BG7TBL GNSSDO-CM55 support

[jclark]

The new BG7TBL GNSSDO-CM55 is nice and works well with satpulse. It has a u-blox M8T inside, and the RS232 port gives direct access to that. But the PPS output is not just passed through from the M8T: it is disciplined. (On an oscilloscope you can see the pulse width is 10ms, but the UBX configuration is the normal 100ms pulsewidth.) This makes this a great fit for PTP.

It also offers the potential to support holdover (see #152).

However, it requires a bit of configuration to work optimally:

• configuration shouldn't mess with the time pulse
• we want the pulse width for handling intel NICs' timestamping of both edges, but the pulse width that the LEA-M8T reports doesn't match the pulse width we get, so we should not use that
• it would make no sense to use the quantization error from UBX-TIM-TP

We can already do this by manually configuring using satpulsetool gps:

satpulsetool gps -s 9600 -d /dev/ttyUSB0 --binary --pvt-out time,leap,survey,tai,off --save

and then in satpulse.toml doing

[gps]
# this is the default
configure=false
# useful only for intel
pulseWidth=0.01

I think we could improve the output of the box experience by adding a TOML configuration option (tentative name disciplined, default false), which, when true, has the effect of

• not configuring the time pulse
• not getting the time pulse width
• not enabling UBX-TIM-TP, but relying on UBX-NAV-TIMEGPS instead (this will mean we won't get pulse corrections)

I believe we could implement this entirely in daemon/gps, although we would need to extend gpsevent.SetMsgOptions.

It's not a huge win, since the user should really user satpulsetool gps anyway to increase the baud rate to 38400, so that get satellites output.

[SpandrelPanel]

@jclark It doesn't look like satpulse gps is able to assert the UBX configuration on the LEA-M8T used in GNSSDO-CM55 - at least for me. In case it makes any difference, I am using MAX3232 to UART on a CM5 and already set the baud rate to 115200 via TCP. Satpulse was built from source. Maybe a case of UBX protocol version mismatch? - the LEA-M8T is running 22.00.

Also, it might be good to include UBX-NAV-SAT messages as this restores satellite data in the satpulse web interface (as well as u-center).

I have found some interesting things about the GNSSDO-CM55 which might be of general use, shall I post them here?

[jclark]

What command did you try and what result did you get? I just did:

satpulsetool gps -s 9600 -d /dev/ttyS0 --speed 38400 --binary --pvt-out time,leap,survey,tai,off --sats-out sat --save

on mine, and it worked fine. My one also is firmware 22.00. At the moment, I have it connected via a real serial port on a PC. But I have also tried it with a CM4 using https://www.waveshare.com/cm4-io-poe-box-b.htm which has a terminal block with an RS232 interface.

I agree including UBX-NAV-SAT is good. satpulsetool gps --sats-out sat will do this.

I would be very interested to hear what you have found out about the CM55. I created a discussion for this #164