you are not supposed to bend cat 5 cable at right angles,
jus sayin...
jus sayin...
Why you don’t make right-angle traces and why lightning rods are pointy
- Thread starter etheory
- Start date
Help Support GroupDIY Audio Forum:
Andy Peters
Well-known member
CJ said:
Especially the type with a single strand per conductor -- the wire will break!
-a
Twenty Log
Well-known member
Absolutely.. Indeed, after working with the University of New Hampshire's "Interoperability Laboratory" (or "IOL", the World's premier Ethernet interoperability testing laboratory for folks like Cisco and 3COM et al) in the mid 1990s, Cat5 was reasonably new and indeed was an antennae (this was just before GigE was coming out and 100Base-T was reasonably new, and current technology for 100M used scramblers to take the Manchester Encoded Ethernet paradigm and make it more FCC friendly)... Never mind the mechanicals.... (of course, it is funny, since the 802.3 specification was released, no one actually adhered to the spec 100%)...
Yes, your design at 1.5 "Giggle Hertz" (GgglHz) or so could indeed work with right-angle traces... However, I wonder of the power budget (especially with battery operated equipment).... Indeed the Signal Integrity could show some ringing on the line with right-angles, but, especially if series termination resistors are employed, then the power budget is dissipated in said resistors (for the forward path and the attenuated backwards reflection path).... Of course, FPGAs, ASICs and the like have variable termination schemes in their I/O cells on silicon (I'm more of a GaAs and SiGe dude) that could implement the same by software control, rather than actual layout control (you might not know about the termination schemes if the design was compartmentalized within the business organization, at least that is how our PCB service bureau was operated in the late 1990s and early 2000s)... I wonder if the design were simulated in something like. say, SPECCTRA CCT (Cooper and Chang Technology, or in my recent case, Altium SI with IBIS models) or by actual measurement if there would not be excessive energy used for ringing and damping with termination schemes? All equating to battery life (perhaps incremental.. depends on the power budget)...
Whilst we were designing power amps for the cell phone radio section, we were increasingly being thwarted by power budget where, OK, we increased the power added efficiency (PAE) of the amplifier, only to have the designers of the cell phone say, "Oh yeah, we added a camera ------ oh and a flash too ------ ".... Ah crap, there goes the battery life anyway, even though we made another 10% efficiency on the GSM and DCS/PCS amplifiers in the MCM...
It all comes down to "routing channels"... These are the physical spaces between microchip packages on the PCB reserved for traces and there are tools to see what kind of SI (Signal Integrity) issues one might come across especially with tools like SPECCTRA (friggin' Cadence, which I hate @ $100k +/- per seat back then) that allow even a "place holder" type package with IBIS model basics for initial "critical net" rough placement et cetera... IBIS models are notorious for being "incorrect" although I suspect they have been getting better since 1999 when I was first using them.... Had to speak a little Japanese to talk with one of the top folks working in this arena....
Yes, your design at 1.5 "Giggle Hertz" (GgglHz) or so could indeed work with right-angle traces... However, I wonder of the power budget (especially with battery operated equipment).... Indeed the Signal Integrity could show some ringing on the line with right-angles, but, especially if series termination resistors are employed, then the power budget is dissipated in said resistors (for the forward path and the attenuated backwards reflection path).... Of course, FPGAs, ASICs and the like have variable termination schemes in their I/O cells on silicon (I'm more of a GaAs and SiGe dude) that could implement the same by software control, rather than actual layout control (you might not know about the termination schemes if the design was compartmentalized within the business organization, at least that is how our PCB service bureau was operated in the late 1990s and early 2000s)... I wonder if the design were simulated in something like. say, SPECCTRA CCT (Cooper and Chang Technology, or in my recent case, Altium SI with IBIS models) or by actual measurement if there would not be excessive energy used for ringing and damping with termination schemes? All equating to battery life (perhaps incremental.. depends on the power budget)...
Whilst we were designing power amps for the cell phone radio section, we were increasingly being thwarted by power budget where, OK, we increased the power added efficiency (PAE) of the amplifier, only to have the designers of the cell phone say, "Oh yeah, we added a camera ------ oh and a flash too ------ ".... Ah crap, there goes the battery life anyway, even though we made another 10% efficiency on the GSM and DCS/PCS amplifiers in the MCM...
It all comes down to "routing channels"... These are the physical spaces between microchip packages on the PCB reserved for traces and there are tools to see what kind of SI (Signal Integrity) issues one might come across especially with tools like SPECCTRA (friggin' Cadence, which I hate @ $100k +/- per seat back then) that allow even a "place holder" type package with IBIS model basics for initial "critical net" rough placement et cetera... IBIS models are notorious for being "incorrect" although I suspect they have been getting better since 1999 when I was first using them.... Had to speak a little Japanese to talk with one of the top folks working in this arena....
exactly. ???
Hey.. sorry to butt in on you people who obviously know a lot more about this than I do, but I had a thought pop into my head.
Just because the trace makes a right angle, does that mean the flow of electrons makes a right angle turn? I would suggest not... that trace isn't/can't be infinitely narrow, it has a few mm of width. We picture electrons zooming along, making a neat right angle turn, and zooming off again, but doesn't current follow the path of least resistance, so wouldn't that path be more of a curve? Especially if the field effects described above were imposing their force on it.
Just a thought, carry on!
-Mike
Just because the trace makes a right angle, does that mean the flow of electrons makes a right angle turn? I would suggest not... that trace isn't/can't be infinitely narrow, it has a few mm of width. We picture electrons zooming along, making a neat right angle turn, and zooming off again, but doesn't current follow the path of least resistance, so wouldn't that path be more of a curve? Especially if the field effects described above were imposing their force on it.
Just a thought, carry on!
-Mike
millzners
Well-known member
- Joined
- May 7, 2010
- Messages
- 219
MikeClev said:
That would likely be the case with DC but remember the hot topic here is regarding signals over right angle traces, where it's better to think in terms of waveforms than flow of electrons. In that case the typical argument made against right angles is one of two things: an impedance discontinuity due to the change in trace width in that corner or EMI radiation due to energy being concentrated on the sharp corner.
In times like this when laying out my boards, I often consult "Signal Integrity Issues and Printed Circuit Board Design" by Douglas Brooks. He summarizes each point separately and then concludes:
"On circuit boards, at least up into the gigahertz range, there is no particular performance reason to avoid 90-degree corners. We at UltraCAD avoid them anyway. We think mitered corners look better, but we cannot argue a performance reason for taking the time for the mitering process."
Therein lies the crux of the issue: there is no real theoretical or emperical evidence that suggests they are a problem, yet they are still commonly avoided for various reasons like asthetics, superstition or outright misunderstanding.
Twenty Log
Well-known member
Yes, I can see both sides, but it may also come down to amount of power and any VSWR (Vertical Standing Wave Ratio) that could be going through the right-angle...
We've blown up 0603 components (which are pretty big for some of the microwave 3 Watt power stuff we were doing) when tuning on a straight trace, never mind a right-angle... (we would use bends rather than mitered corners).
In a nominal digital design, there may be series termination that can reduce VSWR floating around on the trace from impedance discontinuities, mitigating issues at the corners but at the expense of resistor heat/power, PCB board space and driver microchip heat...
Interestingly the older PCI bus in computers would actually have reflections (VSWR) allowed on the buses so as not to have too much cost (power dissipation, PCB real estate and economically) for termination schemes... They would clock/latch in the data with specific timing after most reflections would have already died down upon bus state changes being completed to prevent logic errors...
I actually have to go through various steps to get right-angle traces in my CAD (Altium) system, especially when routing multiple traces at the same time (interactive bus routing with via fanout, all "automatic"/interactive)... default is 45 degree... but has hot keys to toggle through different styles including any-angle and bends and accordions to delay arrival times of signals (preventing signal skew)...
electrons at these higher frequencies follow the path of least impedance (usually least inductance)... Inductance can be reduced by having a solid uninterrupted ground plane directly underneath the trace of interest for the whole run of the trace... Otherwise the return currents, that normally flow directly underneath the trace within the ground plane would be diverted if the ground plane were interrupted thus increasing the loop / inductance... Signal Integrity analysis can show the ringing, critically damped or undershooting due to reflections or too much stray capacitance which could cause funny logic things to happen that might seem "intermittent" as chips heat up and drift and a nightmare when debugging...
There is also skin effect where the electrons ride on top of the conductor rather than in it... Low frequency skin effect is apparently trivial, but not insignificant for some coaxial cables even when considering 21 kHz (yup 21 kHz) or so we analyzed once in EE class... can't remember which RG-xyz it was though.....
We've blown up 0603 components (which are pretty big for some of the microwave 3 Watt power stuff we were doing) when tuning on a straight trace, never mind a right-angle... (we would use bends rather than mitered corners).
In a nominal digital design, there may be series termination that can reduce VSWR floating around on the trace from impedance discontinuities, mitigating issues at the corners but at the expense of resistor heat/power, PCB board space and driver microchip heat...
Interestingly the older PCI bus in computers would actually have reflections (VSWR) allowed on the buses so as not to have too much cost (power dissipation, PCB real estate and economically) for termination schemes... They would clock/latch in the data with specific timing after most reflections would have already died down upon bus state changes being completed to prevent logic errors...
I actually have to go through various steps to get right-angle traces in my CAD (Altium) system, especially when routing multiple traces at the same time (interactive bus routing with via fanout, all "automatic"/interactive)... default is 45 degree... but has hot keys to toggle through different styles including any-angle and bends and accordions to delay arrival times of signals (preventing signal skew)...
electrons at these higher frequencies follow the path of least impedance (usually least inductance)... Inductance can be reduced by having a solid uninterrupted ground plane directly underneath the trace of interest for the whole run of the trace... Otherwise the return currents, that normally flow directly underneath the trace within the ground plane would be diverted if the ground plane were interrupted thus increasing the loop / inductance... Signal Integrity analysis can show the ringing, critically damped or undershooting due to reflections or too much stray capacitance which could cause funny logic things to happen that might seem "intermittent" as chips heat up and drift and a nightmare when debugging...
There is also skin effect where the electrons ride on top of the conductor rather than in it... Low frequency skin effect is apparently trivial, but not insignificant for some coaxial cables even when considering 21 kHz (yup 21 kHz) or so we analyzed once in EE class... can't remember which RG-xyz it was though.....
LF skin effect also matters in high power mains distribution conductors, but these are all esoteric phenomena and not of much importance wrt audio PCB layouts (IMO).
JR
JR
> VSWR (Vertical Standing Wave Ratio
Often spelled Voltage Standing Wave Ratio. Maybe different for gigglehertzers.
> does that mean the flow of electrons makes a right angle turn?
Of course not.
Dig a ditch with a hard right-angle turn.
For the DC case, pour water to flow through. Water has high mass, thus inertia. It will tend to slam into the end and fall off to the side, though some will bend around the inside of the corner. Electrons have low mass and repel a lot. The first ones stand in the corner and the rest will swerve-curve to avoid them.
For the AC case, flood the ditch and wiggle a paddle. If you make 1-foot paddle-waves in a 4 foot wide ditch, there will be strong wave cross-patterns in the corner. However if you make slow 10-foot waves in a narrow 1-foot ditch the wave just flows through the corner without fuss.
In air and water ducts we can show some increase of friction in a sharp corner. Often this is less than the straight-line friction in the typical runs between corners. Some handbooks ignore most turns and over-state the straight-line friction so most results are good-enough with much less figuring.
I have a heating duct which I have run round-elbow and hard-bend... makes little difference. It replaced a smaller duct with _six_ elbows; that didn't flow well but I think the under-size was the main reason.
The test-boards in those papers clearly show the "extra friction" where the connector comes onto the PCB, but don't show any strong "friction backup" at the trace corners.
Often spelled Voltage Standing Wave Ratio. Maybe different for gigglehertzers.
> does that mean the flow of electrons makes a right angle turn?
Of course not.
Dig a ditch with a hard right-angle turn.
For the DC case, pour water to flow through. Water has high mass, thus inertia. It will tend to slam into the end and fall off to the side, though some will bend around the inside of the corner. Electrons have low mass and repel a lot. The first ones stand in the corner and the rest will swerve-curve to avoid them.
For the AC case, flood the ditch and wiggle a paddle. If you make 1-foot paddle-waves in a 4 foot wide ditch, there will be strong wave cross-patterns in the corner. However if you make slow 10-foot waves in a narrow 1-foot ditch the wave just flows through the corner without fuss.
In air and water ducts we can show some increase of friction in a sharp corner. Often this is less than the straight-line friction in the typical runs between corners. Some handbooks ignore most turns and over-state the straight-line friction so most results are good-enough with much less figuring.
I have a heating duct which I have run round-elbow and hard-bend... makes little difference. It replaced a smaller duct with _six_ elbows; that didn't flow well but I think the under-size was the main reason.
The test-boards in those papers clearly show the "extra friction" where the connector comes onto the PCB, but don't show any strong "friction backup" at the trace corners.
Twenty Log
Well-known member
PRR said:
This is very true... it is very often "Voltage" but I used the more generic "Vertical"...
http://www.maxim-ic.com/glossary/definitions.mvp/term/VSWR/gpk/815
bruno2000
Well-known member
CJ said:
Fiber doesn't like that either. ;o)
Best,
Bruno2000
Twenty Log
Well-known member
I recommend a whole fiber diet 
> nice to have all the braniacs here,
ummm... Like, I just work here... 8)
Essentially in the GaAs power amplifier for cellular handset market, we were hellbent to get every added gram of power added efficiency by sliding caps up and down the transmission line to get the sweet tuning spot (to deviate or verify simulation) for both power output and efficiency... otherwise things would just be wasted in heat or worse off destruction, lifetime degradation and the like...
> nice to have all the braniacs here,
ummm... Like, I just work here... 8)
Essentially in the GaAs power amplifier for cellular handset market, we were hellbent to get every added gram of power added efficiency by sliding caps up and down the transmission line to get the sweet tuning spot (to deviate or verify simulation) for both power output and efficiency... otherwise things would just be wasted in heat or worse off destruction, lifetime degradation and the like...
