I was wondering if the complement of an obtuse trace would cause issues when routing, would it be better to have a 90 degree trace or one that comes in like this?

obtuse and acute

or 90 degrees

90 degree bend

EDIT: Added a third photo for clarity here edit picture

  • \$\begingroup\$ Possible duplicate of Why is there such a strong preference for 45 degree angles in PCB routing? \$\endgroup\$ – Scott Seidman Nov 10 '16 at 21:54
  • 1
    \$\begingroup\$ They are both fine. You have no acute angles because you came in by way of the corner. Judging from the picture. I didn't read the text very carefully. \$\endgroup\$ – mkeith Nov 10 '16 at 21:57
  • \$\begingroup\$ @ScottSeidman No this isn't a duplicate, I understand why to use 45 degree angles, in the first image I uploaded the trace is has an extremely sharp bend on the right, and a very open bend on the left \$\endgroup\$ – and3212 Nov 10 '16 at 21:58
  • \$\begingroup\$ @mkeith I am confused, what do you mean "way of the corner" and why does that make the acute angle not matter \$\endgroup\$ – and3212 Nov 10 '16 at 21:59
  • 4
    \$\begingroup\$ Argh! I wrote a answer referring to your two pictures, and you edited the question to replace those pictures with something completely different. That's serious abuse of the volunteers you seek free help from. Put the original back, or I'll do it for you (and probably downvote while I'm at it due to the aggrevation). \$\endgroup\$ – Olin Lathrop Nov 10 '16 at 22:07

It's not going to matter much which one of the choices you use.

Generally it's good to avoid corners less than 90°. Note you don't seem to have a problem with the small 90° corners at left where a thinner trace leaves a fatter pad. Down to 90° is going to happen. Don't worry about it.

There are two issues here, acid traps and inductance. In some bad PCB processes, chemicals can get trapped inside sharp corners of the copper. This isn't going to be a problem if you use a competent board house.

A conducting path going around a corner has more conductance due to it acting like a coil. You can think of a 90° bend as a ¼ winding. Unless you have high frequency signals, this too is a minor issue. However, it's usually no big deal to use two 45° bends instead of one 90° bend, as you do in the red traces in the bottom of your pictures.

Remember that when hand-routing, you're not stuck with multiples of 45°. What I'd do for the case you show is to have the trace go straight from the via to the left-bottom corner of the pad. There is no reason for a bend part way like you show.

It may also be that in your bottom layout that the trace is too close to the left pad. Did you run a DRC check? It seems odd that all your traces are fatter than the minimum distance between traces. You expect power traces to be fatter, but signal traces are usually the minimum.

Here is a example of what I'm talking about:

This resulted from a combination of hand-routing and auto-routing. Note the various traces leaving pads at corners. This guarantees greater than 90° angle. There is also two 90° Ts at the top, and other places where traces leave the edges of pads. That's OK, but try to avoid less than 90° angles.

The red trace going around also illustrates turning a 90° corner by using two 45° corners (near top right).

  • \$\begingroup\$ DRC says it is fine, why should they be skinnier, is 16 mil too big? \$\endgroup\$ – and3212 Nov 10 '16 at 22:47
  • 1
    \$\begingroup\$ @and: 16 mil isn't too big if you got everything to route. It is bigger than necessary for most ordinary signals. I usually use 8 mil trace, 8 mil space design rules unless there is a specific reason not to. That's about as small as you can go but still have any board house in the world make the board without extra charges or chance of screwup, even with a sloppy process. \$\endgroup\$ – Olin Lathrop Nov 11 '16 at 11:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.