I would like to know if it is possible to set the track ends to be straight instead of rounded. I saw some replies regarding drawing zones for rectifying this issue. Can anyone please help this in detail.
PCB manufacture uses an optical system, that plots tracks with a scanned circular aperture. This means that track ends cannot be truly straight, at best you could plot a rectangle with rounded corners.
It will just create a rectangle right? can I fill the rectangle to looks like a track?
Yes, a rectangle, but the corners will always be a little rounded. This is usually not a problem.
“Draw rectangle” from the “Select Item” list on the RHS of the workspace will allow filling.
Why do you want rectangular ends of tracks? Usually they are buried inside the pads.
The 50-ohm impedance of the track is determined by its central width and gap, right?So, if the end points are rounded, how will this impedance remain the same?nd if I’m using it for EM waves, there’s a chance of reflection at the rounded ends.
You may have a look at RF Tools
3 Likes
Do your tracks not start and finish with component pads?
What I remember from 80s photo-plotter used round and square apertures (rectangle pads were made by square apertures).
Since raster photo-plotters are used apertures are only theoretical beings just used to determine state for each pixel.
I have never designed RF PCBs but it was also my question. Do track ends are left floating and not come to pad that still can be rectangle?
Please don’t apologise.
I was only curious. Tracks usually end with pads. I was only interested in your use case. 
What is it?
1 Like
It is a simple PCB with a central portion where we can place a chip with my circuit. You can see the tracks on both sides, and at the end of these tracks, I can assign my ports. However, the region where the tracks meet the chip portion is rounded. So, my confusion is whether the impedance will remain 50 ohms in this area, and if so, whether there is a chance of wave reflection due to the rounded edges
Typically tracks are covered by solder mask. To be able to connect to your track ends the simplest solution seems to make footprint containing one pad (pads typically are not covered with solder mask) and just place it at track end.
If pad will have exactly the same width as your track than it will simply be rectangle end of track and without solder mask to allow for connecting to it.
I’m not sure of that kind of photo plotter still exists, I think they were starting to get replaced by out somewhere in the '80-ies
As piotr already mentioned (and wikipedia) the switch was to raster imaging devices.
The latest and greates apparently use LDI (Laser Direct Imaging) in which lasers directly write on the photo sensitized PCB without using intermediate films. Orbotec makes such machines with apparently 48 lasers in them. Probably very similar to the simplified method Marco Reps uses with a simple laser plotter: https://www.youtube.com/watch?v=gm5P74vcB84
I almost forgot: The square track ends, there is a feature request for that on gitlab.
Even direct write laser has a finite circular spot size and I suspect that it is not too small or you would risk stripes in the tracks and very slow PCB exposure
Yes, there will always be some rounding. That is one of the reasons that IPC recommends rounded corners for solder stencils. It’s better to have a defined radius, then an arbitrary radius that depends on the machines the stencil is made on. (The other reason is for better paste release)
Copper zones in KiCad can have “sharp” corners, but even then KiCad rounds the corners:
This rounding is defined by the “Minimum Width” parameter in the copper zone. That Minimum width is twice the used radius.
I did a short look into photoplotters, and the first side I bumped into was: Orbotech LP7008 Photoplotter | Artnet Pro Inc.
They claim a line width of 15um, and a resolution of 1.25um
Even a truly square end is not going to make the trace 50 Ohm at this end, in the RF world we try to use tapers as described by @maui upthread and either keep them very short ie <wavelength/10 or longer and part of an impedance transformer design