Hi,
my first board made with KiCad was delivered today from China. I am very happy with it. The only little Disappointment is a little error on the silkscreen. I don’t know, where the error lies. The gerber files seems to be ok. I have a guess. Like Joan_Sparky says: Don’t use Circles on silk screen. Cheap chinese Boardhouse have an issues with it.
The only circle i use, is the small Pin 1 Indicator on the BM280 -> U1.
To all newbies like me -> Don’t use circles 
The drill alignment on that board looks a little off, especially the vias.
I know but i think its not a problem of the gerber files or KiCad. I hope
No, it would be the Chinese fab.
Typically this is because you have an annular ring too little for your PCB Fab…
So the Fab is slight moving the drill to have a bigger annular tolerance…
@Zh4ng, any chance to tell tell us the via settings you used for those?
Might be interesting for others in case they have to deal with similar boardhouses - just to get a figure for when things start to become borderline…
Yes of course. Later in the afternoon (in Germany) i will post the settings.
I didn’t think anyone did this anymore. To be done properly such an adjustment needs to be done on a per hole basis, far too time consuming for most boards these days. It’s far easier to just reject the board and get the customer to fix it.
Anyway, the reason why I think it is simply a misalignment is because the holes are all offset in the same direction. And as you can see near the text “C4” the hole almost separates the pad from the trace. If this is a case of adjusting the annular ring it’s still a very poor job.
sw can do it automatically i.e.
probably it is… this Fab is having also problem with circles in gerber files…
That checks for minimum annular ring width and generates reports, it does not make any adjustments to your drill file.
Anyway, I said the holes are misaligned, and they are! It doesn’t matter if it is just poor alignment or an intentional adjustment for annular ring width.
Not the first one we’ve seen this week.
from the report the system can shift the corresponding drill, so the alignment will result as in the image…
Anyway the point is that it is not because of KiCad, but because of the Fab…
Yes, and if you scroll up about 7 posts you’ll see that is exactly what I said. At no point was the blame directed at KiCad. 
Ok, i fix the annular ring size. Bevor it was 0.6mm with a 0.4mm drill. Now it is 0.8mm … done
annular.py is also crying at some pads but i don’t understand the reason. Where i have to tweak this?
annularpy.py should tell you the coordinates of the pads… Check them and see if there is a too little annular…
on U1 it looks as if you have a cicrle on silkscreen that will end up under the chip.
not good
annular rings:
typical normal vanilla via, drill = 0.25 , pad = 0.45
0.1mm surrounding the drill, thats typically the minimum for fabrication without going to extremes, just plain fabrication.
your shop was not top notch
Ok maui, this is maybe be a stupid question, so don’t slap me. 
Do normal smd pads have an annular ring? annular.py sees a violation of 0.1 on a normal SMD pad at a QFN Footprint. It is not a TH-Pad. I am confused… 
Pad 15 for example…
Then you found a bug 
in my annular script … so don’t slap me. 
I’m going to fix it 
There was nothing wrong with his annular rings to start with, it’s just not a very well made board. The silkscreen issue, that could have been avoided.
