I figure this a good thread to report a recent test.
TLDR: The Base Assist Mode is a NICE feature, and works well with a HAS/E6 corrected Base if you don’t have a NTRIP source available. I’m also impressed with the simplicity of using the ESP-NOW radios with Sparkfun RTK devices for local Base/Rover.
I wanted to test a PostCard & HAS/E6 with the “new” Base Assist Mode.
For reference, this is the documentation I found, but there may be more:
and
I started with a Factory Reset on the PostCard, on V3.3 FW. I let the PostCard cook in Rover mode using HAS/E6 for ~30 minutes on a tripod and 2M pole. The reported Rover Accuracy was eventually 0.050m (RTK Float) with a mini choke ring antenna.
With a Bluetooth serial terminal, I switched the PostCard to operate in Base Mode using Base Assist. I use a wireless serial terminal to not monkey with the USB-C, the PostCard’s power source.
I watched the terminal and copied the coordinates it used for the Fixed Base Position (it’s HAS solution), to compare later.
This PostCard (now a Base, using it’s HAS/E6 Position) started transmitting RTCM corrections to another PostCard Rover using their ESP-NOW radios.
The Rover performed as expected with RTK Fixed solutions.
Now the interesting part:
Later I used PointPerfect Flex RTCM to determine the coordinates of the choke ring antenna and compared that to the HAS/E6 position that Survey-In captured in Base Assist.
The 30-minute HAS/E6 solution used by the Base Assist Mode was fairly close to the PointPerfect solution :
| HAS vs PP (truth) | Measurement (cm) |
|---|---|
| North Displacement | -3.40 cm |
| East Displacement | 4.09 cm |
| UP Displacement (Vertical) | -13.55 cm |
| Total 3D Distance | 14.56 cm |
Horizontal Only - we could call this a ~5cm accurate Base Position from HAS/E6
3D - we can say it easily falls within the 20cm “published” expectation for HAS/E6 in the USA.
The table above includes the differences in the reference frames from the 2 correction sources. Google says GTRF and ITRF are considered interchangeable, but Google also says there can be up to a 3cm difference between these reference frames. A measured horizonal accuracy of 5cm for HAS isn’t bad when you consider the possibility of up to 3cm of error between the reference frames that we are comparing. But to be completely fair and unbiased - that could also mean 5cm + 3cm => 8cm Horizonal 
. Either way, it’s still a great result for a free & over-the-air correction source.
Note: The table above was with a PostCard on RTK Everywhere V3.3.
I also used the custom config script with QGNSS on a LG290P breakout, for better vertical results after 30 minutes of convergence.
| Custom HAS vs PP | Measurement (cm) |
|---|---|
| North Displacement | 2.11 cm |
| East Displacement | 5.83 cm |
| UP Displacement (Vertical) | 0.20 cm |
| Total 3D Distance | 6.20 cm |
The same script can be used on a Postcard, but you need to change the baudrate in QGNSS to 230400 when connecting to “Serial B”.
All Tests were performed during “good” local atmospheric conditions this morning.
