Wondering which of the different chipsets in the sparkpnt fp’s will work best under tree canopy and in urban canyon environments ?
yes , i know it is not a simple issue or answer , but opinions / experiences are soughtPierre
Essentially all can be setup as a base or rover; for non-permanent/temporary bases with multipath issues we currently recommend a HAS-capable base + ESP-now setup as demonstrated here (2 RTK torches are used in this demo)
Thsnks. Ntrip would do the same , would it notb? Or am i missing something ? We are looking for cm level accuracy.
Correct. Having a local base station does not directly reduce or eliminate multipath errors at the rover. That Video explains how it’s helpful, when using the free HAS service, to have a local base with wide-open sky. The rover would be expected to drop its PPP Convergence during a momentary signal loss. Having the local base prevents needing to wait for reconvergence if HAS is the correction source, since the local base will maintain signal lock due to it’s better location.
For MultiPath, the best thing you can do is use a choke ring antenna. That’s obviously not really an option with the Facet FP Lineup or any all-in-one solution.
A ground plane would be the next alternative.
After that, it’s up to the GNSS receiver and firmware to recognize and mitigate the multipath errors.
Out of the 3 receivers in the FP lineup, the Mosaic X5 is the best at multipath, at the hardware level.
It’s hard to quantify how the 3 chip’s firmware compare.
I would “guess” that U-blox (X20P) spends the most money on FW development.
But I’d also say that Quectel (LG290P) is the most responsive to users for FW progression.
If a client asked me to build or source the best (inexpensive) ROVER for multipath conditions, I’d start with a Mosaic X5 receiver and a mini choke ring antenna. However, the Facet FPM (X5) would be substantially lighter and easier to use in the field (if the Rover is on a Rod).
But to muddy the water even more : Your workflow is the best protection against multipath. The best measurement is always a redundant measurement if you truly care about the position (especially elevation). I feel a check-point is absolutely required in marginal conditions. In the RTK workflow we Dump the antenna to force a mathematically independent integer ambiguity solution from scratch to collect the check-point.
Because RTK algorithms can occasionally lock onto a false fix (a mathematically incorrect integer ambiguity solution caused by multipath), a single initialization cannot be fully trusted in marginal environments. Every GNSS receiver can make this mistake, but it’s rare to make the same mistake twice. IE: the check-point likely wont give the same wrong results as the first initialization. When you don’t have agreement, you know something is fishy.
The best general advice is to choose a device with the form-factor that works best for 75% of your missions. Tweak your workflow for the remainder, instead of letting the edge cases drive your equipment selection. That applies to selecting a mechanical pencil, all the way up to which airplane to purchase - and everything in between 
Thanks Ryan. Right on the money !
I have an old pair of trimble 4000 sse units with geodetic antennas w/groundplanes. Sure were bulky beasts to carry around !
Pierre.