will see that EKF introduce a lot of precision and accuracy in trajectory due to EKF itself so not really usable for photogrammetry purpose. I already compared in post flight reach rtk vs pixhawk navigation and there is a huge difference in some turns.
Example : (green = reach, red = pixhawk 1 arduplane logs)
About accuracy and precision of F9T : it depends. Multi band allow quick fix but most of commercial GNSS unit switch from multi to single band if baseline is shorten than 10km to skip noise in phase.
Thanks for the graph and explanation. This brings up a few questions:
I wonder how my drotagx gets the gps data? It’s my understanding that it does not interact through the FC, so I am not manually tagging or checking through the logs.
On the other hand I found tagging data to be worthless anyways, because the elevation reference in ucenter is bogus and not related to a level surface. This is because no geoid files are used by ucenter for the height computation.
Ucenter has a EGM96 geiod model but it’s large grid averaged so elevations are only ellipsoid related. It’s off 5-8 meters in my town, that varies due to geoid heights.
Ground control still the best for now.
Thanks again for the f9t explanation.
kikislater,
This post is ocnfusing to me. The F9P has an EXTINT for precise GEOTAGGING. Are you talking about something else?
If any of you are trying to configure an F9P, this Youtube video was a huge help to me: https://www.youtube.com/watch?v=FpkUXmM7mrc. I had manually done all the steps to get my C099-F9P boards working over the built-in WiFi and it was very hard for me, but I finally got it. I didn’t want to use WiFi in the long term, as I used LoRa for RTCM3 corrections on a C94-M8P system now. Someone pointed out this relatively new video, so I thought I would give it a try. It was super helpful. I think the only changes I made were baud rates.
I am very happy with my F9P system. I have used both the Ublox C099-F9P boards and one Ardusimple SimpleRTK2B. All work awesomely.
1 Like
Ok didn’t know they use this pin on P series nowaday but even with that for rover, it’s better to have high accuracy timing and you could only have it with T series.
1 Like
Hi, I’m looking to replace my M8N with
this
or (looks like this one may work by itself mounted on the copter/drone???)
will either of these work as a replacement for a M8N? or do I have to have a ground station with it?
kind of wish I didn’t upgrade the M8N to 3.01!!!
I just started reading about this RTK tech but I’d like to order the hardware I need since I’m in the USA, Long delivery time…
any info would be greatly appreciated…
thanks
Hi, I’m also wondering if any of the M8"s will be a good upgrade to the M8N, one that can handle GLONASS+GALILEO, etc… at 5khz? I’m not worried about an onboard magnetometer but its nice to have.
If one uses the base and rover configuration while doing photogrammetry, or really any other job where high precision location accuracy is needed, then you will be very satisfied with the results. If later you find that the base location was incorrect, then it would be a simple matter in post to adjust all the rover’s locations by the same deviation mathematically, in post. Easy. In the meantime, all your rover time and location stamps will have high accuracy, even if they are all off, they are all off by the same amount.
Thanks @tridge. Have you tried to use a pair of F9Ps with one as a base?
I’m curious as I tried to use both C94 boards from Ublox and S1216 from Navspark in the same config, one board is a base, another identical one is a rover. What I found strange is, the altitudes reported during the day drift a lot, and the altitudes at the same location can be 10-15m different if recorded on different days. Both systems behaved the same, with navspark having a bit poorer altitude estimation. It would be interesting to see what F9P is capable of before investing in another RTK pair.
I’ve only used pairs of F9P in a moving baseline configuration. I haven’t used a F9P as a static base, although I’m sure lots of other people have.
That is strange. Were they reporting a RTK Fixed solution?