You could draw them as polygons on the copper layer, but that sounds pretty painful. If I were doing this, I’d just use free-angle routing and very small segments to approximate the curves. (well, I’d actually use a nightly build, but…)
1 Like
Kind of a hack for this but you could draw your needs in Inkscape and use this plugin to translate it to edge cuts and copper layers. It’s really meant for making PCB badges and artwork. The Freecad plugin would probably work as well.
1 Like
svg2shenzhen is great ( and I prefer it for artistic stuff and board outlines…) but if you’re working with something that correlates to a schematic @Jarret 's stretch plugin might be the better way to go https://github.com/JarrettR/Stretch .
I don’t usually comment, but your insistence that “only 4 people use blah”, and “only two people use diggidy dap” is completely ridiculous reasoning for why we should not have features other programs have. That is, unless you have done a full market survey of the user base, a market analysis of what features people will want over the next 5 years, and analyzed the cost of development of these features vs others in demand - which you haven’t…
I welcome these new features, and trust the developers of this program to continue the rapid pace of development and growth of feature sets.
That’s all… have a nice day
3 Likes
For very small PCB’s curved traces also allow for a much tighter packing of traces, you reclaim most of the area consumed by those 45 / 90 degree corners, currently been using free hand routing as an interim, but being able to specify a common center point will make things much easier.
1 Like
3 Likes
Dont be too to hard on hobbyists - I like KiCad at home very much 
But you are right, as a hobbyist I could stay with V5.x nevertheless KiCad must proceed to commercial users…
I am wondering if curved tracks could be dragged later on.
sad but true… I hope it gets supported in future.
There some advances, but for v6.99, I think.
Roberto’s MR will get in for 6.0 I think, but it does not allow dragging a track with arcs, just dragging a single arc in a track to resize it (still useful, but not the same as requested in 6544)
Man… I love KiCAD … Have I mentioned that lately? … nah, I don’t think I have… I’ve been too busy ranting (whining? :‒) ) about oooh, I don’t like this, ooooh, I don’t like that (net classes, etc. in my most recent post/thread)! :‒)
I am sooo looking forward to V6, and I want to applaud the effort of developers and all the great features KiCAD already has and the ones I’ve seen that you’re planning to add in V6… You have no idea how much I wish I could find time available to work towards joining the team of developers!! Oh well… maybe one day…
1 Like
Yes, the title of the MR is confusing. It should be called “interactive resizing of arc tracks” instead. It just gets triggered when you do the ‘drag’ operation (both 45 degree and free angle).
I am still fixing a problem with the MR as is (if you move the mouse to the wrong place it creates an invalid arc). Hopefully will have it ready for merging soon.
1 Like
‘This RF superstition was debunked years ago’ That’s a very sweeping and erroneous statement in my opinion. The problem of discontinuities and reflections caused by sharp corners is very real one. The cited paper has clearly been written by an ‘EMC expert’ not a professional rf/microwave engineer. Any antenna engineer will tell you that the absolute accuracy of measurements taken in an anechoic chamber is around +/- 1.5dB. All measurements are < 1GHz, yet the TDR plots clearly show a measureable effect. What this paper misses in spectacular fashion is the cumulative effect of cascaded mismatches, so it is important to minimize them. The effect is at its worse when reflections add in phase. There is no full two port s-parameter measurement given in the frequency domain and not a single reference to any field theory. The geometry of the test pieces is not defined - so it is impossible to verify these results. In short this paper is not particularly scientifically rigourous and should be treated with caution.
2 Likes
Rounded corners also help with high currency applications: https://blog.samtec.com/wp-content/uploads/2020/06/05_14_2020_geek_speek_perils_RA_turns.pdf
1 Like
I hadn’t thought about the mechanical aspects - but clearly a 90 degree corner can act to concentrate stresses. A fact sadly demonstrated by the DeHavilland comet, whose square windows had disasterous consequences!
It is not just flexis where sharp corers must be avoided. I have seen regular PCB tracks break where a track joins the pad of a connector that has flexing stresses. Bolted on terminals are prone to this
One of the things that teardrops solve. (The other is a bigger tolerance for drill location (translates to smaller possible via’s).
A bit weird that this issue for teardrops is still open after 12 years (2008-01-27).
On the plus side, It’s got a milestone for KiCad V7 and several scripts for teardrops already exist. I’m guessing that one of the reasons it took so long was the absence of arcs in tracks, which is now coming in V5.99. 
1 Like
The preferred ways to implement teardrops depends on arcs (in tracks and/or polygons, depending on if we implement teardrops as a collection of tracks or as a single polygon) and also groups: if we implement teardrops as tracks instead of a polygon, we want to group those tracks together so that they can be easily deleted and re-generated.
1 Like