Thanks camiel and jasem for feedback.
It's been windy but clear here so not ideal for tight guiding but I continued with some tests to rule out variables, here's an update.
I tried starting server with -vvv option and enabled verbose logging, it created client Ekos log and server log both attached following a kind of dummy session i.e. startup, track/sync to target, plate solve, guide calibration ending in guide fail. So then I did some guiding movements N,S,E,W to ty to check instructions are being followed.
On first read logs are a bit overwhelming but I guess they will mean more to you. I noted several things during session
a. On screen, I could see a high precision RA/DEC stream, like 12 decimals in the server SSH window
b. in the Ekos log at about 19:37 calibration fails (typical state)
c. but at 19:40 I got calibrated, but this soon loses track and fails after about 2 mins (typical)
So, after i tried adjusting guide variables without success. exposure, reticle, guiderate 0.5/1.33, rapid on/off, RA only etc etc - no success.
It seems in general from the logs that guide instructions are being sent and acted on. As I noted before when I do a simulated guide by manually hitting NSEW buttons I can see movement on screen (star moves), in client and server panels RA/DEC output moves and this is confirmed in the log. See around 23:32 guide south movement is shown albeit to low precision (but as you say this is just output not calcs which appear to be done on high precision):-
2017-01-16T23:32:29.605 - DEBG - SkySensor2000PC : "RES <-01�50> "
2017-01-16T23:32:29.628 - DEBG - SkySensor2000PC : "VAL [-1.83333] "
2017-01-16T23:32:30.612 - DEBG - SkySensor2000PC : "CMD <#:GR#> "
2017-01-16T23:32:30.686 - DEBG - SkySensor2000PC : "RES <05:40.7> "
2017-01-16T23:32:30.711 - DEBG - SkySensor2000PC : "VAL [5.67833] "
2017-01-16T23:32:30.740 - DEBG - SkySensor2000PC : "CMD <#:GD#> "
2017-01-16T23:32:30.777 - DEBG - SkySensor2000PC : "RES <-01�51> "
2017-01-16T23:32:30.802 - DEBG - SkySensor2000PC : "VAL [-1.85] "
2017-01-16T23:32:31.751 - DEBG - SkySensor2000PC : "CMD <#:GR#> "
2017-01-16T23:32:31.839 - DEBG - SkySensor2000PC : "RES <05:40.7> "
2017-01-16T23:32:31.863 - DEBG - SkySensor2000PC : "VAL [5.67833] "
2017-01-16T23:32:31.882 - DEBG - SkySensor2000PC : "CMD <#:GD#> "
2017-01-16T23:32:31.920 - DEBG - SkySensor2000PC : "RES <-01�51> "
2017-01-16T23:32:31.946 - DEBG - SkySensor2000PC : "VAL [-1.85] "
2017-01-16T23:32:32.921 - DEBG - SkySensor2000PC : "CMD <#:GR#> "
2017-01-16T23:32:33.029 - DEBG - SkySensor2000PC : "RES <05:40.7> "
2017-01-16T23:32:33.055 - DEBG - SkySensor2000PC : "VAL [5.67833] "
2017-01-16T23:32:33.082 - DEBG - SkySensor2000PC : "CMD <#:GD#> "
2017-01-16T23:32:33.107 - DEBG - SkySensor2000PC : "RES <-01�51> "
2017-01-16T23:32:33.128 - DEBG - SkySensor2000PC : "VAL [-1.85] "
2017-01-16T23:32:34.312 - DEBG - SkySensor2000PC : "<HaltMovement> "
2017-01-16T23:32:35.011 - DEBG - SkySensor2000PC : "CMD <:Qs#> "
2017-01-16T23:32:35.035 - DEBG - SkySensor2000PC : "Movement toward South halted. "
SO, assuming for now the guide algorithm is working correctly (can you check?) with my setup - I see about 1.3sec between capture and receive guide frame, the guide exposure being 1 sec. I decided to rule out any lag in the network and moved Ekos/indi onto the Pi/Ubuntu at the observatory and ran in local mode.
I also setup openPHDs on the server and ran that directly - I got calibration and guided for 2-3 mins, >4" errors which is to be expected windy, no PEC, no darks etc. It showed orthogonal axis error so I changed Skeysensor guide mode to X-Y (was RaDec) - was better.
I ran Ekos/indi and got calibrated , again 2 mins then lost guide. I ran out of time and didn't setup logs but it looks more promising.
I'll test the server/local setup correctly tonight, check everything and send the logs. Hopefully this will prove local setup working at least.
Aside from not getting guiding working the whole setup is working very well, impressive.
Andrew.