Some small update:

1) Yes, when you check "Lock", the filter DOES change for the focus. sorry, I have said nothing. :blush: :blush:

2) I had a hard crash of Ekos. This happened when I clicked the "capture" button in the capture module. Don't know what happened here. Here is extract of backtrace and log.

GDB backtrace

Thread 1 "kstars" received signal SIGSEGV, Segmentation fault.
0x0823ccc5 in FITSData::findCentroid<unsigned short> (this=0xb8540b0, boundary=..., initStdDev=<optimized out>, 
    minEdgeWidth=5) at ./kstars/fitsviewer/fitsdata.cpp:1298
1298    ./kstars/fitsviewer/fitsdata.cpp: No such file or directory.
(gdb) bt
#0  0x0823ccc5 in FITSData::findCentroid<unsigned short> (this=0xb8540b0, boundary=..., initStdDev=<optimized out>, 
    minEdgeWidth=5) at ./kstars/fitsviewer/fitsdata.cpp:1298
#1  0x082441f8 in FITSData::findCentroid (this=<optimized out>, boundary=..., initStdDev=<optimized out>, 
    minEdgeWidth=<optimized out>) at ./kstars/fitsviewer/fitsdata.cpp:1168
#2  0x0824431f in FITSData::findStars (this=0xb8540b0, boundary=..., force=false) at ./kstars/fitsviewer/fitsdata.cpp:2046
#3  0x0825260c in FITSView::findStars (this=0xad66378, algorithm=ALGORITHM_CENTROID) at ./kstars/fitsviewer/fitsview.cpp:1197
#4  0x082fad89 in Ekos::Focus::setCaptureComplete (this=0xad3a388) at ./kstars/ekos/focus/focus.cpp:1144
#5  0x081c016b in Ekos::Focus::qt_static_metacall (_o=0xad3a388, _c=QMetaObject::InvokeMetaMethod, _id=44, _a=0xbffaef54)
    at ./obj-i686-linux-gnu/kstars/moc_focus.cpp:373
#6  0xb568d66f in QMetaObject::activate(QObject*, int, int, void**) () from /usr/lib/i386-linux-gnu/sse2/
#7  0xb568dbad in QMetaObject::activate(QObject*, QMetaObject const*, int, void**) ()
   from /usr/lib/i386-linux-gnu/sse2/
#8  0x081a429a in Ekos::DarkLibrary::darkFrameCompleted (this=0xb505db8, _t1=<optimized out>)
    at ./obj-i686-linux-gnu/kstars/moc_darklibrary.cpp:148
#9  0x082995ab in Ekos::DarkLibrary::subtract<unsigned char> (offsetY=<optimized out>, offsetX=<optimized out>, 
    filter=<optimized out>, lightImage=<optimized out>, darkData=<optimized out>, this=<optimized out>)
    at ./kstars/ekos/auxiliary/darklibrary.cpp:258
#10 Ekos::DarkLibrary::subtract (this=0xb505db8, darkData=0xbdc33f0, lightImage=0xad66378, filter=FITS_NONE, offsetX=0, 
    offsetY=0) at ./kstars/ekos/auxiliary/darklibrary.cpp:186
#11 0x0829d1f6 in Ekos::DarkLibrary::newFITS (this=0xb505db8, bp=0xa9e99c98) at ./kstars/ekos/auxiliary/darklibrary.cpp:341
#12 0x081ae056 in Ekos::DarkLibrary::qt_static_metacall (_o=0xb505db8, _c=QMetaObject::InvokeMetaMethod, _id=2, 
    _a=0xbffaf144) at ./obj-i686-linux-gnu/kstars/moc_darklibrary.cpp:85
#13 0xb568d66f in QMetaObject::activate(QObject*, int, int, void**) () from /usr/lib/i386-linux-gnu/sse2/
#14 0xb568dbad in QMetaObject::activate(QObject*, QMetaObject const*, int, void**) ()
   from /usr/lib/i386-linux-gnu/sse2/
#15 0x081a6dca in ISD::GDInterface::BLOBUpdated (this=0xb32c1c8, _t1=<optimized out>)
    at ./obj-i686-linux-gnu/kstars/moc_indistd.cpp:383
#16 0x0820011f in ISD::CCD::processBLOB (this=<optimized out>, bp=<optimized out>) at ./kstars/indi/indiccd.cpp:1703
#17 0x081f4b45 in INDIListener::processBLOB (this=0xa88b3f0, bp=0xa9e99c98) at ./kstars/indi/indilistener.cpp:374
#18 0x081c082e in INDIListener::qt_static_metacall (_o=0xa88b3f0, _c=QMetaObject::InvokeMetaMethod, _id=18, _a=0xa7cb9588)
    at ./obj-i686-linux-gnu/kstars/moc_indilistener.cpp:182
#19 0xb568a3f0 in QMetaCallEvent::placeMetaCall(QObject*) () from /usr/lib/i386-linux-gnu/sse2/
#20 0xb568e183 in QObject::event(QEvent*) () from /usr/lib/i386-linux-gnu/sse2/
#21 0xb613c04a in QApplicationPrivate::notify_helper(QObject*, QEvent*) () from /usr/lib/i386-linux-gnu/
#22 0xb614184c in QApplication::notify(QObject*, QEvent*) () from /usr/lib/i386-linux-gnu/
#23 0xb565d4ad in QCoreApplication::notifyInternal2(QObject*, QEvent*) () from /usr/lib/i386-linux-gnu/sse2/
#24 0xb566036e in QCoreApplicationPrivate::sendPostedEvents(QObject*, int, QThreadData*) ()
   from /usr/lib/i386-linux-gnu/sse2/
#25 0xb5660877 in QCoreApplication::sendPostedEvents(QObject*, int) () from /usr/lib/i386-linux-gnu/sse2/
#26 0xb56b7e03 in ?? () from /usr/lib/i386-linux-gnu/sse2/
#27 0xb42ee4d9 in g_main_context_dispatch () from /lib/i386-linux-gnu/
#28 0xb42ee779 in ?? () from /lib/i386-linux-gnu/
---Type <return> to continue, or q <return> to quit---
#29 0xb42ee844 in g_main_context_iteration () from /lib/i386-linux-gnu/
#30 0xb56b81f3 in QEventDispatcherGlib::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) ()
   from /usr/lib/i386-linux-gnu/sse2/
#31 0xb188a521 in ?? () from /usr/lib/i386-linux-gnu/
#32 0xb565adad in QEventLoop::exec(QFlags<QEventLoop::ProcessEventsFlag>) ()
   from /usr/lib/i386-linux-gnu/sse2/
#33 0xb566407f in QCoreApplication::exec() () from /usr/lib/i386-linux-gnu/sse2/
#34 0xb5b417b1 in QGuiApplication::exec() () from /usr/lib/i386-linux-gnu/sse2/
#35 0xb6138cc4 in QApplication::exec() () from /usr/lib/i386-linux-gnu/
#36 0x08094a60 in main (argc=<optimized out>, argv=<optimized out>) at ./kstars/main.cpp:313

==> log_09-36-14.txt <==
2017-04-11T09:36:26.416 - WARN - "Object named NGC 78 not found"
2017-04-11T09:36:26.488 - WARN - "Object named NGC 6050A not found"
2017-04-11T09:36:27.878 - WARN - "Object named NGC 4666 not found"
2017-04-11T09:36:28.526 - WARN - "Object named NGC 7318 not found"
2017-04-11T09:36:28.816 - WARN - "Object named NGC 7839 not found"
2017-04-11T09:36:29.013 - WARN - "Object named NGC 7840 not found"
2017-04-11T09:36:29.249 - WARN - "Object named IC 1127 not found"
2017-04-11T09:36:29.542 - WARN - "Object named IC 2497 not found"
2017-04-11T09:36:29.775 - WARN - "Object named IC 4703 not found"
2017-04-11T09:36:29.992 - WARN - "Object named JupiterComet Shoemaker–Levy 9 Wikipedia page not found"

==> log_21-38-29.txt <==
2017-04-11T21:45:45.543 - DEBG - SBIG CCD :  "Taking 1s exposure on main camera... "
2017-04-11T21:45:50.305 - DEBG - Focus:  "Capturing dark frame..."
2017-04-11T21:45:50.326 - DEBG - SBIG CCD :  "Taking 2s exposure on main camera... "
2017-04-11T21:45:55.213 - DEBG - fits MIN:  927  - fits MAX:  32639  - pixel range:  31712  - bin width  50.9021  bin count  624
2017-04-11T21:45:55.302 - DEBG - FITHistogram: JMIndex  0.999997
2017-04-11T21:45:55.317 - DEBG - Focus:  "Dark frame received."
2017-04-11T21:45:55.338 - DEBG - Focus:  "Dark frame saved to /home/cboy/.local/share/kstars/darks/darkframe_2017-04-11T21-45-55.fits"
2017-04-11T21:45:55.470 - DEBG - Focus:  "Image received."
2017-04-11T21:45:55.473 - DEBG - SNR:  31.3477
2017-04-11T21:45:55.473 - DEBG - The threshold level is  1031.21 (actual  1031.21 )  minimum edge width 5  minimum edge limit  3

3) It seems the robofocus is still an issue. Control box is version 3.1. It seems the correct focus jumps up and down around 200 ticks for the same filter, most of the time ending in not focused stars. (I use "gradient", and "polynomial", 0.40% tolerance, 30 ticks step size). Filters are +-85 ticks parfocal. But I mainly observe with my present setup with the V-filter, so I can probably find a workaround for that. I think the "gradient" option is best if you have an optical system that produces a dohnut because of the central obstruction?

Another focuser is not really an option, it's a) expensive and b), robofocus is connected directly to the axis of the built-in focuser of the vmc200l. Like this:
I did have a moonlite years ago. I wasn't impressed by it because it had light leaks and focus shifted when a heavy sbig-camera was attached. I'm not exactly impressed with the robofocus either, but for now, I 'm pretty much stuck with it.

I can borrow another robofocus control-box from a friend soon, I'll test with that one. When I enable the driver info verbose of the robofocus, it doesn't show any errors.