reflection in PCB traces

Go To Last Post
22 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I know I am not supposed to have right angles in PCB traces, and all the PCBs I have seen use two 45 degree bends to make a 90 degree change in direction.  I also seem to know that reflections at 90 degree bends happen at higher frequencies.  What frequency does it start to be a problem to have 90 degree trace bends?  Furthermore, isn't this also a problem at vias in multi layer boards?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

MarkThomas wrote:
What frequency does it start to be a problem to have 90 degree trace bends?

1.21 jigahertz, My God, Marty!

 

This only starts to be an issue with differential pairs where the sig's need to arrive at about the same time.

For ordinary single path signals, say DC to many MHz, it's not something to worry about, as electrons are near zero mass and will make the turn just fine regardless of angle.

Are you working on an interesting project where it may mater, or just curious? 

 

Jim

If I'm wrong, someone will point it out, quickly.

 

Click Link: Get Free Stock: Retire early! PM for strategy

share.robinhood.com/jamesc3274
get $5 free gold/silver https://www.onegold.com/join/713...

 

 

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ki0bk wrote:
This only starts to be an issue with differential pairs where the sig's need to arrive at about the same time.

 

I gotta disagree with this one.  When you are connecting to RF antennas on PC boards you always use right angles as the signal is actually traveling on the surface of the trace, so having the two 45's in there reduces/eliminates reflections.

 

East Coast Jim

 

 

I would rather attempt something great and fail, than attempt nothing and succeed - Fortune Cookie

 

"The critical shortage here is not stuff, but time." - Johan Ekdahl

 

"Step N is required before you can do step N+1!" - ka7ehk

 

"If you want a career with a known path - become an undertaker. Dead people don't sue!" - Kartman

"Why is there a "Highway to Hell" and only a "Stairway to Heaven"? A prediction of the expected traffic load?"  - Lee "theusch"

 

Speak sweetly. It makes your words easier to digest when at a later date you have to eat them ;-)  - Source Unknown

Please Read: Code-of-Conduct

Atmel Studio6.2/AS7, DipTrace, Quartus, MPLAB, RSLogix user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


Example:

Sorry Jim, even with RF signals, 90's can be used as shown.

Again, I'm not an expert here, just going on what I have seen.

Jim

 

Click Link: Get Free Stock: Retire early! PM for strategy

share.robinhood.com/jamesc3274
get $5 free gold/silver https://www.onegold.com/join/713...

 

 

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ki0bk wrote:
Sorry Jim, even with RF signals, 90's can be used as shown.

Sure, but LOOK at the traces...the edges around the corners are rounded, not square.

 

JIm

I would rather attempt something great and fail, than attempt nothing and succeed - Fortune Cookie

 

"The critical shortage here is not stuff, but time." - Johan Ekdahl

 

"Step N is required before you can do step N+1!" - ka7ehk

 

"If you want a career with a known path - become an undertaker. Dead people don't sue!" - Kartman

"Why is there a "Highway to Hell" and only a "Stairway to Heaven"? A prediction of the expected traffic load?"  - Lee "theusch"

 

Speak sweetly. It makes your words easier to digest when at a later date you have to eat them ;-)  - Source Unknown

Please Read: Code-of-Conduct

Atmel Studio6.2/AS7, DipTrace, Quartus, MPLAB, RSLogix user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ki0bk wrote:
Are you working on an interesting project where it may mater, or just curious? 

Unfortunately, not doing anything interesting.  A friend of mine is doing a board with swooping traces because he read that right angles are a problem.  This is on a board where the highest frequency is the 16 MHz crystal, and there is one trace from the CLKO pin to a couple of other chips.  Otherwise, everything is 3906 Hz PWM.  I think it is a little crazy, but maybe there are reflections at 16 MHz with right angles.  That is almost RF.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Though "what's happening" is the same, we tend to think about beveled PCB trace corners a bit differently for digital and for analog (quasi-sineweve) signals. It really comes down to what bad thing you might expect to happen.

 

In the case of digital, its the classic "pulse on a transmission line" problem. There is a small impedance discontinuity at every corner. That discontinuity tends to be more pronounced with a right-angle corner. For a bevel corner, the magnitude of the impedance change is lower and it is spread out, distance wise. The effect on pulses is fairly similar though it appears somewhat "softer" with the beveled corner. So, when do you have to pay attention to this? The critical factor is the rise and fall time of the digital signal. In free space, the speed of light is about 3E8 m/s which means 0.3m/ns = 30cm/ns or 12"/ns. On a circuit board the velocity is about 0.6C (depends on the dielectric constant of the board material), or about 18cm/ns or 7.5"/ns. Now consider a logic signal with a rise time of 10ns (typical of MCUs that can be clocked at 10MHz to 20MHz); at 18cm/ns, the rise of that pulse will be spread out over 180cm. A right-angle corner that might have an effective length of 1cm will generate reflections because of the impedance discontinuity, but those reflections will be lost in that much-longer rise or fall. When you start to get down to rise or fall times of 1ns, now it starts to become important.

 

In the case of analog, it depends on the signal frequency. Here, you can talk about things like "insertion loss" or "reflection coefficient". In fact, a small amount of the signal energy will be reflected by the discontinuity of the corner. This impedance size of the discontinuity is smaller with a bevel corner. Thus, the insertion loss is smaller. The effect of a corner depends on the ratio of the effective size of the corner compared to a wavelength of the signal. 2.4GHz WiFi/BT/etc has a wavelength of 12.5cm. Thus a right-angle corner with an effective length of 1cm is very significant. The effective length of a bevel corner is actually larger, but the size of the impedance discontinuity is a lot smaller, so you are actually ahead with the bevel. The bevel corner will actually still have some loss, because a small amount of the signal will be reflected, but the effect is less severe.

 

So, there are your criteria. For digital, its all about the rise time compared to the effective size of the corner. For analog, its all about signal loss due to reflections which depend on the effective size of the corner compared to the wavelength of the signal.

 

So, there are NO hard-and-fast rules about when you need bevel and when you don't. Its one of those engineering things that is all about "how much can you tolerate"? For example, and I am just picking some numbers out of thin air, suppose that your WiFi signal has to go around a corner and it looses 0.085db with a bevel corner and 0.097db with a right angle corner. Clearly, the right angle is "worse". But is it bad enough to require a bevel? That is up to you to determine!  Hint: it makes a big difference if that is a 1uW signal vs a 1MW signal.

 

For me, personally, I tend to use bevel corners when ever I can, even for circuits that don't "need" them. Why? I think they look better and EAGLE makes it easy. That is a good enough engineering reason for me, in my circumstance.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

Last Edited: Wed. Sep 11, 2019 - 06:11 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
For me, personally, I tend to use bevel corners when ever I can, even for circuits that don't "need" them. Why? I think they look better and EAGLE makes it easy. That is a good enough engineering reason for me, in my circumstance.

I do the same thing here, they just "look" better to me at least.

Jim

 

Click Link: Get Free Stock: Retire early! PM for strategy

share.robinhood.com/jamesc3274
get $5 free gold/silver https://www.onegold.com/join/713...

 

 

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
Thus a right-angle corner with an effective length of 1cm

I'm not sure what you mean by the "effective length" of a corner.

 

The same reasoning must apply to vias too.  It sounds like making two 45 degree angles is just good practice even though it probably doesn't matter with a 16 MHz mega.

 

Thanks Jim.  I appreciate it.  I didn't think about rise times, but I can see that has a high component frequency.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

MarkThomas wrote:
What frequency does it start to be a problem to have 90 degree trace bends?
10Gbps digital

MarkThomas wrote:
Furthermore, isn't this also a problem at vias in multi layer boards?
Yes; the via's inductance is its primary effect.

 

Who's Afraid of the Big Bad Bend?

by Dr. Howard Johnson. First printed in EDN magazineMay 2000

[bottom]

...

Overall, contemporary microwave designs can be 1000 times more sensitive to right-angle bends than are digital designs.

 

As digital designs push toward higher speeds, you may eventually reach a point where the right-angle bends begin to matter. For example, corners are just beginning to affect the design of 10-Gbps serial connections, and they also contribute perceptibly to skew in certain poorly routed differential pairs. ...

http://www.sigcon.com/Pubs/Keyword.html#vias

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
For me, personally, I tend to use bevel corners when ever I can, even for circuits that don't "need" them. Why? I think they look better and EAGLE makes it easy. That is a good enough engineering reason for me, in my circumstance.

 

ki0bk wrote:
I do the same thing here, they just "look" better to me at least.

 

When you say "bevel" do you mean two 45's, or something else that Eagle has?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

This is what I mean by bevel

 

 

Some use a mitered corner which does not come out very well in EAGLE. I have to make the grid REALLY FINE to get anywhere close to a miter but its pretty poor.

 

 

Jim

 

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

Last Edited: Thu. Sep 12, 2019 - 04:44 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

 know I am not supposed to have right angles in PCB traces

Maybe you are referring to the "old days" when right angles were somewhat verboten due to etching problems when making PCBs.  Supposedly, they would  forme "acid traps" & give a poor etch.

 

Now, processes are far better, so maybe the alarm is set low.  

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Take a look at this article,

 

https://resources.altium.com/pcb...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks waulu.  Interesting article.  I suspect we will all continue to use beveled turns anyway, even if it probably doesn't make a difference, especially at mega speeds.  I'm not sure I understand Sergey Ivanov's comment, or if he is just a sexist.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Sergey appears to be referring to "wiggles" added to pcb transmission lines to match the electrical length between pairs of lines (accordion). The reference to aperture is an apparent reference to the old Gerber plotter setting that controls the apparent pen size of the plotter; it does not mean much, any more, with laser plotters. But then it might refer to the aperture time of the sampler in the TDR. Not very clear about that, at all.

 

Not quite sure that I'd call it sexist. Certainly brusque, though. I put much more weight on the original blog report than the comments.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I was taught the old 'acid trap in the corners' problem, but it also put me into the habit of using right-angles on power traces (usually thicker) and 45deg. bends on signal traces (typically thinner.  Of course, bending the rules can be almost as much fun as bending the traces...).  Ever since it has been sorta handy to be able to look at a board and have some idea what that particular trace is supposed to be carrying.  S.

 

PS - Eagle can also do nice curved traces, around holes and things, as long as the radius isn't wildly bigger than the board itself.  Cute, if not always entirely helpful.  S.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hm - never tried curved traces. Not sure how that would fit my mental "ethic" about what a "real board" should look like. If I can remember, I'll give it a try on my next board.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hm - never tried curved traces.

I did and it was a bit of a pain...one PCB edge was sorta  "S" shaped in places & a few parallel traces needed to hug this edge...got everything looking pretty. 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 2

It is my personal opinion that there still is art in circuit layout.  If I have the time, I will often take a few days to just look at the layout after every signal is connected, and think of ways to make it a touch more elegant, a bit prettier, a little more organized.  Push a via around a little, straighten a few traces, swap a pin, &c.

 

And sometimes, during those days, I find I've committed a horrible blunder*, trivial to fix in the layout stage and rather more difficult later on, so it's not just dreamy aimless time noodling about.

 

Pretty designs work better, are easier to understand, and less likely to be badly wrong.  In my personal opinion and experience.  S.

 

* Actually, this is incorrect, because when I err it is a trivial oversight, and when someone else errs on my circuit board design it's a horrible blunder.  wink S.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It is my personal opinion that there still is art in circuit layout.  If I have the time, I will often take a few days to just look at the layout after every signal is connected, and think of ways to make it a touch more elegant, a bit prettier, a little more organized.  Push a via around a little, straighten a few traces, swap a pin, &c.

Absolutely, worth the detailing.  I had a PCB house telling me "great news" that they were starting my layout & should have it completed later that day or the next...rather than being happy at the "great news", I was rather upset & told them I expected it to take much longer than that.   What a ratsnest they made. They must have done it in a dark room with the monitors turned off.   Connector bodies hiding their pin labels---arrrg.  I get very suspicious when they say right off the bat this will take 4 or 6 layers....that is no substitute for an organized effort. Please put your ADC RC filters near the micro, not at the circuit way over there. 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Totally agree with Scroungre on art in circuit layout. I try to follow much the same practice, though I am not always so successful. But, like writing, it really helps to put it aside for more than a few days, then come back and review it.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net