It’s getting to Mission Planner and therefore getting to the autopilot so long as the telemetry protocol is correct.
Make sure the SERIALx_PROTOCOL for your telemetry radios is set to 2. If it isn’t, that’s the problem.
Otherwise, connect the autopilot via USB, make sure you are outside, and see what happens. This way we can eliminate the telemetry radios as an issue.
To my delight, during today’s test flight with the Quectel GPS[0].Lng and POS.Lng (and Lat) were in good agreement, in contrast to earlier flights:
And the landing was now also accurate to within 15 cm.
I can’t yet say what the reason for this difference was. Individual flights certainly don’t say much. So either I extend the small copter so that the Quectel and the Ublox GPS fit both on it or I build a mount so that I can quickly swap them. The waypoint flight route remains the same before the copter has to land again at the home position via RTL.
The timing is still much more irregular than with the UBlox GPS.
Ublox:
Rolf
Good news. I checked the protocol for serial 7, the telemetry radio, and it had been -1. I changed it to 2 (mavlink2), took it outside and let it find 25 satellites and NTRIP was started in setup, and I got “GPS: RTK Float” in the HUD pretty quickly, followed some time later with “GPS: RTK Fixed”!!! Thanks so much, Yuri, you are, as always, awesome. I believe this means that NTRIP correction data is getting into the LC29 and correcting the GPS position. Concerns are that the little rover on the map is floating all over the place, perhaps a circle of 2-3m diameter or so, with physical gps stationary. The HDOP at the bottom of the map was varying from about 1 to 2. That seems a little high for a RTK Fixed position?!? So, Yuri, I still see lots of posts that seem ambiguous about the LC29, including Rolf’s just above this post. So to all of you with experience with the LC29, after looking at your own and other people’s results, what is your final opinion about the LC29? Can it be used for a robot mower? Can it achieve accuracy of 1cm with NTRIP fixed position? Is there anything else I need to do to improve accuracy? The nearest Florida NTRIP base station is 8km away. Per the title of this thread, budget is important. The F9P may be better but $300 more is not worth it to me to do this project at all.
You’re asking me to make a judgment call on a GPS I configured once, flew a couple of times, and otherwise have seen very little on which to base a good opinion (no discredit to those who’ve contributed, but Rolf is getting mixed results, and there isn’t much else to go by).
If pressed to make that call today, I’d call it marginally better than an M8P, and significantly worse than an F9P, UM982, or Mosaic X. As such, it will work, but I wouldn’t put it on a large vehicle or critical application.
That’s only worth a grain of salt. We just don’t have enough good test data, and I’ve been busy with other projects and a real job.
BTW, an ArduSimple SimpleRTK2B starts at just under $200. I use them on most everything.
Thanks so much for the insight, even though its early. I understand the lack of data so far. I’m certainly going to continue with my project and might even have some results to share later. I saw Rolf’s results which are better than previous but still discouraging. Even in the first 10 minutes, I’m a little concerned that my rover location was wandering in MP after RTK fix and HDOP was high. My “large mower” is only about 40# MOL and about 2’ wide; my backyard large, fenced, and free of most obstacles and hazards.
You mentioned F9P, UM982 and Mosaic X as 3 top RTK GPS’s. And LC29 much lower, with M8P slightly worse. If I were to call the top 3 GPSs as 10/10, and maybe LC29 5/10, M8P 4/10. Where would the ArduSimple SimpleRTK2B fit? If I have trouble with the LC29, and the rest of the project works well, I might consider an upgrade. $200 is annoying, but after all the time, money and effort, might be worth it.
I understand time constraints, but Yuri, I hope you, Rolf and others might continue the investigation as time and energy permits to help us all get the best results.
And most of all, a personal HUGE THANKS to Yuri primarily and the others who have also contributed and worked this project. The community of builders/developers is unbelievable. Without that personal help, I would have a paperweight.
RTK Float is only marginally better than not having differential GPS at all. You’d expect to see some drift until an RTK Fixed state.
I’ll watch for the RTK Fixed. I see an ArduSimple SimpleRTK2B Budget for $197 without soldered pins. It seems to have the same chip as the F9P. I assume that would just replace the LC29 board, as is? I’ll remember that if the LC29 doesn’t pan out.
You want one with a Zed-F9P if you’re using u-Blox based boards for precision GNSS. The board you referenced is based on that module and is the one I use most often and is a direct replacement (and upgrade).
Yuri, Fantastic. I’ll make a note and buy that if the LC29 looks inadequate after reasonable testing. To configure the SimpleRTK2B, I assume I just set baud rate to 460K and plug it in? (
) After building a tank to do testing, getting RC eqmt, FC, GPS, 915mHz radios, and all the connectors, chargers, battery, etc, I feel like I’m committed. Tomorrow its finish config, install elec on tank and send it out for an autonomous run. Then to re-watch all your videos on fences, routes, waypoints, etc, etc. It looks like many of the errors you found a couple years ago have been fixed.
Does YGWYPF sum up this module?
Pretty much. It’s decent for the price, but it’s not amazing anyone.
It’s right next to buildings and very low to the ground. Expected behavior with any GPS.
Well, I got my LC29 with GPS fixed to a 6 waypoint autonomous route today. It went back to within an inch (or better) of where it was supposed to. Compass seems a problem, though, since it wanted a wide curve to go from one waypoint to the next. Should I spring $40 for the RM3100? My current compass is the $6 Amazon special QMC5883L. Or is it likely that the 5883 will be OK with repositioning on the rover?
That sounds like a tuning problem, not a compass problem. Once again, start a new topic for that.
I had intended to carry out systematic measurement flights with both GPS (Quectel LC29HEA compared to u-blox F9P) To do this, I had marked a spot on the airfield where I would arm the system, so the homepoint is located just at this spot.
After a few waypoints the copter should land at this spot by RTL. For each flight, I wanted to note wind, distance to landing point measured by a folding rule and measured by GPS.
But during the first flight the number of satellites indicated by LC29HEA suddenly decreased rapidly without recognisable reason, the GPS switched to RTK-float. Due to the positional inaccuracy EKF switched to LAND mode.
The connection to the NTRIP server was uninterrupted. The UART[n].Rx/tx log datas (GCS telemetry) showed no indication of an interrupted data connection. The tlog file also showed no indication of a connection lost to the NTRIP server (GPS_RTCM_DATA.len). Even if the telemetry had been interrupted, the number of satellites should not have dropped so drastically.
The fault occurred at a height of approx. 18 meters in an open field without any shadowing. I had previously flown this copter for hours with the UBlox with an otherwise completely identical setup without any problems. I can’t find any other reason for this behaviour apart from the GPS.
Rolf
Hello Guys!
Im really impresse with the work done on this topic, that even get me to register on the foroum.
I have a couple of question for you guy maybe someone can give me more clarity, I see for the module LC29Hxx have different version one even called LCH29HBA, that include RTK+DR.
Is DR a determine factor important at the moment of choosing the chipset?
I will use a couple of drone I have to do a light show, but seems to be GPS+RTK is need it in order to do it.
Link of the product: Lc29h Gps Bds Galileo Glonass Module Lc29h Aa Lc29h(aa) Lc29h(ba) Lc29h(ca) Lc29h(da) Lc29h(ea) Lc29hea - Buy Lc29h lc29h Aa lc29h Gps Bds Galileo Glonass Module Product on Alibaba.com
DR usually means an integrated IMU. At a glance, it’s probably not important at all, as we would have to determine the message set used and potentially develop firmware support.
Thank you so much for the pointer. I will use the one you’re working on, the LC29HEA. Once I receive the board, I’ll get back here and help with any development you may need.
hi, try to set baud rate to 460800 when using NMEA type message.
To reduce serial data throughput issue.
