I’m currently using KiCad V5 and I’m wondering if autorouting was possible in this version. I’ve heard that it was removed in V6, but I’m not sure if that’s true.
If anyone knows the answer, please let me know. I’m also interested in any tips or tricks for using autorouting in KiCad.
Thanks!
I think the last version that had a kind of builtin autorouter was v4. After that you’d use the DSN-SES round trip to interface to an external autorouter.
KiCad V7 can still export and import “Specctra” files as an interface to an external autorouter.
As far as I know, KiCad has never had a built in autorouter. Older KiCad versions (V4 or V5?) used to have a more direct interface to directly interact with an external autorouter (notably freerouting, but these days you have to work with those Specctra files as intermediate step.
An autorouter is not a high priority for the “KiCad team”, for several resons. First, they are not the magical tool beginners often think they are. Second, to make an autorouter that is good enough to be actually useful takes many years of software development from several people, and KiCad does not have the resources (yet) for that.
There have been some long threads about autorouters and their usefulness in the past.
https://forum.kicad.info/search?q=autorouter
From what I remember, the link below is the most comprehensive discussion:
I started with a V4 build and there was no autorouter built in. Over the years KiCad has improved its interface to the 3rd party Freerouting tool.
Meanwhile KiCad has improved the built in one at a time track completion
Ah. Fortunately I never used v4. I’m too young, or old?, to remember it. 
Freerouter isn’t the only game in town though. There are other routers mostly (all?) paid for, like topological router which claims to handle difficult jobs. The DSN-SES interface also seems to exist in other ECAD suites.
A Freerouting fork has had some attention, making it work with recent Java and fixing several bugs. This is the version used in the KiCad plugin.
Some of the commercial routers are better, but far too expensive for the fresh hobbyists who seem to be the main autorouting requesters
Hi
I used to believe I wanted an autorouter. But as a general sweeping statement - you don’t really need one and you get better result doing it manually anyway.
There are a lot of good practices and rules of thumb out there which most autorouters just don’t have. Check out ‘Phils Lab’ and 'Rick Hartley’b(I think on the Altium YT site) to name but a few.
Is it possible to utilize the autorouting feature in KiCad v5 while considering third-party interference?
Hi @rayan
A good autorouter can save many hours of work.
A good autorouter also requires many hours to set up correctly to get good results.
Setting up requires the user to know how the autorouter works and its limitations.
Understanding the autorouter and its limitations requires the user to be fully conversant, and totally competent, with manual routing and parts placement.
I think something like a PC motherboard with fast and wide buses requiring length matching makes it worth using a good autorouter. The setup for this will take several days. This is a very small part of the PCB layout market, so a clue why there are so few vendors left
I never attempted to use auto routing in KiCad.
I tried autorouting a few times some 30 years ago, and I found the result so horrible that manually fixing the PCB afterward took more time then routing it manually in the first place. In contrast to that, I love the interactive router in KiCad which is a great help to make PCB’s denser then I’d ever dare to attempt to make with an autorouter. Making PCB’s small and compact makes them extremely difficult to route. At some density there is just not enough room to lay all the tracks, regardless of the tools you use. For a beginner I recommend to stay well away from that.
Did you read the (long) tread I linked to earlier?
It has a lot of the pro’s and con’s of autorouters. Maybe you don’t like the length of the thread, but autorouters are a quite involved topic.
My short opinion is that beginners in PCB design should not use an autorouter at all, and first learn to do manual PCB routing properly. After you have learned that you have a lot of the knowledge that is needed to set up an autorouter in the first place.
Why stay with old technology. Upgade to kicad 7
I use the freerouting tool, now with KiCAD 7.
But the trick is HOW you use an autorouter.
I view it as only a helpful guide for subsequent manual routing.
I first manually route the power bus layout, then I invoke freerouting a few times to give me a head start on a reasonable layout that gets all connections routed.
BUT, I then manually re-route most (all?) of the traces to position them where I actually want them.
So, autorouting simply takes away the initial “where to start” headache, and also circumvents a lot of “redo” changes of otherwise completely manual routing, that I’d otherwise have to do to manually determine an efficient trace layout.
But everyone has their own view on this.
Hope this perspective helps?
Using one of the best autorouters (Mentor Expedition) for almost 20 years, I do not miss this feature in Kicad. As usual I started the design with push and shoove manual routing. Therefore I ignored any rules what I set for the autorouter. This finishes the design well over 90%. There are a few dozen remaining connections what go anywhere from A to B what are located at opposite end of the board. In this state its usefull to give the few reamaining connections to the autorouter what will finishes the design much faster as you can do.
What I learned from:
If the autorouter fails with any signal, its most time not too hard to do it by hand.
Sometime the autorouter has really good ideas
Frequently I did the following game with the autorouter: Design a board by hand and in late state with autorouter until finished. Mainly the push and shove rules on board edge should be diffrent from default rules. Then make a copy of the design, erase any copper but do not change the placements. If autorouter starts from scratch, he typically fails at around 60-70%. The result is useless, as there is no chance to go on by hand.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.