Right angle traces

[talvor]

Hi, I currently in the process of laying out my first PCB for an 8x8 RGB LED Matrix.

I would like to know if using right angle traces is a bad thing? And if so why.

Thanks
Phillip

[Psi]

It's bad practice.
And 45deg just looks nicer.

The usual arguments against 90deg tracks are..

- They act as an antenna at high frequencies, producing noise and loading your signal (there's some debate about this)
- The copper track itself can break off the pcb easier because of the small area right at the corner. 

In reality it's not going to make any difference if your circuit is simple and slow speed. Like a typical 8bit mcu or 555 timer.
But, other engineers who see your design wont be impressed.
Best to always do 45deg, you never know when you may need to show your PCB to someone, if its got 90deg tracks you probably wont get offered a job.

[sleemanj]

I would like to know if using right angle traces is a bad thing? And if so why.

For your 8x8 led matrix purposes;
  Angles less than or equal to 90 degrees are usually considered to be  aesthetically displeasing (they look naff).
  The sharp angle can create problems with etching, especially if you do it yourself, it's easier to break a trace at a sharp corner.
other than that, copper is copper (for your project's purposes).

[IanB]

I think it's the sharp corners that are worse than the angles.

[BravoV]

Check this out -> NO ! You don't need to worry about 90 degrees corner at PCB trace

If you believe 90 degrees is bad, then never use vias on your board. 

Also quoting from PCB expert :

Right angle bends do not cause signal integrity problems at any practical edge speed
Right angle bends also do not cause EMI
Right angle bends are NOT acid trap
There is no good technical reason to prevent the use of right angle bends to route traces in a PCB.

"Vias used to route signal from one layer to another do not have a significant impact on signal quality even with 200 ps edges."

[talvor]

OK, so right angle bends are a bad idea.  Cool, now I get to play around in KiCad for a few more hours 

So where should I put the 2 x 45 degree bends, close to where the 90 degree one is or further along the trace.  That is should I be trying to make the straight parts of the trace as equal as possible, or does this not matter.

[alm]

In reality it's not going to make any difference if your circuit is simple and slow speed. Like a typical 8bit mcu or 555 timer.

Low speed is up to a few GHz in this case. That's one fast 555!

[GK]

You need a nice radius to each bend, to help the electrons take the corner without flying off. They move really fast you know.

[BravoV]

You need a nice radius to each bend, to help the electrons take the corner without flying off. They move really fast you know.

Not even close, if I'm not mistaken, according to Dr. Howard Johnson at TheAmpHour session, the physical speed of the electron moving inside the conductor is even slower than a walking ant.

[marshallh]

The only electrical problem with 90 degree bends is at high speed (>100mhz) where the small part of excess copper is a minute parasitic capacitance... if you are working with stuff fast enough to worry about it, you already know the problem. At ghz speeds it becomes a waveguide and that's when you see mitered edges on the bends.

45 bends look better though.

[BravoV]

The only electrical problem with 90 degree bends is at high speed (>100mhz) where the small part of excess copper is a minute parasitic capacitance... if you are working with stuff fast enough to worry about it, you already know the problem. At ghz speeds it becomes a waveguide and that's when you see mitered edges on the bends.

Again, in the Dr. Howard Johnson session at TheAmpHour, this is also discussed and he said that at the sharp 90 degrees tip at the corner creates the same effect no more than a small lump of solder at the pcb even at ghz signal.

[marshallh]

Exactly, you'd have to route DDR3-1600 using all right angle bends for it to approach being an issue

[GK]

You need a nice radius to each bend, to help the electrons take the corner without flying off. They move really fast you know.

Not even close, if I'm not mistaken, according to Dr. Howard Johnson at TheAmpHour session, the physical speed of the electron moving inside the conductor is even slower than a walking ant.

Which species of ant? There is a lot of variation you know.

[DutchGert]

The only electrical problem with 90 degree bends is at high speed (>100mhz) where the small part of excess copper is a minute parasitic capacitance... if you are working with stuff fast enough to worry about it, you already know the problem. At ghz speeds it becomes a waveguide and that's when you see mitered edges on the bends.

Again, in the Dr. Howard Johnson session at TheAmpHour, this is also discussed and he said that at the sharp 90 degrees tip at the corner creates the same effect no more than a small lump of solder at the pcb even at ghz signal.

In my experience Dr's unsually don't know much about production.
In the past 90* corners would lead to etching problems and indeed sharp edges come of the board easyer.

In your case it's fine to use sharp edges, it's makes you just look like an absoluut n00b .

[BravoV]

Which species of ant? There is a lot of variation you know.

Ok .. ok ... I get it, must be this one ... 

The Atom Ant !!!!! 

[GK]

Years ago in high school we all had to assemble in the library one afternoon to receive a lecture in electrical safety from some supposed expert that they (school admin) managed to find from who knows where. He was an old fart and shortly after he started yacking, it became quite clear that rather than being an expert on electrical safety, he was a bit of a nutter. Spent the whole time overly enthusiastically trying to scare us, eyes bulging and arms waving and all, with exaggerated examples and stories of real life electrocutions. He concluded his spiel with an account of some bloke who got his fingers in between the joiners of an electrical extension lead - "the electricity shot right up his arm, couldn't take the corner at the top and blew his shoulder right out!". Fair dinkum. Our technical studies teacher Mr Walkley marched out of the room shaking his head. The students thought the whole lecture was totally awesome, of course, except for a few of the girls.

[Mechatrommer]

so... this $800 worth of device from Agilent is a bad practice? or they purposely intending to flew some electrons away to make the device within spec?

[GK]

An absolute poop load of old stuff was layed out exclusively with 90 degree bends - especially digital boards in the '70s full of logic chips. I'm not sure that the manufacturing/etching argument was ever that valid.
 

[BravoV]

In my experience Dr's unsually don't know much about production.
In the past 90* corners would lead to etching problems and indeed sharp edges come of the board easyer.

Also in RF designs it could lead to reflection, thats why a lot of sensitive RF designs don't like via's.

In your case it's fine to use sharp edges, it's makes you just look like an absoluut n00b .

so... this $800 worth of device from Agilent is a bad practice? or they purposely intending to flew some electrons away to make the device within spec?

I guess we will have to wait for DutchGert for that, since he boldly claims Howard Johnson and Agilent are bunch of noobs. 

[free_electron]

The only problem is manufacturability.

Acid traps is still a problem, despite what the experts claim it  no longer is.
Sure, the problem has been solved for run of the mill boards with 8 mil traces... Once you get to hdi boards it becomes a problem again. It all has to do with the thickness of the copper vs the width of the trace.  etching a 3 mil trace in 1 ounce copper is borderline. Throw in a sharp corner and the misery begins.

Sharp corners also have other problems. The soldermask is easier damaged .

That stuff all doesnt maater for your one-off board.. But for industrial production it does as it has an impact on the yield and durability of the end product.

[Psi]

Are there any advantages to a 90deg tracks that you don't get from 45deg ?
Other than maybe routing a tight corner slightly easier i cant think of any.

[Mechatrommer]

Are there any advantages to a 90deg tracks that you don't get from 45deg ?
Other than maybe routing a tight corner slightly easier i cant think of any.

you can measure traces length easier with a ruler and one eye.

[BravoV]

you can measure traces length easier with a ruler and one eye.

You're right, like in calculating the resistance across the trace in high accuracy, eg. using copper trace as low ohm resistor. 

[DutchGert]

The only problem is manufacturability.

Acid traps is still a problem, despite what the experts claim it  no longer is.
Sure, the problem has been solved for run of the mill boards with 8 mil traces... Once you get to hdi boards it becomes a problem again. It all has to do with the thickness of the copper vs the width of the trace.  etching a 3 mil trace in 1 ounce copper is borderline. Throw in a sharp corner and the misery begins.

Sharp corners also have other problems. The soldermask is easier damaged .

That stuff all doesnt maater for your one-off board.. But for industrial production it does as it has an impact on the yield and durability of the end product.

This and the fact that using 45* usually leads to shorter traces -> more free space for components or planes: win-win

Aparrently my RF story was a big load of crap as I just found out.

And ofc it simply looks nicer with 45*

[Psi]

you can measure traces length easier with a ruler and one eye.

You're right, like in calculating the resistance across the trace in high accuracy, eg. using copper trace as low ohm resistor. 

Accuracy would be better measuring that with a low ohm meter.
Rather than relying on distance measurements and assumptions about the copper thickness being exactly what it should be.