Transmission of surges through right-angle termination 4

[electricpete]

This would be again considering 0.1 microsecond rise time pulses in a power system. There are two different types of concern:
1 - surges from power system propogating toward the motor.. we'd like to lessen them.
2 - surges generated by partial discharge inside the motor which propogate outward into terminal box, make a right angle turn to get to coupling capacitor filter network for sending partial discharge.

In both cases, there is occasions where the surge may encounter a 90 degree turn in the copper buswork within the terminal box. I'm guessing the buswork is 1/4" thick.

What effect does a 90 degree turn have on pulses that I assume are initially propogating in transverse electromagnetic mode.

What about a 90 degree tap. i.e. surge has the choice to go straight or make a 90 degree turn... will any of it go down the 90 degree tap>?



=====================================
(2B)+(2B)' ?

 

[electricpete]

The purpose was to give a qualitative demonstration showing more reflection for shorter rise time and less for slower rise time so we could see the mechanism by which the relationship between rise time and conductor width affected the result. For that I chose one rise time less than conductor width and one more, but neither dramatically more. What I think I learned (and kind of already suspected) is that for fast rise time relative to width, there is very little time for diffusion (per TLM or Huygen's principle) of the wave toward/into the 90-degree direction, so that most of the wave hits the far wall and almost doubles/bounces before it has had a chance to diffuse. In contrast the longer wave has more opportunity to diffuse and never hits the wall head on.

To make it easy and general purpose, I tried to model all my boundaries using and impedance mesh. Low impedance in conductor area and high impedance everywhere else including a 1-cell-wide strip around the edges. Then I blocked out the values in the center because they were distracting from the picture. There are some features of the boundaries that are quite obviously not exactly correct like the ridge on the left side of the wave, but I didn't spend a lot of time correcting them... maybe later. Another is that simple 2-D mesh creates anisotropy in the medium (wave propagates sqrt2 faster along the mesh than 45 degrees from the mesh). There is a correction available, but for simplicity I ignored it. There are the reasons I called it crude.

I am well aware of the relationship between rise time and dimension, but the dimensions I selected suited my purposes better (demonstration, not calculation of a magnitude). I don't think that picture of doubling and reflection would have been as clear comparing two rise times which were both many orders of magnitude higher than the width. I learned the speed of light in 3rd grade, and this speed will not be too much lower. But thanks for your assumption that I did not have that 3rd grade knowledge... we all know what they say about assumptions.

=====================================
(2B)+(2B)' ?

 

[electricpete]

My instinct says that the waveform in the .avi sims is showing way too much 'internal' reflections for the scales to be correctly matched. Almost as if you could shave the corner at 45° to 'bounce' the waveform edge around the corner like a beam of light.

That is what the model uses... high impedance in the non-conducting areas, so wave will bounce off of those. Perhaps the real world has a blurrier boundaries due to fringing field effects, but as a first approximation that is correct, isn't it?

=====================================
(2B)+(2B)' ?

 

[VEBill]

(All of this is within the context of this and your other thread on this forum. I'm assuming that both threads are related, and bother are on the subject of surges within bent bus-bars.)

If the purpose of this exercise is to refine your instincts, then it's best to start in the middle before working out to the extreme cases.

Assuming that the rise time of surge of interest is still around 0.1 microsecond, then your model bent conductor is the size of a building. That's pretty big. It's not that you didn't know that (reading minds is not my specialty), it's that the implications of this extreme scale ratio choice are significant.

The mental construct of an internal "wall", where the waveform bounces around internally, does not exist within correctly (more real-world) scaled conductors. Most conductors are essentially one-dimensional structures, often bent into other dimensions.

If I may re-purpose a famous phrase - Within most conductors as typically applied, "God does not play..." billiards with charge carriers.

 

[electricpete]

For 2" conductor, 3E8 m/sec(round numbers... should be lower) speed, the wave travels one foot per nanosecond and 1000 feet per microsecond. 1000 feet would be 6000 times the conductor width. That poses some challenges in my programming environment, but I tried attached a rise time 40x the conductor width.

To keep the avi file size managable, snapshots are taken every time the wave advances 20 grid points (before it was 5). If you look carefully you can see the leading edge jumping along the path for the first 14 snapshots (1st 5 seconds of avi) and you can see the trailing edge (where it turns constant) passing through between 20 and 25 seconds.

There is no discernible (from the graphic) reflection by this 2-D model even for rise time 1/40 of conductor cross width. How well that related to real life scenarios I'm not sure.

=====================================
(2B)+(2B)' ?

 

[electricpete]

Correction:

There is no discernible (from the graphic) reflection by this 2-D model even for rise time 1/40 of conductor cross width. How well that related to real life scenarios I'm not sure.

should've obviously been

There is no discernible (from the graphic) reflection by this 2-D model even for rise time 40 times the conductor cross width. How well that related to real life scenarios I'm not sure.

=====================================
(2B)+(2B)' ?

 

[electricpete]

I have made plenty of mistakes and I don't mind being corrected/questioned whether I'm right or wrong. However when I correct/question people, I generally don't try to do it by turning it into a joke at their expense.



=====================================
(2B)+(2B)' ?

 

[VEBill]

"...a joke at their expense."

Sorry.

Do you mind to point out exactly where I crossed the line?

 

[electricpete]

Hmmm. First pictures and quotes comparing me to this eccentric scientist. That alone could be good or bad, so we look for the context that follows.....

What follows is your discussion where you percieve my simulation to be based on complete ignorance of the scale (*). And to up the comic value of your comments, let's throw in "Well worth stealing for the copper value; bring a crane and a truck."

I hope you can see where that might strike me negatively. At any rate I suspect now it was not intended that way, maybe I just need more sense of humour and lighten up a little.

(*) By the way, I hope you can see that comparing 40xwidth to 4000x width would show nothing at all interesting. The comparison of 1 / 2 width to 4x width showed a lot more that we might possibly learn from the simulation. Relevance of that "learning" to a particular scenario's is an important item to judge and an open question for me.

The mental construct of an internal "wall", where the waveform bounces around internally, does not exist within correctly (more real-world) scaled conductors. Most conductors are essentially one-dimensional structures, often bent into other dimensions.

It may be the case. What is the explanation for the impedance bump that you referred to earlier?

=====================================
(2B)+(2B)' ?

 

[electricpete]

It may be the case

It appears to be the case from the simulation. 40x -> nothing reflected. Much longer rise times relative to cross section -> even less.

For right angle bend, there is also the fact that the effective width gets longer as you pass through the turn. But this is seen in the simulation and does not cause any reflection at 40x and beyond.

I wonder if there is some effect similar to self inductance... the wave at position X has different linkage to the wave at position X-deltaX when the line is curved ? That might be mixing quasistatic/field concepts with wave concepts... I'm not sure.

=====================================
(2B)+(2B)' ?

 

[electricpete]

I should have waited for you to answer. My last post was trying to address the previous post question "What is the explanation for the impedance bump that you referred to earlier?"

I would be interested to hear more about what kind of impedance bumps you see... including wiring configuration and rise time of incoming wave.

=====================================
(2B)+(2B)' ?

 

[VEBill]

Pete, my reference to the world-famous meme ("Please excuse the crudity of my model...") was intended as an amusing comparison of your quoted statement, certainly not of you personally. I apologize if it caused offense.

It's worth noting that your expertise is held in the highest possible regard by myself and many others on Eng-Tips. In comparison, all that I can offer is the occasional real-world 'boundary condition' that may occasionally guide and constrain the theoretical analysis.

...

What is "...the explanation for the impedance bump..." of a right angle bend?

Series inductance.

90° is obviously one-quarter of a full turn (excuse the embedded humour, it's unavoidable).

As a gedanken experiment, replace the 90° bend with a very smooth and continuous curve that continues to curl around and around. Make it as many turns as you wish.

It's a coil; an inductor.

Lots of series inductance.

Plenty of 'impedance bump' to restrict (choke) the flow of higher frequencies.

And not a single "wall" anywhere.

 

[Skogsgurra]

There are situations where the impedance discontinuity of a right angle bend may cause noticable effects. One such case is the DIN 41612 connector. See

http://www.erni.com/DB/PDF/DIN-PDF/ERNI-DIN-HFE160.pdf

Lots of data, geometrical and electrical, plus results from measurements on actual connectors to test different simulations against. The measurements show all aspects you can think of - including TDR.

The phenomenon does not have any practical consequences in power applications where voltage levels are orders of magnitude higher than the voltage induced in the discontinuity. Add to that that rise-times very seldom are shorter than 100 ns in power circuits while sub-ns are met in 'signal' electronics.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[electricpete]

VE1BLL - thanks for the kind words. They are probably not deserved, but I won't dwell on that. I certainly appreciate your knowledge and help. I'll have to think about that quarter turn. Makes some sense.

Gunnar - That is pretty much the same conclusion I was reaching. For most purposes the right angle makes no difference in power systems. Compared to computer systems, the dimensions are larger which makes wave effects more likely, but the rise times are lower by a much larger factor which makes wave effects less pronounced.

In contrast, tees/branches should have an effect on pulses even in a power system, although I can understand now that it doesn't make much difference whether you go straight through the Tee or make a turn.

=====================================
(2B)+(2B)' ?