Yes, if we have details of the protocol. Dave, can you hunt around and try to
find the NMEA sentence definitions for the C3/C4 units?

Alternately, capture the NMEA output from your C4 and send it to me as we may be
able to reverse engineer it.

I thought about connecting a C4 to XCSoar back in 2004. Since then I've been out
of the game until the last year or two, but I think my conclusions are still
valid. I am currently flying a Std Libelle fitted with an SDI C4 running version
4.05 software which is driven off a Garmin GPS II+. What follows is straight out
of the V4.05 manual combined with what I've seen by looking at the GPS II+
output stream.

The serial interface between the C4's main and repeater units isn't documented
apart from listing the pin-outs, so I don't think its either accessible or
useful for the XCSoar interface: think of it as only carrying knob twiddling
from repeater to main and passing display information back to the repeater.

The only other data feed to or from a C4 is GPS input to it. It uses a standard
NMEA 0183 interface and requires only the GPRMB, GPRMC, GPROO and GPWPL
sentences. AFAIK it ignores all other sentences.

All GPS units, even blind ones like the Garmin GPS35, output GPRMC. This gives
the basic GPS location and time plus derived ground speed.

The other three sentences can only be produced by something that knows the task
data: GPRMB gives range, bearing and cross-track for the next way point,
GPROO passes the list of waypoints making up the task while each GPWPL gives the
location of one of the waypoints. Each second the GPS outputs GPRMB, GPRMC and
one of the other sentences. The latter cycle round in the sequence GPROO,
GPWPL,....

In other words of this data set, XCSOAR will normally just get GPRMC from the
GPS if it has no waypoints set in it: this would be the norm for a blind GPS.

As each GPRMC is received by XCSOAC it should be copied to the C4. If a task is
defined it is done. Otherwise it should immediately assemble and send the GPRMB
and whichever is next in the GPROO, GPWPL,... sequence. I think the C4 can do
wind calculations from the GPRMC data but it needs the remaining three sentence
types to do final glide calculations for the next waypoint and the task finish.

There seems to be no way to pass information about MC, bug and ballast, etc. to
the C4, so IMO its best to let XCSOAR handle final glide calculations though it
would be useful to build and pass the complete sentence set to the C4 because it
can calculate wind when its cruising as well as thermalling and some pilots may
still want to have the C4 calculate final glide in parallel to XCSoar.

I hope this is helpful information. Feel free to ask for anything I've omitted.

Here's a correction to my previous note.

I've now found that a C4 cannot calculate wind unless the GPS has an active waypoint, i.e., it needs to get all four sentences, GPRMB, GPRMC, GPROO and GPWPL, to do so. My previous note still stands except that there is no point in XCSoar sending anything to a C4 vario if no task has been selected and a default airfield has not been set in the airfields file.

I discovered this last season when I replaced my former setup, which used a Garmin GPS II+ to feed the C4 and as my prime navigation aid, with a Garmin GPS35 blind GPS to feed my C4 and a Binatone PNA running XCSoar as my main navigation aid. The GPS35, which only outputs a GPRMC sentence, was removed as soon as I found out that wasn't sufficient for the C4's wind calculations.