KiCad gets support for length matched traces

[ConnorGames]

Commit message from 8 hours ago: "Merged the differential pair router & track length tuning tool."

It seems to work, although it is a little clunky right now, and you can't annotate length matched traces on the schematic far as I can tell, but that will probably be easy to add. To try it out, get a <8 hour old build of KiCad, open up Pcbnew, switch to opengl canvas, and go to route>tune track length or route>differential pair. Right click for preferences.

Looks like I now have one less excuse for not learning about high-speed digital design yet. What are your thoughts? What features does KiCad need next if it is aiming to be competitive with professional packages?

[c4757p]

Give it time before you dive in. Those of us poking at in on freenode:#kicad have found at least three crashy bugs so far.

It's a massive merge.

[krivx]

The demo video is worth a watch and should answer some questions:

[firewalker]

Yep. There are some nasty segfaults.

Alexander.

[JohnnyBerg]

Looks like I now have one less excuse for not learning about high-speed digital design yet. What are your thoughts? What features does KiCad need next if it is aiming to be competitive with professional packages?

I use KiCad a lot. I wouldn't need new features. I just wanna be more productive. There are a lot of things that could be more automated or done more logical.
When I make a change in the schematic for example, before I get that change in the pcb .. so many clicks.

Fix that first, then talk new features.

[c4757p]

I use KiCad a lot. I wouldn't need new features. I just wanna be more productive. There are a lot of things that could be more automated or done more logical.
When I make a change in the schematic for example, before I get that change in the pcb .. so many clicks.

Fix that first, then talk new features.

That is new features. Planned, too, I believe - the whole cvpcb step is to be removed along with some of its cruft, and full foward and back-annotation. Not sure if that's to come before or after the feature freeze that's coming soon to prepare for the next stable.

[nuno]

Yeah, I recently downloaded the "old stable" and it has some very annoying bugs on eeschema item selection, I guess "stable" is just for feature freeze (the KiCAD already crashed on me too). Still prefer my 2009 version. I would also vote to get things really stable and bug free instead of new features (or at least to avoid breaking the basic stuff). Software developers just can't stand not being constantly adding more bugs, I mean, features (I'm a software developer, so I'm allowed to joke :p). I should get involved. Wait, nahh... too many things already  .

[workwithme]

Full disclosure - and I'm a bit biased because I work with them, but Upverter has LMT pretty down pat.

You could find more info at https://upverter.com/solution/high-speed-interfaces/

If you've got any questions, toss em' my way!

Yusef

[timofonic]

Full disclosure - and I'm a bit biased because I work with them, but Upverter has LMT pretty down pat.

You could find more info at https://upverter.com/solution/high-speed-interfaces/

If you've got any questions, toss em' my way!

Yusef

I tried to use Upverter because of being able to import Eagle projects, but it had many quirks on it.

Cloud services can be useful for sharing projects between classmates and teachers, but I prefer native software (better if Free as In Freedom) and local.

I imported an Eagle 7.2.0 project and some stuff failed, so I gave up.

Anyway, this forum section is about KiCad. If you want advertising, pay to Dave

[Korken]

Does the length matching take via depth into consideration?
So if I have a trace that changes layer, it should add the electrical distance of the via to the trace length.

[krivx]

Does the length matching take via depth into consideration?
So if I have a trace that changes layer, it should add the electrical distance of the via to the trace length.

I believe traces are routed in pairs, so that both would change layers at the same time. Is it common to run matched differential traces on different layers? It sounds like you would have to trust the fab too much.

[miceuz]

I've just importes eagle board into kicad pcbnew yesterday, panelised, sent gerbers to pcb manufacturer. Everything went surprisingly smoothly

[Korken]

I believe traces are routed in pairs, so that both would change layers at the same time. Is it common to run matched differential traces on different layers? It sounds like you would have to trust the fab too much.

I was thinking about single ended traces.
It is very common, for example in DDR routing, to have length matched traces on different layers.
This because of the sheer number of signals so they all can't fit on a single layer.