Layout curved tracks/traces


Please refrain from personal attacks. (Or is it necessary that the moderators step in and close this?)


A personal fight deleted.

There is actually quite a lot of RF experience experience on this forum


I think you got a little carried away with the whiteout there.


It is hard to delete several posts from a thread that reply to others and have others replying to them.
I won’t lock the thread yet, but let’s all stop flinging mud about.


However this is not evidence of what we are discussing here. If line impedance doesn’t match load exactly (never will), changing even line length will change measured impedance and you don’t need curves for this. Then add impedance tolerance, etc. You would need to de-embed exactly the corner to make those assumptions.


What is not evidence? I realize that measuring equipment not anywhere close to being evidence as Google search results, but let’s just imagine for a second that it is. I thought everybody here were big proponents of empirical data just a couple of posts before?

I know everybody is an RF expert here today, but let’s do it anyway, it’s going to be fun, I promise. After all, we are after empirical evidence here… or so I thought.

Here is a plot (that apparently you didn’t need) of two 50 Ohm loads connected by a 48 Ohm line (which all of us would agree is not a perfect 50 Ohm line) at 915Mhz, which, according to the esteemed article should have no noticeable effects while utilizing right angle traces .

Since you could envision this in your head, and didn’t even need a plot, can you explain it for everybody, not so well versed in RF murky craft? Can you tell us what happens to the measured impedance as the line length increases? (Let me know if you need me to magnify the center of the chart for you :slight_smile:

Wow, wow! Slow down there! What exactly are we adding and what is included in that mysterious “etc.” I’m sorry, I thought we were pretending to be scientists here who are about to debunk every RF textbook ever written. If that’s our goal here, we should probably be a bit more specific.


My reason for challenging the youtube video was the claim “the RF traces are all nice and curved… the reason they do that is if you put sharp corners, it’s going to radiate energy from those sharp points”.

Now, there may be other reasons for not using “sharp corners”, but I believe the claim about radiating energy is bogus. When people make claims based on bogus reasons, I wonder what other bogus reasoning is being used.

I didn’t claim to be an RF expert, nor that rounded corners make no difference in any scenario. I do think if there is a significant effect, it should be easy to demonstrate it in a real world example. I’m really not sure why some people got so invested over it…


I didn’t read the article. All I’m claiming is that effect of 90 deg corner is very much less then what imperfect transmission line would result in.


I guess I didn’t explain it very well. Here is a picture from:


Ok, please now apply that to the cross section of a copper track on a PCB. A track is typically 35 um thick, with a width of 0.2mm, the aspect ratio is ~6000:1. That means that all tracks on a PCB have incredibly sharp edges, and should have a massive effect, if we follow the logic of your theory.

I’m afraid that fishing diagrams out of high school courses is not going to advance understanding very far.


I think since now on I will use only carefully bent wrap wires.


+1 for the picture! :wink:

I am curious however, out of all the pictures you had to choose from on google why did you pick the one that was wrong? Anyway, had you read the page you linked, or even just the paragraph under the picture, you would understand that this only applies under electrostatic conditions. Meaning the charges are not moving, there is no current flow. In this state the charges distribute along the surface until equilibrium is reached. The distribution is proportional to the curvature of the surface. Generally when we design a PCB we’re not too interested in the electrostatics of the board, we tend to be a little more concerned with the electrodynamics of the board, that is charges that move resulting in current flow. In this topic we have been discussing high frequency AC currents in particular, and the influence, if any, of 90 degree corners in PCB tracks.


It’s a two dimensional drawing that works foe a two dimensional trace. I don’t expect you to understand it I didn’t take a field theory course until the last quarter of my junior year. I never said it was reason to do curved traces, Only that it is not a simple as you seem to think.


Please refrain from personal attacks. @davidsrsb i get the feeling we really need to shut this down if it goes on that way.


The drawing is a cross section of a 3D object. In the real world, there are no 2D traces, everything is 3D, you can’t ignore that.

I understand electrostatic field theory, I took the course at A level. I didn’t suggest anywhere that “it is a simple”…quite the opposite, it appears!


A pity as there is some scope for serious analysis or measurement, but I agree.
Closing now