Changing Line Width of a Trace in a Net without DRC Errors

neck_rules.zip

Here's an example of a net routed with the 8 mil neck rules but the basic line width is 100 mils

Hi Redwire,

Thanks for the demo files. That works for me.

What I notice is that a trace has to contain at least a segment of 100mils. Otherwise, I have the same errors. I guess they stay true to the meaning of a necked trace.

In your file, if I delete the last trace segments to R1 and R2, this turns into a connection between these two. If I use Neck Mode to route this connection end to end, this 8mils trace has these same errors. In my design, these are long winding traces. Following the same concept, I have to find a spot for each trace to drop a 100mils segments somehow. It's going to look like pythons after having big meals.

It would be great to have a clean way of doing this, just 8mils traces. If not, I guess I just waive these errors and done with?.

TC

Hi Excellon1,

What you said is a very good practice. In this design, I have a few BLMxx ferrite beads for EMI and net separation. I also use nano PTC fuses for protection. It's similar idea but was not really intended for the routing purpose (didn't see these errors coming).

Over the last few days, I moved things around a bit and used many copper polygons to enclose those high-current nets. The Neck Mode routing together with longer neck length as you suggested did the job. The board is now fully routed without error. I need to learn how to post process it as my next task. I am pretty sure I will need help on this.

Thanks for your help again.

TC

Hi Redwire,

I see DRC Update icon on the top left, not sure if it is the same one you talked about. Unfortunately, every time I click on it, the software simply closes/crashes. I am not so sure what it is causing it. It also does the same when I click on Update DRC button in the window after Check - Design Status.

I am new to this OrCAD PCB, still learning but so far I cannot really say it is the most user friendly or efficient software. It sure needs a lot of mouse clicking, too many really for a single task. It seems like I have clicked over a few millions Oops and Done doing this board.

TC

Hi,

maybe you are using"product" software without patch ? It can explain crashes; "base" software without update are generaly not stable.

Help->about: return

17.2-2016 S059

with actual updates

Hmmm...correct button.  Crash should not be happening.  As JC Tess mentions, check your release.  The other thing that occasionally can happen is a corruption to the pcb database.  Run the tool called database doctor (standalone) or Tools->Database Check which will scan the database for some incorrect entries.  Personally, I just don't see a crash on this so something sounds whacked.

Hi redwire/jc teyssier,

It's not the software version but the database instead.

This software was licensed and installed in July with hotfix (per their install instructions). I just updated to the latest moments ago (17.20.059). The latest also crashes. I used the standalone DB Doctor and it found 8 errors of the same type as:

ERROR: in SHAPE (no segment)
class = BOUNDARY
subclass = BOTTOM
WARNING(SPMHA1-62): Illegal parent identifier.
Error was fixed, object was deleted.

I did not see any error marker on the layout. I guess it's something internal. Anyhow, the software does not crash anymore on the updated board file.

Thanks for the inputs.

TC

Hi redwire/jc teyssier,

It's not the software version but the database instead.

This software was licensed and installed in July with hotfix (per their install instructions). I just updated to the latest moments ago (17.20.059). The latest also crashes. I used the standalone DB Doctor and it found 8 errors of the same type as:

ERROR: in SHAPE (no segment)
class = BOUNDARY
subclass = BOTTOM
WARNING(SPMHA1-62): Illegal parent identifier.
Error was fixed, object was deleted.

I did not see any error marker on the layout. I guess it's something internal. Anyhow, the software does not crash anymore on the updated board file.

Thanks for the inputs.

TC