Seems to be an area of great controversy, although I don’t honestly know why. I successfully used an auto-router in the 80’s for a dense military design. The art, and it is an art, is to specify enough rules so that it is possible to get what you want, without ending in a locked no solution mode. Like layout itself, it requires a bit of intuition and skill. It can be done, indeed it was done in the 1980’s by Mentor Graphics, because I was one of the first users of that tool. Can you get a mess, sure. Does it absolve the designer of reviewing the layout, NO! Even if it gets you only 99% of the way there it is worthwhile since you can manually fix the rest. This collapses development time, which means either you, or your group is more productive. Despite all the horror stories contained in the threads posted, my experience with commercial auto-routing has been positive. However, one cannot blindly trust it. The router needs to be constrained with sensible, necessary and sufficient rules. All I know is when I adopted new tools to help me with my designs, I was able to complete the designs in less time. To ignore technology like this just doesn’t make sense. Instead, we should be trying to make this sort of tool more useful and accessible. Personally, I think it would benefit the community.
I get the old school part of this argument. There are some things I’d rather do in traditional ways, if the results are good, and the experience gratifying. But sometimes you have to use modern tools, or you simply won’t be as productive as the person that does use them. So you won’t get the job, because it would take too long, or cost too much. Or you give up, because it will take too long. Quite honestly, in the past, I have given up on creating personal PCB’s because the software was painful to learn. At this point, I do have some additional motivation to learn the tools, so I will go through the process.
As a hobbyist, sometimes I am forced to do things the old way, because the new way is too expensive, or perhaps incomprehensible. But tedious and unenjoyable stuff, I want to automate, and if a good auto-router was around, I’d use it. Because in my case, I used the auto-router, verified the output, and it did a great job, (I checked all critical routes and sampled 50% of the rest of the routes) and it did it in a fraction of the time that the designer could do. So my experience is favorable. 98% of the time one doesn’t need an optimal layout, just one that is good enough. As long as the board performs good enough, and isn’t a mechanical nightmare, I’m happy.
I will figure it out eventually. You have gone through the learning curve, I have yet to. Will get there. My personal journey…
