the Leica GS18's I mostly use (and everyone other one I looked back that I've used last decade) are mm differences only.
It depends on the azimuth and your mask angle. With a 10° mask the GS18 maxes out at about 4mm at certain azimuths. With a 0° mask that runs up to almost a centimeter.
And I misremembered about the Javad, it maxes out at about 14mm, not a tenth of a foot.
Are you looking at antex or antinfo? Assume you are looking at L1 differences? I really need to look on PC with a proper screen much harder to make sense of on phone!
Also I would make the assumption that for base/rover RTK which I thinking here using less than a 10° is borderline negligent and I'm often using 15° (at least at the rover) which is the 2mm level in the antex file for GS18 as best I could see.
I can accept that if you are doing static GNSS using well adjusted high quality tribrachs then pointing all the receivers in the campaign to north is important.
But for RTK GNSS on a 2m pole (no matter how much you baby it) the run out/bubble going off adjustment (a minor bump during the day your assistant forgot to tell you about) is a bigger source of error to me; one the 180° observation 'pairs' solve. One is a cm issue and the other is mm issue. However if your pole is well adjusted (both bubble for plumb and run out accounted for) along with well adjusted tribrach under the base then having base and rover facing north will yield the most precise results.
As an aside run-out of the pole is not so much of a problem if you adjust your bubble so the tip and top are plumb at a fixed height and only take observations using that height, perhaps a good argument for fixed height poles? However this kind of like adjustment requires a total station to sight up and down pole plus lots of iteration compared to those pole jigs on the office wall so less likely to be done as often as you really need to.
@lukenz Also this will probably throw you for a loop. static. if you process with an absolute antenna cal file and do not orient the antenna correctly it can throw off the vertical. test show up to a cm in that as well. I agree that in rtk mode DEPENDING on receiver especially newer ones it is possible to be in the mm range. and lets say rtk is 8 mm accuracy + 1PPM just pulling a spec. thats in a perfect world and say the offset is 3mm its in the weeds. BUT it is something I can control and that 3 mm turns into 6 mm that's .02 ft plus the random error of the 8mm. Now let's say that we have only a few mm in the specs built between the same model base and rover. If no attention to detail is paid we have more random error and if we pay attention to details we control some of that error to me it's just makes for sound field procedures and takes no more time and allows me to control and be a bit better when things get close. if it were only .01 ft I am gaining then so be it. Yes pole run out 40 min buble. we deal with all sorts of errors we have as surveyors for years. Direct and reverse readings on total stations and transits multiple readings methods were designed to reduce those. I dont topo orienting it but when I am now establishing RTK control and a 1 mm + 1 ppm edm is shooting distances from that control and I am getting .025 at most 95% of the time on ground verifications at distances between 300 to 1000 ft on my job sites with good tribrachs and traverse kits I can trust that my procedures are doing what I want and I am happy. Yes we get the tenth or so sometimes we get .05 ft sometimes. Vertical is coming in very tight as well. I only do it for control and property corners and in the east we do more RTK for control why MONEY TIME but I dont buy it I did static fast static a lot when I was out west and wish we did more of that here but its just get rtk control and move on. Many of our projects here in my opinion should be static for primary control but its not in the budget they say.
