I did a bit more investigation over the weekend.
I was lucky enough to borrow another HPWC from a Model X owner while he moves house. I wired a 32 Amp blue "commando" plug onto it so that I could connect it to the outlet that I normally charge the Roadster from.
Firstly I checked that the new HPWC could charge the Roadster. With the DIP 2 in up (normal) mode, the Roadster paused for around 20 seconds before recognising the pilot signal and starting to charge, as others have reported. I did not see any HCS warnings like reported in the original post, but perhaps this is down to new 3.0 firmware. With DIP 2 in down (legacy) the car charged right away.
Next I tried all of the combinations of cars (Ampera - Roadster, Ampera - Zoe, Zoe - Roadster) on the 2 units. Everything worked except the Zoe - Roadster combination where the Roadster was the second one to start charging. At the point the ESS contactors closed, the RCD (GFCI in American parlance) in the distribution board upstream tripped out. This was repeatable. The Zoe is *very* susceptible to ground faults and deliberately trips the RCD when it detects one in the supply. I suspect it was detecting an isolation fault in the Roadster (more on this later*).
Next I connected the 2 units together in master-slave mode using the only thing I had to hand which was doorbell wire (I twisted it using a drill). Initially I set my HPWC to slave and the borrowed one to master, however on power up the one that should have been slave insisted on becoming a master too. This happened 3 times. They obviously knew about each other because when I tried to charge 2 cars they faulted with the code for 2 masters in the network.
I suspected maybe the wire wasn't good enough but just to be sure I swapped the order of master and slave. This time they powered up with the correct display of lights. I then got down to business testing the cars and observing the load sharing behaviour. It should be noted that DIP 2 was in the down (legacy) position on both units.
With the network set to 20 Amps max and using the Roadster and Ampera - because I can see the current draw on both cars on OVMS - I can see that the load sharing does indeed work. However I was able to confirm what the other owner above (Model S / Audi) found, namely that with one car plugged in but not charging the system seems to keep 6 Amps available for that other car. So with the Ampera plugged in but not charging the Roadster receives 14 Amps in this case. With both cars charging they get 10 Amps each. Once the Ampera is unplugged, the Roadster climbs back up to 20 Amps.
However I observed two strange events:
1) If both cars are charging and you command the Roadster to stop, either by the VDS or OVMS, the HPWC tries to restart it. First you get an Invalid Pilot Signal warning on the VDS and then after a few seconds that goes away and the car starts charging again. This is a pain, because sometimes you want to just charge a bit (e.g. cooldown charge) and then stop it. After the second time of this I got a new error, DMC FW: HCS Faulted warning, then the car went to white ring mode and I had to disconnect and reconnect the HPWC to the charge port. I was once told that the Roadster doesn't recognise a pilot signal below 7 A, so if the unit is trying to do a soft start from 6 A then perhaps this explains the error. Further work needed here.
Note this behaviour does not happen if you are using a single HPWC. If you command a stop, it stays stopped*.
2) If the Ampera is charging happily and you plug in the Roadster to the slave, the Roadster immediately grabs all the power (20 A) and the Ampera stops, with its master HPWC flashing the code for overcurrent protection. I then switched the Ampera to the other slave HPWC and plugged the Zoe into the master and again the Ampera was stopped while the Zoe took all 20A. I didn't have time to check the Roadster - Zoe combination but will follow up with this. I also want to see if the Ampera starts charging again once the other car is finished.
That was all I managed to test before packing the kit away on Saturday.
*On Sunday we took the car to the beach and back, around 100 miles. I wanted to investigate that auto-restart issue further so when we got back I plugged the borrowed HPWC back in again and set about checking it.
On the first attempt the HPWC faulted at the point of ESS contactor closing with the code for a ground fault. I unplugged the car and plugged back in and this time it worked, and I was able to confirm that after a few minutes of charging the car does indeed stay off when you command a stop and the HPWC is not in a network.
Then I unplugged and plugged back in and once again the HPWC stopped with a ground fault at ESS contactor closure. I again unplugged and plugged back in which cured it.
It is curious that I saw nothing like this during the Saturday testing. Either I was lucky or the drive has moved something which causes the fault. I have been getting 283 errors on the car for a couple of years, which have not been fixed, but this is normally during cold and muggy weather. Yesterday was warm and dry. I need to investigate this further because if I am going to obtain 3 of these units they must be able to reliably charge all the cars.