RTK base station placement: why 'full satellite count' doesn't mean good signal

What happened

An owner troubleshooting repeated positioning loss and missed mowing rows on a YUKA initially ruled out the RTK station's location: it was under "pretty light" tree cover, and the station's own status page reported 40+ satellites received, which seemed to rule out a signal problem¹.

To settle it, they physically relocated the RTK base station from under that light canopy to a fully open spot in the middle of the yard, with no structures or trees nearby, then remapped the same area and re-ran the same mowing task. Under the original canopy location, the mower had dropped to vision-only positioning through most of a no-go zone and missed multiple rows; after the move, the mower held a full satellite position lock through the same no-go zone and missed nothing on the equivalent pass¹.

The key takeaway: satellite count isn't signal quality

The RTK station's "40+ satellites visible" reading did not reflect how well those signals were actually getting through. Even light leaf cover was enough to attenuate the signal strength on enough satellites to degrade the mower's position lock, despite the station still reporting a high satellite count. The owner's explanation, consistent with how GNSS reception works: L1 frequency (used more heavily close to the mower) is attenuated more easily by obstructions than L2. So a station sitting under even modest cover can show plenty of L1/L2 overlap at the station itself while the mower, working under the same or worse cover elsewhere in the yard, struggles more¹.

Practical implication: if the base station's satellite count looks fine but the mower is still losing position lock or dropping to vision-only navigation, don't rule out placement: a light obstruction overhead (not just obvious heavy tree canopy) can be enough.

Don't put the RTK station in an attic

The video specifically calls out attic installation (roof decking, a waterproofing layer, and shingles stacked on top of the antenna) as a setup that will attenuate the GNSS signal far more than the light tree cover that caused problems in this test. The stated rule of thumb: the RTK station needs a clear, open view of the sky, not just "outdoors" or "under some roof."

Mammotion's own support account says the same thing. Asked by a LUBA 2 AWD X owner which of three spots to use, it advised against the one under a roof: "the roof cover will significantly weaken the signal, which may cause the robot to drift in positioning and fail to return to the base normally", and recommended a spot "close to the socket + without tall obstructions around"³.

Moving the RTK station destroys all saved maps

Confirmed directly in this test: as soon as the owner told the app the RTK reference station had moved, it warned that all existing maps would be cleared, with no option to preserve or migrate them¹. This matches the cross-model fixes-that-work-across-models page's own "moving an RTK base station" entry, which notes other owners have disputed whether relocation always forces a remap. This test is a second, independent data point on the "yes, it does" side, not a final resolution of that dispute. Plan around this before moving a station: if the current location is questionable but maps are still usable, consider remapping once, intentionally, rather than discovering the wipe mid-troubleshooting.

Don't pull the base station pole out by force

A secondary tip from the same test: the owner tried to wiggle the mounting pole out of the ground by hand, found it was planted solidly, and warns against doing this: forcing or wiggling a firmly-set pole risks breaking it at the base. If a station needs to come out of well-settled ground, work it out gently (the owner ended up using a weed-puller tool) rather than wrenching on the pole itself.

Supporting data point: RTK range itself is not usually the bottleneck

A separate, shorter range test on a LUBA 2 AWD Mini found good positioning maintained at roughly 200m (650 ft) from the base station, including with the station out of direct line of sight behind a rise in terrain, and on slopes up to 28°². That test didn't include tree canopy (explicitly deferred to a follow-up), so it doesn't contradict the attenuation findings above, but it does suggest that for most residential lots, horizontal distance and line-of-sight to the station are rarely the limiting factor; what's directly over the station and over the mower's working area matters more.