I’m creating a footprint for a ti H-Bridge using a VQFN-HR (16) package, setting out the solder resist mask and the solder paste layers are causing me some issues . . .
https://www.ti.com/lit/pdf/mpqf631b
Where I 'm having issues is with the radiused corners of the solder resist mask apertures, for example Pad 4 . . .
I have the pad and a square cornered mask but this is not technically correct . . .
So I converted the polygon to lines then added a correct 0.05mm rad to the relevant corners and then converted the shape back to a polygon . . . and got this . . .
What were rads are now a mix of rads and chamfers, it’s not very pretty.
Is this a known issue ? am I doing something wrong ? is there a workaround ? this is on 8.0.1 latest release build for Windows.
What I have though about doing is making this feature from 3 pads, one central rectangular one with square corners and two smaller ones, one at each end with radiused corners.
As far as I know there is no open gitlab issue regarding this behaviour on gitlab. It looks like a arc approximation artefact. Remember the used radius: 0,05mm
It’s not really possible to produce such a small radius as a “real” shape - the produced result will always differ from a ideal 0,05mm sized arc.
If you repeat your shape modification / shape creation commands with a 0,25mm arc radius you will get better looking results (albeit still not perfect).
making this feature from 3 pads, one central rectangular one with square corners and two smaller ones, one at each end with radiused corners
This approach will work. You could even work with only two pads (one long from left<–>right, one from top<–>bottom) overlapping each other. This allows easy pad placement - both share the same pad coordinate point.
I get nice rads on pads when specified as Rounded Pads . . .
I’m making progress regardless . . .
Good point 
I get nice rads on pads when specified as Rounded Pads
yes, pads seem to have nearly perfect round edges, even with very high zoom factors.
I think there are two different code paths internally, one for pads and one for polygon/arc calculation (with much less approximation accuracy).
As always: if this annoys you enough → gitlab 
1 Like
It took me some minutes you likely mean radii when you write rads. I already guessed it was not radio active and suggestions of Reactive Attachment Disorder or Reactive Airways Dysfunction Syndrome also seemed implausible. Using unusual abbreviations on a public forum is not a very good practice. It also degrades search results on this forum.
There are already a bunch of related open issues on gitlab related to polygon geometry.
If I were to make this footprint, I would use multiple overlapping pads with the same pad number. This would have no issues with rounded corners.
Yup, this is what I have done . . . it’s close enough for now.
I’ll see if any of those existing issues fit with what I’m seeing, if not I’ll probably open an issue.
Sorry, it’s a standard engineering term for radius here in the UK so didn’t give it much thought, but yes I do agree with your point.
0.05mm = 0.002 inch. You won’t find any PCB vendor that can do that.
An average Human ‘Hair’ is 0.003 inches thick.
An average Human ‘Hair’ is 0.003 inches thick
but it is round at least.
2 Likes
Basically, confirming @mf_ibfeew …
I have no problem drawing 0.05mm Radius - it shows clean radius until it’s placed over Cu, then it (the Aperture) shows Straight (Chamfered) line. A graphics thing and not necessarily a PCB thing…
EDIT: Correction - the Straight-Line happens when converting from Polyline to Filled Polygon. That said, after playing around, I discovered that Setting the Line-Width (on a corner that does not have a Radius Corner) will present a Radius’d corner based on the Line’s width… Last TWO Screenshots, below… Thus, if making a Rectangle, user can create a Radius’d Corner by tweaking the Line Width and Filling the Rectangle will keep that desired result. Last Screenshot has 0.1mm line-width on an Aperture with 0.52mm Width
EDIT - UPDATED / CLEAN-UP: What is the Smallest possible Aperture in a Mask?
I was able to dial-it-down to very small, too small to see the Radius’d corners in the 3D-Viewer so, I increased the size to be able to see them.
Line-Width = 0.01mm. Last image shows Radius’s corners. Aperture Width=0.035mm
