Autorouting as a first pass layout in PCBnew

hmm…Awhile back I used FreeRoute as a way to get a rough/1st pass at the pcb layout. Is this a good way to do this? I’m interested in your thoughts about starting w/ autorouting. And if you think it is an ok idea, what do you do to autoroute?

I’m having a challenge getting FreeRoute to run. I’m running windows kicad on my mac using Parallels Desktop for Mac (yes, I know there is a Mac version of kicad. I tried it and it was far too unstable for me to continue with it, so I invested in Parallels and all works well…ok, except freeroute :slight_smile: ).

I can’t launch FreeRoute via Java Web Start. When I go to the page, it notes the web version is no longer available…is anyone currently using freeroute?

(Thanks very much for insight)

Personally I never autoroute anything. I seem faster and produce a better layout than the autorouter, especially if you have some critical areas (Switch mode supply or RF stuff). The new Push and Shove router that was recently merged into KiCad makes it much easier to layout by hand and also fix mistakes.

You also get a better feel for what pins can be swapped or even tweak the design to make the layout cleaner. I’m not sure you’d get this with the autorouter. I know it’s daunting laying out a board the first time, but you will be a better designer if you push through the learning curve.

Well thats just my 2c

Hi, Margaret. I believe Freerouting was killed because the creator used to be an employee at Zuken long ago and the company thought he was using their IP in his software. I haven’t checked recently to see if the situation has changed, so the rest of my commentary may be moot.

I like to use an autorouter. For some designs (low speed, low density) it just makes sense, time-wise. For more challenging designs, setting-up the router control parameters takes some time. However, you’ll learn what works and what doesn’t and that knowledge carries over to your future designs. The problem with manual routing is that you’re always starting each design from scratch.

An autorouter can also help with placement. You can test various component placements and see which has the best routing completion, fewest vias, shortest traces, etc.

Thanks very much. Ah yes - now that you mention it, I recall hearing something about the IP issue. A shame this feature is embedded into kicad. As you note, I like the idea of an autorouter as a learning tool. In that sense, perhaps autorouters miss their mark. If I had lots of experience, my need for feedback from an autorouter would be minimal. If I am on a layout learning curve, autorouting replaces a TA in the “here’s one approach.” I see this as true for the majority of cases with LTSpice. It is amazing to learn from. If I had 20-30 years, I have the “intuition.”

Thank you VERY much for taking your time to share your thoughts.

Thanks Stew. Ah yes, you are far more advanced than me! I also agree it is much, much better to do without the autorouter as a crutch. The challenge I have is with the learning curve. I’m always looking for something that avoids me having to not waste a person such as yourself time with n00b layout questions and such. In the days of the dinosaurs (i.e.: when I was in software) we called it RTFM (Read The F**KING MANUAL) Now it is “hey sport, google is your friend”)…

And indeed I am pushing through the learning curve. I appreciate your reply.

Hi, Margaret. I tend to use the autorouter as an idea tester: what happens to the routing if I arrange things like this? How does the routing change if I add more layers? Etc…

In the sense of using it for learning, then I think autorouters aren’t a good idea. The quality of their results is very dependent upon how you setup their parameters. If you don’t have much experience, then that’s difficult to do and you end up learning from a bunch of poor examples.

With layout, you’ll go through the same three learning stages that you did with writing programs:

  1. Does it compile? Or, in the case of PCBs, does it pass the design rule checker? This involves knowing the syntax of a PCB: i.e., how close can this thing be to that thing?
  2. Does it meet the specs? Or, in the case of PCBs, is it manufacturable, economical and realizes the necessary performance? This is where the majority of your learning will occur. Books by Howard Johnson and Eric Bogatin cover the principles of building high-speed circuitry. There are handbooks on principles of PCB design for manufacturability. The backs of datasheets often provide advice and layout tips for specific types of chips like ADCs and switching regulators.
  3. Is it elegant? Most of us never reach this point, but we know it when we see it.

With programming, the best way to learn is to read other programs. With PCBs, you need to start looking at other people’s layouts. Luckily, this is easy to do now: google for images of “PCB layout”. Try to see if some layouts look better than others, sort of like wandering through an art museum. Once you see a lot of examples of how PCBs are arranged, then you can start to backfill your intuition with information from texts, datasheets and blogs.

I wish there were an automated tutor for this, but I don’t know of one. However, if you record the steps you take on the way to mastery, others could benefit from the path you’ve blazed.