One interesting thing I have noticed since the Postcard got firmware enabling RMS is that on my facet typically once I turn it on and get 20+ satellites my horizontal RMS drops to 0.014m or the theoretical maximum whereas my Postcard with the L5 antenna drops in around 0.018 to 0.031m.
What is strange to me that doing 3 minute observations my delta average on surpad will be tighter on the Postcard typically in a 3 to 8mm horizontally and my Facet will typically be 5mm to 11mm horizontally.
I understand that u-blox and quectel probably compute RMS differently (Isnāt it a standard formula?) or is something else different. I will say the PDOP, HDOP, VDOP values on Postcard are always lower and the Facet my verticals would never really go below 1.
Just trying to wrap my head around why the facet shows more accuracy in RMS but generally gets beaten by the Postcard
Since the Um980 that is in the Torch is made by unicore (edited) I am curious if it too pegs to 0.08 with a good RTN connection and maybe itās just poorly formulated on the Quectel chip.
The only technically correct comparison would be to connect both receivers to the same antenna with a splitter, collect observations over a significant timeframe, run them through the same third-party processing tool, find the error distribution for each dataset, and compare those results with what the receivers report. Otherwise, itās comparing unknown to unknown.
Shouldnāt RMS be a standard error though? Calculated as such. I do recall the facet being an hungry antenna so maybe it can get right down to 14mm almost all the time.
āShouldā isnāt a good ground for any technical assumption. There is a pretty straightforward metrological approach to this, if you are interested in a solid answer.
I fully agree with you which is why I am starting to go down the rabbit hole. Unfortunately I have a facet and I have a postcard with l5 antenna, I would need to assemble a ZED and connect it to the l5 antenna to test these hypothesis. I am curious if someone like @sparky might have some scientific data to avoid spending more. I also believe @rftop has more equipment than I do.
The only hypothesis without equipment i can come up with is that Iāve been told the facet has a hungry antenna but my l5 antenna does lock onto more satellites due to increased bands.
At the end of the day I know which device has given me a better nad83/91 coordinate on a publisher benchmark and better geoid 18 heights but the RMS doesnāt correlate.
A fairly easy test that doesnāt require buying a good-quality RF splitter, matched DC isolators, and a bias-T would be to simply swap the antennas and compare results before and after. At least, it might hint at any sort of hardcoded coefficients. You never know what some creative firmware developers might have done, like adjusting the uncertainty circle simply based on the set of frequencies a receiver can āhearā. You also likely know as well how āmore satellitesā doesnāt mean āhigher actual precisionā - thatās why elevation mask exists, etc.
This is why Iām not worrying too much about whatever my receiver reports (even though it seems fairly accurate) except for the dB levels.
Yes and I thank you for the discussion. This is why Iāve been going out and putting equipment through itās paces on verifiable items whether itās established control or points weve captured with expensive Trimble equipment. I have done simply put my pole out with the l5 and postcard spun it off and placed the facet on and collected 3 minute observations compared etc. This is where and until now postcard had no RMS for RTK and the display floating higher (16 to 21) millimeters than the typical 14mm the facet almost always shows I began asking why. True the error circle is only one thing in the equation but repeatability and spread being another. And again granted this is on ORGN RTN. I will perform some tests again and try to get a spreadsheet of data.
@AORPLS , Iām expecting the std deviation is coming from the GST message.
As you know, the āUncertaintyā Values from any GNSS should be taken with a grain of salt.
If the receiver knew itās actual location to perform the calculation, there would be no uncertainty at all
I think you are simply noticing the difference in the Manufacturers and how they produce the metadata.
Itās funny to me because our Trimble constantly is saying RMS of around 0.02ā to 0.04ā horizontal yet I find the postcard hitting right up in there with it but quoting 0.10ā horizontal. Almost like the RMS isnāt accurate much at all. Iām gonna have to burn some more points with our Trimble and copy the calibration to surpad and see what I get. Granted these two will never equal each other but some distances between multiple points could be interesting to see.
And this comes from Trimble saying 8mm + ppm
Postcard 8mm + ppm
And facet 14mm + ppm
The ppm part for me is tough due to voodoo RTN from ORGN is versus single base solution.
Another thing that just dawned on me is I have 3 control points near my house I established with a total station (3 sets of rounds) the other day. Using OCRS low distortion I could go tag these points with gps and compare distances and angles to see what kind of errors I have. And do it with both the facet and postcard. Then Iām comparing measurements versus plane coordinates.
@TS-Russell how do I twist your arm to just send me a torch and I can provide lots of real world review and tests from a licensed land surveyor as well as impress the local chapter that Iām president of
And get a nice write-up slash tests in the Oregon surveyor publication as a nice advertisement but really a technical review.
do you know what these GST messages from the postcard layout to be?
above is RTK
here is on DGPS
$GNGST
235356
2.109
1.617
1.302
121.726
1.536
1.396
3.124*7F
$GNGST
235356.5
2.147
1.611
1.297
121.726
1.531
1.391
3.114*78
I am assuming the FW is pulling the 2nd to two last values which might be horizontal lat/long sum of squares being 28mm for both which would correlate closely to the almost 0.10ā I have been seeing in surpad.
