I’m making a high-power RF inductor and want to curve the tracks to reduce radiation.
I know there have been many past discussions on this, but as KiCad is ever improving, I thought I would bring it up again.
Currently there is an ok workaround:
Draw the track with the normal routing tool
Delete the segment you want to replace with an arc
Draw the arc in its place
I’m using KiCad 7.0.1, which allows me to draw arcs on the copper planes (Yay!). Older versions might not allow that.
However, the DRC doesn’t think this is so good. It fails to recognize that the arc is connected to the trace, and it doesn’t allow me to route over the arc to fake it.
I did find an old issue on the GitLab that shows routing with arcs has unresolved DRC issues, which is probably why it’s not in the release build yet.
Once you draw the arc, couldn’t you just place a route over that made up of small segments?
Side question: Is there any validity to the idea that curved traces reduce RF emissions? I mean, has anybody ever made identical designs, one with curved traces and one without, and measured the emissions? There’s a similar issue with curved traces for high-speed designs and that’s pretty much been shown to be unnecessary. A chamfer gives you most of what you need.
I suggest you start with updating to KiCad V7.0.7. Increments in the third digit are only bug fixes, without adding new functionality, and there have been quite a lot of bug fixes in the last 6 months. See the Release Notes | KiCad EDA for more info.
Are you aware there are several (at least 5) scripts for designing PCB Inductor spirals? There is even a plugin for designing the inductors of a BLDC motor on a PCB.
But unfortunately, support for arcs in tracks is going very slowly in KiCad. At the moment it still does not support independent arcs, but only fillets, and this has several annoying limitations. You can’t drag a single endpoint, or change it’s radius without changing both endpoints. A fillet also does not support a 180 degree arc.
What you can do is draw graphical arcs, modify their properties (duplicate, drag end pints, copy paste, etc) and then select them, right click and select: Create from selection / Create tracks from Selection
I never done it myself, but from what I have read it starts having an influence if your signals go beyond about 3GHz. You also won’t save much space with arcs compared with 45 degree chamfers, and considering the difficulty KiCad still has with arcs, I doubt if it’s worth it to use them (except for scripted approaches) at the moment.
You also may benefit from KiCad-Nightly V7.99 Apparently it can assign net names to graphical items, so you can use “real arcs” instead of having to convert them to fillets first.
for HF boards there is an rf tool plug-in at high frequencies, the material of the board is of great importance due to the wave impedance and fr4 is not suitable there already … in general, the HF part is a single direction of electronics with its own magic and nuances … complex, confusing with non-standard approaches. …
In v7.0.7, the Arc and Fillet tools works well enough to accomplish various Arc-making goals. Each aspect of Arc making is somewhat different in each Arc tool…
Video shows common usage but, I haven’t explored every possible tweak and pref setting…
(And, I could have done a better job with the video but, it’s good enough to demo…)
The DRC errors show only unconnected tracks and no board shape (I didn’t connect anything nor draw the shape). No actual error with the Arc’s.
Cool!
I didn’t know there was a “make filet” in the right-click menu. This works fine for my purposes.
Also very neat that you can edit it with the drag tool.
As for whether the curve is strictly necessary, who knows? It does give me more clearance as I come out of the coaxial connector footprint.
Along with the filet tool, which is limited IMO, the 3rd party plug-in Round Tracks works amazingly well. I highly recommend adding it. It’s found in the Plugin and Content Manager.
I do power and not RF. But I will bet that a sharp angle in a track represents a discontinuity which will cause some reflections; much more so than a nicely radiused curve. But if we assume that you are running RF along such a track and you get perceptible reflections from a 90 degree angle (or from two 45’s) that if you were to implement a successive series of nine 10 degree bends that might work pretty well. How about 15 six degree bends?