Maker Pro
Maker Pro

Transverse Electric mode Waveguide on PCB?

A

amdx

Jan 1, 1970
0
I have a friend asking for waveguide information, in reference to the
proposed PCB layout on this url:
http://img.villagephotos.com/p/2005-4/986073/TEAlaserellipsoidtest.jpg

He asks the following;
........................................................................................
What I need is the time-delay between spark generation and sympathetic
spark occurrence,
which will depend on spacing D and relative dielectric constant of the board
material. Another way to ask is "what is the propagation velocity within
such a waveguide?", or "what is the wavelength of the EM energy within such
a
waveguide?".

There will be a LOWER "cutoff frequency" for such propagation, too. So the
sparks will have to be fast.
......................................................................................
I think he's looking for the math to lead him in the right direction.
Thanks, Mike
 
B

Baron

Jan 1, 1970
0
amdx said:
I have a friend asking for waveguide information, in reference to
the
proposed PCB layout on this url:
http://img.villagephotos.com/p/2005-4/986073/TEAlaserellipsoidtest.jpg

He asks the following;
.......................................................................................
What I need is the time-delay between spark generation and
sympathetic
spark occurrence,
which will depend on spacing D and relative dielectric constant of the
board material. Another way to ask is "what is the propagation
velocity within such a waveguide?", or "what is the wavelength of the
EM energy within such a
waveguide?".

There will be a LOWER "cutoff frequency" for such propagation, too. So
the sparks will have to be fast.
.....................................................................................
I think he's looking for the math to lead him in the right
direction.
Thanks, Mike
Dunno about the math.... But he can get the "Dielectric Constant" from
the PCB manufacturer. A search on "microstrip" might help as well.
 
B

Benj

Jan 1, 1970
0
I have a friend asking for waveguide information, in reference to the
proposed PCB layout on this url:http://img.villagephotos.com/p/2005-4/986073/TEAlaserellipsoidtest.jpg

He asks the following;
.......................................................................................
What I need is the time-delay between spark generation and sympathetic
spark occurrence,
which will depend on spacing D and relative dielectric constant of the board
material. Another way to ask is "what is the propagation velocity within
such a waveguide?", or "what is the wavelength of the EM energy within such
a
waveguide?".

There will be a LOWER "cutoff frequency" for such propagation, too. So the
sparks will have to be fast.
.....................................................................................
I think he's looking for the math to lead him in the right direction.
Thanks, Mike

I really don't know what he's trying to do here, (Switching delay for
TEA laser discharge perhaps?) but what he has is an open traveling
wave structure as it isn't closed like a waveguide. In some
circumstances the traveling waves can be confined under the ground
planes so that the edges do not enter in significantly, but I'm not
sure it's the case in his device.

Generally speaking when you need a 'waveguide' on a PCB tri-plate line
is a geometry of choice. There are lots of data and formulas available
for such structures and even if he doesn't use one, it still should
give him some hints as to propagation velocities and the like. And
yes, waveguides have a low frequency cut off (waves switch from
propagating to evanescent).

That's all I know.
 
A

amdx

Jan 1, 1970
0
Benj said:
I really don't know what he's trying to do here, (Switching delay for
TEA laser discharge perhaps?) but what he has is an open traveling
wave structure as it isn't closed like a waveguide. >
Generally speaking when you need a 'waveguide' on a PCB tri-plate line
is a geometry of choice.

There are lots of data and formulas available
for such structures

Do you have any URLs for such data and formulas?
Thanks for your help,
Mike
 
W

Wimpie

Jan 1, 1970
0
I have a friend asking for waveguide information, in reference to the
proposed PCB layout on this url:http://img.villagephotos.com/p/2005-4/986073/TEAlaserellipsoidtest.jpg

He asks the following;
.......................................................................................
What I need is the time-delay between spark generation and sympathetic
spark occurrence,
which will depend on spacing D and relative dielectric constant of the board
material. Another way to ask is "what is the propagation velocity within
such a waveguide?", or "what is the wavelength of the EM energy within such
a
waveguide?".

There will be a LOWER "cutoff frequency" for such propagation, too. So the
sparks will have to be fast.
.....................................................................................
I think he's looking for the math to lead him in the right direction.
Thanks, Mike

Hello,

The propagation velocity is determined by "epsilon r" (c0/sqrt(er)).
It is in fact a parallel plate transmission line (with oval shape and
some shorts). For TEM propagation, there is no cut off frequency
(even DC power can be transported with a parallel plate transmission
line).

In you situation only the shorts (reflectors?) will impede low
frequency energy transport.

At wavelengths << size of structure, there will be focusing because of
reflection on the open edges. But one thing that appears strange to
me. The horizontal distance between the spark gap and the reflecting
short is not large with respect to the dielectric thickness.

Because of the short horizontal distance, only very high frequency
components of the discharge current will be effectively launched in
this structure (in case of 1.6mm thickness, er=4, you should think of
50 GHz). The longest wavelength will be about 4*(hor. distance from
gap to short). In that case there will be significant radiation losses
because of dielectric thickness in not very small with respect to
wavelength. So you might consider shorted edges.

As most of the discharge energy is in relative low frequency, I doubt
whether you get sufficient power in the wave-guide to get a real spark
at the right side of the PCB. A design challenge is to find a good
structure to efficiently couple the spark energy into the wave guide
(probably via high voltage coaxial entry?).

Best regards,

Wim
PA3DJS
www.tetech.nl (Dutch)
 
J

John Larkin

Jan 1, 1970
0
Hello,

The propagation velocity is determined by "epsilon r" (c0/sqrt(er)).
It is in fact a parallel plate transmission line (with oval shape and
some shorts). For TEM propagation, there is no cut off frequency
(even DC power can be transported with a parallel plate transmission
line).

In you situation only the shorts (reflectors?) will impede low
frequency energy transport.

At wavelengths << size of structure, there will be focusing because of
reflection on the open edges. But one thing that appears strange to
me. The horizontal distance between the spark gap and the reflecting
short is not large with respect to the dielectric thickness.

Because of the short horizontal distance, only very high frequency
components of the discharge current will be effectively launched in
this structure (in case of 1.6mm thickness, er=4, you should think of
50 GHz). The longest wavelength will be about 4*(hor. distance from
gap to short). In that case there will be significant radiation losses
because of dielectric thickness in not very small with respect to
wavelength. So you might consider shorted edges.

As most of the discharge energy is in relative low frequency, I doubt
whether you get sufficient power in the wave-guide to get a real spark
at the right side of the PCB. A design challenge is to find a good
structure to efficiently couple the spark energy into the wave guide
(probably via high voltage coaxial entry?).

Best regards,

Wim
PA3DJS
www.tetech.nl (Dutch)

I occasionally add an SMA footprint to a multilayer pcb layout, so I
can TDR the power planes. I've never observed edge-of-board
reflections, presumably because the FR4/copper structure is pretty
lossy at the sorts of frequencies involved.

On the board we just finished, I have a 2.5 volt power plane, about
5x7 inches, 12 mils from ground. One test SMA is in the center, one
sort of near a corner. So I can TDR and TDT the combo, and see how
things propagate and/or reflect inside the planes. I'll do that in a
week or so and post if anything interesting shows up.

I agree that the proposed board will have no useful focussing effect.

John
 
W

Wimpie

Jan 1, 1970
0
I occasionally add an SMA footprint to a multilayer pcb layout, so I
can TDR the power planes. I've never observed edge-of-board
reflections, presumably because the FR4/copper structure is pretty
lossy at the sorts of frequencies involved.

On the board we just finished, I have a 2.5 volt power plane, about
5x7 inches, 12 mils from ground. One test SMA is in the center, one
sort of near a corner. So I can TDR and TDT the combo, and see how
things propagate and/or reflect inside the planes. I'll do that in a
week or so and post if anything interesting shows up.

I agree that the proposed board will have no useful focussing effect.

John

Hello John,

I would expect reflections (in general when dielectric thickness <<
wavelength) because the edges behave like an open transmission line
with not that high radiation loss. Of course dielectric losses are
present, but in the frequency domain you can measure standing wave
patterns at large patches over a larger ground plane (for example a
halve wave patch antenna).

The problem with the setup as given in the drawing is that the
dielectric thickness is no longer small with respect to 0.25 lambda.
In addition (assuming 1.6mm dielectric), the frequency for effective
launching of a wave will be in the extremely high GHz range. Shorting
the edges will give less radiation loss because of end effects and the
dielectric, but doesn't reduce dielectric losses.

Maybe they want to do the experiment on low loss dielectric (PTFE,
ceramic, etc). The concept isn't new (with shorted edges), it is
used in microwave antennas to generate a continuous lines source
emitter (fan beam radiator, I have a small one for 10 GHz).

I am looking forward to your TDR/TDT results.

Best regards,

Wim
PA3DJS
www.tetech.nl (Dutch)
 
A

amdx

Jan 1, 1970
0
Wimpie said:
Hello John,

I would expect reflections (in general when dielectric thickness <<
wavelength) because the edges behave like an open transmission line
with not that high radiation loss. Of course dielectric losses are
present, but in the frequency domain you can measure standing wave
patterns at large patches over a larger ground plane (for example a
halve wave patch antenna).

The problem with the setup as given in the drawing is that the
dielectric thickness is no longer small with respect to 0.25 lambda.
In addition (assuming 1.6mm dielectric), the frequency for effective
launching of a wave will be in the extremely high GHz range. Shorting
the edges will give less radiation loss because of end effects and the
dielectric, but doesn't reduce dielectric losses.

Maybe they want to do the experiment on low loss dielectric (PTFE,
ceramic, etc). The concept isn't new (with shorted edges), it is
used in microwave antennas to generate a continuous lines source
emitter (fan beam radiator, I have a small one for 10 GHz).

I am looking forward to your TDR/TDT results.

Best regards,

Wim
PA3DJS
Hi Wim and all,
Take a look at this later drawing and see if my friend is any closer.
http://img.villagephotos.com/p/2005-4/986073/ellipsoidtest3.jpg
Your feedback has been encouraging,
Thanks, Mike
 
J

John Larkin

Jan 1, 1970
0
Hi Wim and all,
Take a look at this later drawing and see if my friend is any closer.
http://img.villagephotos.com/p/2005-4/986073/ellipsoidtest3.jpg
Your feedback has been encouraging,
Thanks, Mike

But why?

At wavelengths short enough for there to be optical focussing, the
different path lengths will smear the pulse that arrives at the focus,
so peak voltage will be down. And for the focus to be localized, the
whole structure will have to be many, many wavelengths in size.

The focussing effect of an ellipse is incoherent.

John
 
W

Wimpie

Jan 1, 1970
0
But why?

At wavelengths short enough for there to be optical focussing, the
different path lengths will smear the pulse that arrives at the focus,
so peak voltage will be down. And for the focus to be localized, the
whole structure will have to be many, many wavelengths in size.

The focussing effect of an ellipse is incoherent.

John

Hi John,

Are you really sure about the non-coherence. I was expecting that the
path length via one reflection at the ellipse is always the same
irrespective of the reflection point on the ellipse. The same
principle is/was used in the medical scene to treat kidney stones
(acoustical waves in water).

Best regards,


Wim
PA3DJS
www.tetech.nl
 
W

Wimpie

Jan 1, 1970
0
Hi Wim and all,
Take a look at this later drawing and see if my friend is any closer.http://img.villagephotos.com/p/2005-4/986073/ellipsoidtest3.jpg
Your feedback has been encouraging,
Thanks, Mike

Hello Mike,

To me this looks better. Now you don't have dielectric loss and
dispersion and because of the metallic walls, you don't have any
radiation loss. John has doubts about the coherence of the
reflection, so probably you should figure this out first.

In fact, the height (B) can be rather large (so above 0.25lambda). The
walls perfectly reflect TEM waves. When you have B>0.5lambda, it
depends on the coupling from the spark gap to the waveguide and
construction of receiving spark gap whether it will work. The
transmitting gap must generate TEM waves and the receiving gap must be
able to "guide" all the TEM energy into the gap. When B is around
0.25lambda or less, it easier to generate mostly TEM waves and to
receive the TEM waves (by the receiving spark gap).

When your complete sphere is many wavelengths large (for example B =
5mm, width 300mm) and the distance between the transmitting and
receiving gap is also large, you may remove the "reflecting" shorts.
As the received E field via the direct path will be negligible with
respect to the E field received by all the single reflection paths.

When the coherence question is demystified, the next point will be the
construction of the transmitting spark gap and receiving spark gap.

How you are going to check whether you get a spark at the receiver?
What is your actual application for this experiment?

Best regards,

Wim
PA3DJS
www.tetech.nl
 
J

John Larkin

Jan 1, 1970
0
Hi John,

Are you really sure about the non-coherence. I was expecting that the
path length via one reflection at the ellipse is always the same
irrespective of the reflection point on the ellipse. The same
principle is/was used in the medical scene to treat kidney stones
(acoustical waves in water).

Best regards,


Wim
PA3DJS
www.tetech.nl

Yes, reflecting (pun!) in the string-loop construction of the ellipse,
all reflected paths have the same length; and only the direct path is
unsynchronized. But it's still going to be dicey to get a sharp focus
of electric field in real life at practical wavelengths. Again, why?

John
 
J

John Larkin

Jan 1, 1970
0
Hello Mike,

To me this looks better. Now you don't have dielectric loss and
dispersion and because of the metallic walls, you don't have any
radiation loss. John has doubts about the coherence of the
reflection, so probably you should figure this out first.

In fact, the height (B) can be rather large (so above 0.25lambda). The
walls perfectly reflect TEM waves. When you have B>0.5lambda, it
depends on the coupling from the spark gap to the waveguide and
construction of receiving spark gap whether it will work. The
transmitting gap must generate TEM waves and the receiving gap must be
able to "guide" all the TEM energy into the gap. When B is around
0.25lambda or less, it easier to generate mostly TEM waves and to
receive the TEM waves (by the receiving spark gap).

When your complete sphere is many wavelengths large (for example B =
5mm, width 300mm) and the distance between the transmitting and
receiving gap is also large, you may remove the "reflecting" shorts.
As the received E field via the direct path will be negligible with
respect to the E field received by all the single reflection paths.

When the coherence question is demystified, the next point will be the
construction of the transmitting spark gap and receiving spark gap.

How you are going to check whether you get a spark at the receiver?
What is your actual application for this experiment?

Best regards,

Wim
PA3DJS
www.tetech.nl

If you're willing to add a distinct receiver to concentrate the field
at the second focus - say, two cones that almost touch in the middle,
making a small spark gap - then it's a spin on Hertz's experiments.
Yes, one should be able to observe a small spark at the receive node,
given enough input power at the transmit focus. Given enough power, an
arc could be achieved without a concentrator. Waveguides do arc over
at kilowatt to megawatt power levels.

A properly constructed spark gap can deliver fast megawatt pulses.


But why?

John
 
W

Wimpie

Jan 1, 1970
0
The proposed use is to adjust the phase of the pulse going to a TEA
laser gap. Take a look at the next drawing.
http://img.villagephotos.com/p/2005-4/986073/TEAlaserellipsoidtest2.jpg

Thanks, Mike

Hello Mike,

Thank you for your feedback with regards to the application.

I'm not familiar with the details of the pulse required for such laser
devices (I am in electronics and antennas). When your goal is to get
a high voltage wave front at the edge of the half ellipsoid, the
ellipsoid is not the preferred shape. In addition, because the wave
fronts do not concentrate as in the full ellipse example, now the
direct path may lead to premature discharge initiation.

Because of the Marx generator, I expect that you will use really high
voltage.

Good luck with the experiments,

Wim
PA3DJS
www.tetech.nl
 
A

amdx

Jan 1, 1970
0
Wimpie said:
Hello Mike,

Thank you for your feedback with regards to the application.

I'm not familiar with the details of the pulse required for such laser
devices (I am in electronics and antennas). When your goal is to get
a high voltage wave front at the edge of the half ellipsoid, the
ellipsoid is not the preferred shape. In addition, because the wave
fronts do not concentrate as in the full ellipse example, now the
direct path may lead to premature discharge initiation.

Because of the Marx generator, I expect that you will use really high
voltage.

Good luck with the experiments,

Wim
A little more cleaned up drawing.
http://img.villagephotos.com/p/2005-4/986073/TEAlaserellipsoidtest3.jpg
Thanks,Mike
 
Top