Thanks for the comment. Autorouting is definitely a hard target.
We try to be fairly explicit about this in our case studies and show the amount of manual rework required after automated placement and routing.
For smaller embedded boards, our current expectation is that routing cleanup is often less than an hour unless there is a significant gap in the first pass. We also provide the DRC results, applied constraints, and a routing report highlighting areas that may need engineer review.
There is also a subjective part to PCB layout. A board can be electrically correct and pass DRC, but an engineer may still dislike how a particular trace was routed. For example, some engineers prefer arcs for differential pairs while others prefer 45-degree bends. So we don't consider "100% routed" to mean "100% finished."
For larger boards, we generally avoid doing everything in one long run. We prefer breaking the work into shorter subtasks, reviewing the output with the user, and then running the next iteration. This helps avoid spending several hours on a run only to discover that the result has drifted significantly from how the engineer wanted the board laid out.
The generated boards still need engineering review and some cleanup, and we try to make that clear rather than claiming otherwise. The same applies to things like silkscreen generation — some silkscreen may still need repositioning after the automated pass.
Our online app autocuro.com/app can generate routing for smaller boards as well, so you can generate a few designs and judge the results for yourself.
automating pcb routing is a pretty hard target. how much cleanup is still manual after a first pass?
Thanks for the comment. Autorouting is definitely a hard target. We try to be fairly explicit about this in our case studies and show the amount of manual rework required after automated placement and routing. For smaller embedded boards, our current expectation is that routing cleanup is often less than an hour unless there is a significant gap in the first pass. We also provide the DRC results, applied constraints, and a routing report highlighting areas that may need engineer review. There is also a subjective part to PCB layout. A board can be electrically correct and pass DRC, but an engineer may still dislike how a particular trace was routed. For example, some engineers prefer arcs for differential pairs while others prefer 45-degree bends. So we don't consider "100% routed" to mean "100% finished." For larger boards, we generally avoid doing everything in one long run. We prefer breaking the work into shorter subtasks, reviewing the output with the user, and then running the next iteration. This helps avoid spending several hours on a run only to discover that the result has drifted significantly from how the engineer wanted the board laid out. The generated boards still need engineering review and some cleanup, and we try to make that clear rather than claiming otherwise. The same applies to things like silkscreen generation — some silkscreen may still need repositioning after the automated pass. Our online app autocuro.com/app can generate routing for smaller boards as well, so you can generate a few designs and judge the results for yourself.