Видео

3yo video prob didn't had it at that time 😅

They take longer to do than a 45 degree angle in a CAD and provide not much benefit except they "look pretty". Now-a-days it's about the speed of turnaround on the routing to final production, functionality and proper operation rather than looks. KiCAD also doesn't allow you to draw arcs on copper layers for some reason.

I think it's either m or w in kicad

@@leandroebner1405 'd' is another way in kicad

@@leandroebner1405 I'm sure I heard him say "D".

Making traces equal length is the kind of thing you only bother to do when you really need to - just because it takes extra time and potentially extra space, so no point wasting that unless you need to. That's only really done if there's a timing margin (for example I routed some Ethernet RGMII interfaces recently and the datasheet for the PHY specified < 11ps margin between each of the five signals going in each direction). For arcs as well, I mean you can it's just it's easier to route a lot of signals with 45 degree corners, I think it's just going from hand drawn to CAD that people stopped bothering. Again it's just a matter of ease and time - the arcs look nice but take longer and don't provide much benefit. Even at microwave frequencies often you can just get by with chamfered corners (the chamfer makes up for the changing width which causes a discontinuity in the characteristic impedance of the track).

@@stephengentle2815 wow, thank you!

They don't do "rounded tracks" any more as those were hand drawn. It also takes up more board space to do an arc instead of a nice 45 degree, 45 degree turn.

Yeah this was super distracting

Would have made that side neater, but might complicate the layout to the left. The other thing is the connection at pins 1 and 16 has such a tight angle. I was always told to make pad and track connections at 90 degrees, avoiding acute angles.

For all boards? I work on too many boards where hitting a price point is more important than electromagnetic factors. At what frequency do you decide the rise time is fast enough to necessitate the extra time required for rounded corners?

​@@BrandonDrury with a KiCad addon it happens in seconds by the press of a button

@@martinvollderpro Isn't it more expensive to manufacture rounded corners? Or am I imagining that.? If it's the same price to manufacture round corners, as it is to have them just make them angled, then it would seem like the automatic default of the designers, would be to always request rounded corners.

@@Inertia888 @darrick.S I didn't get extra charged for using rounded tracks. Why would it be more expensive, a rounded corner is "smoother" than a 45 degree angle thus easier to print

yes, that indeed used to be true 30 years ago gramps, times change, no issue for modern fabs

The only concern with 45 degree angle tracks is at very high frequencies. Reflections will bounce back from a 45 degree angle. I guess I need to pull out the books to see which frequency that starts to matter. If I had to guess, I'd say anything less than 100Mhz is totally safe. I could be wrong on that. Keep in mind that every via is a 90 degree angle and it's impossible to make a dense board without a zillion vias.

@@BrandonDruryI think he means that <90 degree angles can cause problems during PCB etching. My understanding is that this isn’t really a concern anymore with modern PCB production processes.

​@@tookitogoYou're right, it really doesn't matter.

Have you seen some of the 3D printing guys using the UV screen of their resin printer for setting a board before etching?

I wonder if it can be done from the Gerber viewer?

@@DeShark88 oh theres an idea. I could also have a word with the laser softwares writer, see if he has an idea how to convert the kicad svg into something the software can read. kicad exports a white square as board area, the laser engraves it onto the pcb resist, so just one big square of bare copper lol

@@welshdave5263 yes i did think of trying to print the tracks on the board just with my pla printer, but lightly spraying a board with some black rattle can paint and putting it in my laser engraver seems an easier method. this summer i'l have the time (and hopefully money) to build an a3/a2 sized cnc and more powerful laser

@lezbriddon is seen laser etched spray paint done here. Also seen PLA printers with a waterproof permanent marker attached and used as a plotter to draw the tracks etc before etching the still bare copper.

It is a special design for discharging!

You'll see rounded traces in SUPER high speed digital designs. Off the top of my head, I don't recall which rise time that starts to matter. I believe it's in the Ghz range. Below that, it's generally acceptable to use the 45 degree method. The shape of the trace is a factor, but at lower frequencies, it's nothing compared to that bug catcher net absorbing noise created by the area within the loop of the signal AND it's return current. This is where using a ground plane underneath the trace and a stackup with thin prepreg come into play.

@@BrandonDrury ​But with modern PCB design and manufacturing methods, why not just use round bends _always?_ Straight lines are from pen-and-ruler era.

​@@u2bear377That's called pre-mature optimisation with little to no benefit. There are many other things that give more benefits for the time spent.

@@u2bear377 One reason is your EDA tool might not support it since it's a lot more complex to implement. I don't think KiCAD can do rounded bends in the editor yet, although now there are at least plugins/scripts for it and they keep talking about how it's coming to the editor eventually. The straight lines and 45 degree bevels are from the 80s CAD era, look at boards from the 60s and 70s when they were laid out with tape on glass and they're generally very curvy.

@@treelineresearch3387 How is the rounded trace more complex than a truetype font?

Great question! The problem is in the word "optimal". A trace with 20A DC flowing through it on a military board has a radically different definition of "optimal" from a trace with 2uA DC flowing through it on a child's toy which has a radically different definition of "optimal" for a 3Ghz clock. As of now, we don't really have an "Intent" setting on traces yet, but it's plausible to have this in the future. There may be a time when two high speed digital traces could let us set the "Intent" to be time aligned and the tools just do it. As is, we have to recognize the need for time alignment and do that ourselves at this stage of the game. I see vast changes occuring quickly, though. So you may be on to something!

​@@BrandonDrurysome programs have parallel track routing options for high speed signals. Keeping the tracks exactly the same length. Also, many, if not most programs have auto routing, but as mentioned above, it doesn't know what your circuit is supposed to do. It only knows that two contacts or pins are supposed to be connected together. It would be very time consuming to define each and every track's "intent" and have the program try and reroute all of them. Sometimes it's just easier to route them manually

Aside from specific impedance requirements (for which PCB routing tools *do* calculate the proper trace widths), there's never a "right width" for a trace, it's always a trade-off between heat production, space available, distance to other traces, etc. etc. If you have the space, might as well make the traces wider, like, 0.5mm width traces are massive and I have only used those for high power traces, but in a board like in the video there is the space, so why not...

@@rrrrrrrrrrrrrrrrrrrrr because sometimes too low of impedance can cause stability issues things like SMPS or other high frequency switches. It's a balance. Just creating the lowest resistance track is not always the best approach

Welcome to the real world. People make noises. Be a man and ignore it.