M8T problems

0 votes

I have bought uBlox M8T from CSG Shop (http://www.csgshop.com/product.php?id_product=205) and i'm using Sparkfun antenna VTGPSIA-3 (https://www.sparkfun.com/datasheets/GPS/embeddedsma.pdf). I took laptop to my car and after few rides I see 3 problems:

1) Very low accuracy - sometimes error is about 10-15m

2) No fix when driving > 50km/h

3) I cannot convert my files from .ubx to RINEX using rtkconv

What could be a problem? How can I fix it?

I tried to change frequency, only GPS, GPS+GLONASS, SBAS. RAW messages are enabled (I can see it in RAWX tab). I'm using Windows 10 so I had to remove Sensor API driver and use COM port to be able to pull GPS config.
by som4one asked Jan 12
by som4one edited Jan 12
0 votes
Make sure to set appropriate dynamics, a static setting will not work well for moving vehicle.

Use a magnetic mount antenna placed on the roof of the vehicle.

Make sure you have a build of the tool that supports the RAWX format.
by clive1 answered Jan 12
How to set dynamics and which should be appropriate?
One of the UBX-CFG-NAV commands as I recall, set to vehicular. Review the protocol specification for exact details.
How much impact can have antenna on the results? I need to prepare some equipment with antenna inside the plastic box and wonder if this antenna is appropriate for M8T? Would you recommend something else?
The antenna and placement are going to be critical if you are looking for some level of precision.

Being inside a car is far from ideal. Roof or trunk mounting is recommended.
Ok, actually I resolved all of this problems. Thank you! I'm not sure what was the problem with converting to .obs files but when I changed baud rate to other than 115200 files are now able to convert.
I noticed another problem - DGPS. I cannot make it work. I can see that RTCM messages are parsed in UBX-MON-MSGPP however some of them are not accepted. I found somewhere information that you need RTCM 2.3 messages 1,2,3 and 9 and all of them are needed. There is a problem with message number 2 - it is "retired" (https://www.use-snip.com/kb/knowledge-base/rtcm-2-message-list/) and I can't found any provider which can provide message number 2. Is it really needed? I tried to test DGPS with stations even far away from me (even 1000km) just to see if gps will turn into DGPS mode.  Is this too far away?
I noticed that almost all reference stations in networks like euref-ip or igs-ip are streamed in RTCM 3. Is there any firmware to make M8T accept RTCM3 or is it possible to convert RTCM 3 to RTCM 2.3?
It doesn't need all the messages to work.

To be useful and describe the same set of satellites the corrections need to come from something close, the closer and more timely the better. I would probably look for 100 KM or less. Generally this type of DGPS has been replaced by SBAS (ie WAAS or EGNOS). It tightens the CEP by around 50cm, but you still have several metres of CEP to contend with. Now that SA is turned off, you don't have 100's of metres of dither to contend with.

It might be possible to transcode RTCM3 into RTCM2, the latter being pseudo-range based differential corrections. I personally think this is a pointless exercise as the results will likely not provide the accuracy you want/expect, and it will take a lot of effort and work on your part.

If you are looking for sub-metre accuracy, and to use RTCM3 carrier based observations you should use the NEO-M8P.
I wanted to utilize some reference station network which provides DGPS corrections based on virtual station, where you send your localization and get corrections based on your position. They declare accuracy about 0.25m. I would like to achieve accuracy < 1m but RTK and RTN solutions are too expensive for my project.
SBAS could be enough for me but only sometimes GPS can connect to EGNOS. It could be caused because of my location (east Poland).
How better accuracy will I get using differential GPS corrections (not RTK) comparing M8P to M8T?
Pseudo-range base differential corrections still leave a metre or so of unresolved error.

RTK Float might resolve to a 20 cm radius circle.
website banner