Re: [PEDA] OT: DIY: autorouter
Bevan: Not to discourage you, but I think some of the best minds in the industry have been working on autorouters for 20 years and we have 'what we have.' If you can single-handedly come out with an autorouter that Phooey on that. Bevan, I encourage you in your autorouter development quest. PCB design has changed a lot in the last 20 years. Maybe some of the traditional algorithms don't work so well as they once did. Maybe you can discover something new. Maybe a simple algorithm will work better. Given the huge speed increases in modern PCs, maybe it's like a chess problem - you can consider more moves (routes) ahead if you have more computing power. Lots of tech advances have occurred because someone said I think I can do it better than these folks who have been doing it for 20 years... Best regards, Ivan Baggett Bagotronix Inc. website: www.bagotronix.com - Original Message - From: Tony Karavidas [EMAIL PROTECTED] To: 'Protel EDA Forum' [EMAIL PROTECTED] Sent: Tuesday, October 15, 2002 6:41 PM Subject: Re: [PEDA] OT: DIY: autorouter Not to discourage you, but I think some of the best minds in the industry have been working on autorouters for 20 years and we have 'what we have.' If you can single-handedly come out with an autorouter that performs better than what's out there, you will certainly become famous / rich / etc. Someone will scoop your code up and integrate it, but for some funny reason, I just don't see that occuring. -Original Message- From: Bevan Weiss [mailto:[EMAIL PROTECTED]] Sent: Tuesday, October 15, 2002 4:35 AM To: Protel EDA Forum Subject: [PEDA] OT: DIY: autorouter Hi guys, I've been doing some research on the kinds of things to improve the autorouter (as per the desire to create my own). I've come up with some stuff that I'd like to bounce off ya's. Using a path-finding algorithm which assigns weights to various directions to travel (ie assuming that only 45deg angles are allowed, does anyone have a reason this isn't valid??) The default weighting would be to head towards the target pin (closest of the set), however if a large obstacle (ie dense gathering of wires) exists in the default path, then the algorithm would start to look at ways around the blockage, ie either using a via in which case you just perform the same operations on a different layer, or by going around the blockage. Weightings would be assigned to either of these (based on a static disadvantage for a via), if the side-track distance exceeds the disadvantage for the via however, then the via would be generated. The algorithm would take a single step forward (ie a single node point) for each connection pair, and then loop around and do the connection pairs again. I imagine that this would allow for better compromises to be made however other opinions on this are welcome. Just an update, Thanks for your time, Bevan Weiss * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * To post a message: mailto:[EMAIL PROTECTED] * * To leave this list visit: * http://www.techservinc.com/protelusers/leave.html * * Contact the list manager: * mailto:[EMAIL PROTECTED] * * Forum Guidelines Rules: * http://www.techservinc.com/protelusers/forumrules.html * * Browse or Search previous postings: * http://www.mail-archive.com/proteledaforum@techservinc.com * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
Re: [PEDA] OT: DIY: autorouter
It might be interesting to consider the board as an array of spots, and route each track separately, noting which spots get used the most. Then use that data to cause the longer tracks through those points to flee those points. They are already long, making them a bit longer usually won't hurt. When I manually route, I try to get all the critical stuff in first, then the short stuff, then progressively longer tracks, looking for groups of signals that can be routed together. One thing that's hard to do in autorouters, is to consider return current. Current project: I have this printhead, that will take pulses of up to 10A, and dump them into ground. Obviously I would prefer that current route back to the source cap on a dedicated track, but every AR I've ever seen would cheerfully dump it into the plane, and let it meander the entire board before getting home. In order to handle this, you'd have to have a way to flag a net as high current, and to tell the system where it's source is. Stuff like this, is why I almost never autoroute. By the time I could get done setting up all the conditionals and fencing in the AR, I could have routed it myself. * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * To post a message: mailto:[EMAIL PROTECTED] * * To leave this list visit: * http://www.techservinc.com/protelusers/leave.html * * Contact the list manager: * mailto:[EMAIL PROTECTED] * * Forum Guidelines Rules: * http://www.techservinc.com/protelusers/forumrules.html * * Browse or Search previous postings: * http://www.mail-archive.com/proteledaforum@techservinc.com * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
Re: [PEDA] OT: DIY: autorouter
Thanks, I had completely forgotten about those situations where a ground plane isn't being used... There are several ways that I can think of to do this, if you give the router info on the maximum size of the current through the connection, and the maximum voltage drop across the connection (implied as 0.1V otherwise) then it will attempt to route the connection such that it always remains on the low side of the threshold, ie the track will be X ohms in resistance over its length and the track throughout the route would be capable of handling the input current. This means that other tracks would be routed in perhaps a slightly longer route depending on the priority of that particular track with reference to the large power bus... This would require setting up priorities for different connections however, something which can be a little time consuming... Thanks for the feedback guys, I'll have a think about all these problems and will eventually get something worked out about them... - Original Message - From: David VanHorn [EMAIL PROTECTED] To: Protel EDA Forum [EMAIL PROTECTED] Sent: Thursday, October 17, 2002 5:31 AM Subject: Re: [PEDA] OT: DIY: autorouter It might be interesting to consider the board as an array of spots, and route each track separately, noting which spots get used the most. Then use that data to cause the longer tracks through those points to flee those points. They are already long, making them a bit longer usually won't hurt. When I manually route, I try to get all the critical stuff in first, then the short stuff, then progressively longer tracks, looking for groups of signals that can be routed together. One thing that's hard to do in autorouters, is to consider return current. Current project: I have this printhead, that will take pulses of up to 10A, and dump them into ground. Obviously I would prefer that current route back to the source cap on a dedicated track, but every AR I've ever seen would cheerfully dump it into the plane, and let it meander the entire board before getting home. In order to handle this, you'd have to have a way to flag a net as high current, and to tell the system where it's source is. Stuff like this, is why I almost never autoroute. By the time I could get done setting up all the conditionals and fencing in the AR, I could have routed it myself. * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * To post a message: mailto:[EMAIL PROTECTED] * * To leave this list visit: * http://www.techservinc.com/protelusers/leave.html * * Contact the list manager: * mailto:[EMAIL PROTECTED] * * Forum Guidelines Rules: * http://www.techservinc.com/protelusers/forumrules.html * * Browse or Search previous postings: * http://www.mail-archive.com/proteledaforum@techservinc.com * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
[PEDA] OT: DIY: autorouter
Hi guys, I've been doing some research on the kinds of things to improve the autorouter (as per the desire to create my own). I've come up with some stuff that I'd like to bounce off ya's. Using a path-finding algorithm which assigns weights to various directions to travel (ie assuming that only 45deg angles are allowed, does anyone have a reason this isn't valid??) The default weighting would be to head towards the target pin (closest of the set), however if a large obstacle (ie dense gathering of wires) exists in the default path, then the algorithm would start to look at ways around the blockage, ie either using a via in which case you just perform the same operations on a different layer, or by going around the blockage. Weightings would be assigned to either of these (based on a static disadvantage for a via), if the side-track distance exceeds the disadvantage for the via however, then the via would be generated. The algorithm would take a single step forward (ie a single node point) for each connection pair, and then loop around and do the connection pairs again. I imagine that this would allow for better compromises to be made however other opinions on this are welcome. Just an update, Thanks for your time, Bevan Weiss * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * To post a message: mailto:[EMAIL PROTECTED] * * To leave this list visit: * http://www.techservinc.com/protelusers/leave.html * * Contact the list manager: * mailto:[EMAIL PROTECTED] * * Forum Guidelines Rules: * http://www.techservinc.com/protelusers/forumrules.html * * Browse or Search previous postings: * http://www.mail-archive.com/proteledaforum@techservinc.com * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
Re: [PEDA] OT: DIY: autorouter
Not to discourage you, but I think some of the best minds in the industry have been working on autorouters for 20 years and we have 'what we have.' If you can single-handedly come out with an autorouter that performs better than what's out there, you will certainly become famous / rich / etc. Someone will scoop your code up and integrate it, but for some funny reason, I just don't see that occuring. -Original Message- From: Bevan Weiss [mailto:[EMAIL PROTECTED]] Sent: Tuesday, October 15, 2002 4:35 AM To: Protel EDA Forum Subject: [PEDA] OT: DIY: autorouter Hi guys, I've been doing some research on the kinds of things to improve the autorouter (as per the desire to create my own). I've come up with some stuff that I'd like to bounce off ya's. Using a path-finding algorithm which assigns weights to various directions to travel (ie assuming that only 45deg angles are allowed, does anyone have a reason this isn't valid??) The default weighting would be to head towards the target pin (closest of the set), however if a large obstacle (ie dense gathering of wires) exists in the default path, then the algorithm would start to look at ways around the blockage, ie either using a via in which case you just perform the same operations on a different layer, or by going around the blockage. Weightings would be assigned to either of these (based on a static disadvantage for a via), if the side-track distance exceeds the disadvantage for the via however, then the via would be generated. The algorithm would take a single step forward (ie a single node point) for each connection pair, and then loop around and do the connection pairs again. I imagine that this would allow for better compromises to be made however other opinions on this are welcome. Just an update, Thanks for your time, Bevan Weiss * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * To post a message: mailto:[EMAIL PROTECTED] * * To leave this list visit: * http://www.techservinc.com/protelusers/leave.html * * Contact the list manager: * mailto:[EMAIL PROTECTED] * * Forum Guidelines Rules: * http://www.techservinc.com/protelusers/forumrules.html * * Browse or Search previous postings: * http://www.mail-archive.com/proteledaforum@techservinc.com * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
