- atlc - transmission line/coupler design sofware.

All About Circuits - Newsgroup Archive
These forums contain an archive of usenet posts relating to electronics. These usenet posts are a great resource. Click the 'search' link to search for a topic, or browse through the posts.
Back To:

please note - you cannot post on these forums. this is a usenet archive, new posts are not permitted. please visit our forums to create a new post

atlc - transmission line/coupler design sofware.
Goto page 1, 2 Next

All About Circuits - Newsgroup Archive Forum Index -> sci.electronics.design

Author

Message

Dr. David Kirkby
Guest

Posted: Thu Sep 18, 2003 7:45 am Post subject: atlc - transmission line/coupler design sofware.

Some might be interested is some free (issued under the GNU General Public Licencse) of mine for the anaysis and synthesis of transmission lines and directional couplers (which are transmission lines anyway). The source is available for those wishing to build it. The code is very portable, so should run on any UNIX system. Someone else has produced some Windoze binaries from the source code. Anyway, take a look if interested. http://atlc.sourceforge.net/ There are some nice pretty examples around. Improvements to documentation or code are always welcolm -- "The day Microsoft makes something that doesn't suck is probably the day they start making vacuum cleaners." -Ernst Jan Plugge. Dr. David Kirkby, Senior Research Fellow, Department of Medical Physics, University College London, 11-20 Capper St, London, WC1E 6JA. Website: http://www.medphys.ucl.ac.uk/~davek Author of 'atlc' http://atlc.sourceforge.net/

John Larkin
Guest

Posted: Thu Sep 18, 2003 8:20 am Post subject: Re: atlc - transmission line/coupler design sofware.

On Thu, 18 Sep 2003 04:45:47 +0100, "Dr. David Kirkby"
<drkirkby@ntlworld.com> wrote:

Quote:

Some might be interested is some free (issued under the GNU General
Public Licencse) of mine for the anaysis and synthesis of transmission
lines and directional couplers (which are transmission lines anyway).

The source is available for those wishing to build it. The code is very
portable, so should run on any UNIX system.

Someone else has produced some Windoze binaries from the source code.

Anyway, take a look if interested.

http://atlc.sourceforge.net/

There are some nice pretty examples around. Improvements to
documentation or code are always welcolm

Hi, David,

We had a thread recently regarding the stupidity of nearly all
embedded-microstrip calculator programs. Someone pointed me to atlc,
and I've been using it to do my calcs. It's wonderful, and I am
immensely grateful you wrote it.

I've been using it under Windows with Paint, which really isn't bad
once you get rolling. I can snip a chunk off the width of a trace,
save the bmp, sic atlc on it, and save the results in about a minute
per point.

I know you don't like Windows (who does?) but most engineering apps
run under the beast, so lots of us are stuck with it. A GUI Windows
version would be great some day.

Thanks!

John

ps - is there an equivalent anywhere for solving sheet resistance
problems? I tried to fool atlc into doing this (used sheet
capacitance, Er=1e6) but it didn't make sensible results somehow.

Dr. David Kirkby
Guest

Posted: Fri Sep 19, 2003 6:56 am Post subject: Re: atlc - transmission line/coupler design sofware.

John Larkin wrote:

Quote:

On Thu, 18 Sep 2003 04:45:47 +0100, "Dr. David Kirkby"
drkirkby@ntlworld.com> wrote:

Some might be interested is some free (issued under the GNU General
Public Licencse) of mine for the anaysis and synthesis of transmission
lines and directional couplers (which are transmission lines anyway).

Quote:

I'm glad you like it. Over the years I have put quite a lot of work into
that, from a simple 66 line program to its current 10,000 lines of C and
a lot of documentation.

Quote:

I've been using it under Windows with Paint, which really isn't bad
once you get rolling. I can snip a chunk off the width of a trace,
save the bmp, sic atlc on it, and save the results in about a minute
per point.

If you send me a bitmap (compress it first please) of the sort of thing
you are looking at, I could probably write something to generate the
bitmaps for you automatically, like those at

http://atlc.sourceforge.net/generators.html

I assume the program 'create_bmp_for_symmetrical_stripline' can't
generate the bitmaps you want, or 'create_bmp_for_rect_in_rect' If so
those will create bitmaps in under a second, which is clearly a lot
quicker than using paint.

Quote:

I know you don't like Windows (who does?) but most engineering apps
run under the beast, so lots of us are stuck with it. A GUI Windows
version would be great some day.

I'm quite keen to see a platform independent one - probably a Java front
end. But it's clearly not a simple task and one I am not proficient at.
Perhaps someone else would write a front end for it. Any offers of help
much appreciated.

Quote:

ps - is there an equivalent anywhere for solving sheet resistance
problems? I tried to fool atlc into doing this (used sheet
capacitance, Er=1e6) but it didn't make sensible results somehow.

Perhaps you can explain in greater detail what you want. atlc might be
able to do it. Putting Er=1e6 would be like putting a highly conductive
material I think. One obvious problem I see is that permittivity has no
upper bound, but it does have a lower limit of Er=1. That is not so with
resistive materials. So that would screw atlc up I suspect, since many
of the calculations are done by first assuming Er=1.

--
"The day Microsoft makes something that doesn't suck is probably
the day they start making vacuum cleaners." -Ernst Jan Plugge.

Dr. David Kirkby,
Senior Research Fellow,
Department of Medical Physics,
University College London,
11-20 Capper St, London, WC1E 6JA.
Website: http://www.medphys.ucl.ac.uk/~davek
Author of 'atlc' http://atlc.sourceforge.net/

John Larkin
Guest

Posted: Fri Sep 19, 2003 8:52 am Post subject: Re: atlc - transmission line/coupler design sofware.

On Fri, 19 Sep 2003 03:56:35 +0100, "Dr. David Kirkby"
<drkirkby@ntlworld.com> wrote:

Quote:

I'm currently doing embedded microstrip: ground on the bottom,
turquoise FR4 some thickness, vacuum above, and a red trace embedded
somewhere within the dielectric.

Why not a general shape flinger program? Maybe from an indirect
command file if the command line got too long. Allow basic shapes to
be painted any color, anywhere. That could replace *all* of your
specialized shape-makers. Then I could write PowerBasic shells to
solve specific problems, like the embedded microstrip thingie, in
engineering units. Heck, I could even graph families of curves! Piece
of cake. [1]

Quote:

Ah. Maybe that explains it. I first tried to do the classic L-shaped
planar resistor, but got nothing that scaled (or looked) right. So
then I did a simpler case: a rectangular bar of dielectric, red/hot on
one end, green/ground on the other, 10 squares in area (long and
skinny) floating in space, and solved for capacitance, figuring
capacitance must be directly analogous to conductance. Got an answer.
Then I reduced the l/w ratio to 5 squares, expecting twice the
capacitance; but the ratio was 1.445. I was using a huge Er to swamp
the "leakage" of the vacuum. The field plots didn't look right at all.
Oh well...

Works great for transmission lines!

John

[1] Americanism for "really easy."

Dr. David Kirkby
Guest

Posted: Fri Sep 19, 2003 1:15 pm Post subject: Re: atlc - transmission line/coupler design sofware.

John Larkin wrote:

Quote:

I'm currently doing embedded microstrip: ground on the bottom,
turquoise FR4 some thickness, vacuum above, and a red trace embedded
somewhere within the dielectric.

Yes, none of the standard bitmaps generators for atlc
http://atlc.sourceforge.net/index2.html

which are all described at
http://atlc.sourceforge.net/generators.html

that can do that - you need to draw the bitmaps yourself.

Quote:

Why not a general shape flinger program? Maybe from an indirect
command file if the command line got too long. Allow basic shapes to
be painted any color, anywhere. That could replace *all* of your
specialized shape-makers.

That is a good idea. It's a shame I never thought of that before. Can
you think of any commands that would be needed apart from these below.
Where the upper case is the command in a command file and the lower case
the arguments it would need?

BACKGROUND width height Er
CIRCLE x_centre y_centre radius voltage_or_Er
RECTANGLE x_centre y_centre w h voltage_or_Er

so one's command file would have a background statement:

BACKGROUND 200 100 1.0 # 200 wide x 100 high with Er=1

and might have any (or multiples) of the following, where as above I've
used # to denote a comment.

CIRCLE 50 53 6 P # centre (50,53) radius 6, positive conductor.
CIRCLE 50 53 6 N # centre (50,53) radius 6, negative conductor.
CIRCLE 25 30 5 4.8 # centre (25,30) radius 5 and Er=4.8
RECTANGLE 30 40 5 8 P # rectangle centered (30,40) 5x8, positive
RECTANGLE 30 40 5 8 N # rectangle centered (30,40) 5x8, negative
RECTANGLE 30 40 5 8 10.2 # rectangle centered (30,40) 5x8, Er=10.2

there would probably be some point in having a command

ENCLOSED_BACKGROUND width height border_thickness Er
to generate the borders for enclosed problems, to save someone the
hassle of working out the positions of 4 different rectangles.

Any comments on the above ???

Quote:

Perhaps you can explain in greater detail what you want. atlc might be
able to do it. Putting Er=1e6 would be like putting a highly conductive
material I think. One obvious problem I see is that permittivity has no
upper bound, but it does have a lower limit of Er=1. That is not so with
resistive materials. So that would screw atlc up I suspect, since many
of the calculations are done by first assuming Er=1.

Ah. Maybe that explains it. I first tried to do the classic L-shaped
planar resistor, but got nothing that scaled (or looked) right. So
then I did a simpler case: a rectangular bar of dielectric, red/hot on
one end, green/ground on the other, 10 squares in area (long and
skinny) floating in space, and solved for capacitance, figuring
capacitance must be directly analogous to conductance. Got an answer.
Then I reduced the l/w ratio to 5 squares, expecting twice the
capacitance; but the ratio was 1.445. I was using a huge Er to swamp
the "leakage" of the vacuum. The field plots didn't look right at all.
Oh well...

When you multiple dielectrics (say five with Er=1.006, 2, 4, 8 and 10),
atlc does *two* simulations. The first is done at Er=1, despite the fact
your problem has no material with Er=1. That allows atlc to find the
capacitance per metre C' assuming a vacuum. You might think is a useless
bit of information, since you have no vacuum, but 5 dielectrics all with
Er > 1.0. However, that is not so, as it is possible to compute the
inductance per metre from

L = mu0/C', where mu0 is permeability of free space

Inductance is independent of the dielectric.

Next atlc re-runs the simulation using all the dielectrics present
(1.006, 2, 4, 8 and 10). This time it is able to compute the correct
capacitance per metre C, but is unable to determine the inductance per
metre L. However, since the inductance per metre has already been found
by assuming Er=1, it is easy to find Zo:

Zo=sqrt(L/C)

So when there are multiple dielectrics, the simulation is run twice in
order that both the inductance/m and capacitnce/m may be found.

If there is just a single dielectric with say Er=5, then again atlc does
the simulation at Er=1, computes C', computes L, computes C=Er*C' and
works out Zo from

Zo=sqrt(L/C).

I think it these processes that are causing problems if you try to use
atlc for a problem it was not designed for.

Quote:

Works great for transmission lines!

Good, I'm glad of that.

With a few relativley minor changes, atlc could handle the resistive
problem you desire. But to be honest it is not something that interests
me, so it is not something I am likely to do. In any case, I think the
name of the program would need changing too. Perhaps 'apt' for Arbitrary
Potential Calculator or something similar.

Feel free if you want to take atlc's code and change it to do your
problem - I don't mind giving you some guidance. I think it would just
need lots of "if" statements removed, as the finite difference basic
code is there in place, but has all these unwanted branches, peforming
multiple simulations, which you screwing you up.

Quote:

Piece of cake. [1]
[1] Americanism for "really easy."

Yes, we use that in the UK too. There is another four letter word that
is sometimes used to replace cake, but perhaps its only the British who
use that.

--
"The day Microsoft makes something that doesn't suck is probably
the day they start making vacuum cleaners." -Ernst Jan Plugge.

Dr. David Kirkby,
Senior Research Fellow,
Department of Medical Physics,
University College London,
11-20 Capper St, London, WC1E 6JA.
Website: http://www.medphys.ucl.ac.uk/~davek
Author of 'atlc' http://atlc.sourceforge.net/

John Larkin
Guest

Posted: Fri Sep 19, 2003 7:34 pm Post subject: Re: atlc - transmission line/coupler design sofware.

On Fri, 19 Sep 2003 10:15:15 +0100, "Dr. David Kirkby"
<drkirkby@ntlworld.com> wrote:

Quote:

John Larkin wrote:

I'm currently doing embedded microstrip: ground on the bottom,
turquoise FR4 some thickness, vacuum above, and a red trace embedded
somewhere within the dielectric.

Yes, none of the standard bitmaps generators for atlc
http://atlc.sourceforge.net/index2.html

which are all described at
http://atlc.sourceforge.net/generators.html

that can do that - you need to draw the bitmaps yourself.

Why not a general shape flinger program? Maybe from an indirect
command file if the command line got too long. Allow basic shapes to
be painted any color, anywhere. That could replace *all* of your
specialized shape-makers.

That is a good idea. It's a shame I never thought of that before. Can
you think of any commands that would be needed apart from these below.
Where the upper case is the command in a command file and the lower case
the arguments it would need?

BACKGROUND width height Er
CIRCLE x_centre y_centre radius voltage_or_Er
RECTANGLE x_centre y_centre w h voltage_or_Er

so one's command file would have a background statement:

BACKGROUND 200 100 1.0 # 200 wide x 100 high with Er=1

and might have any (or multiples) of the following, where as above I've
used # to denote a comment.

CIRCLE 50 53 6 P # centre (50,53) radius 6, positive conductor.
CIRCLE 50 53 6 N # centre (50,53) radius 6, negative conductor.
CIRCLE 25 30 5 4.8 # centre (25,30) radius 5 and Er=4.8
RECTANGLE 30 40 5 8 P # rectangle centered (30,40) 5x8, positive
RECTANGLE 30 40 5 8 N # rectangle centered (30,40) 5x8, negative
RECTANGLE 30 40 5 8 10.2 # rectangle centered (30,40) 5x8, Er=10.2

there would probably be some point in having a command

ENCLOSED_BACKGROUND width height border_thickness Er
to generate the borders for enclosed problems, to save someone the
hassle of working out the positions of 4 different rectangles.

Any comments on the above ???

Close to what I was thinking. The only part I don't get is how the
dielectric constants would be connected to atlc, at least in the
general case. I assume the "4.8" works because the new splatter
program knows it's turquoise, and atlc knows turquoise is 4.8.

You could - maybe have to - do both filled and hollow circles and
rectangles, which would supercede the ENCLOSED_BACKGROUND thing,
although it looks handy by itself.

Please make everything case insensitive; real men don't use lowercase.

Quote:

Feel free if you want to take atlc's code and change it to do your
problem - I don't mind giving you some guidance. I think it would just
need lots of "if" statements removed, as the finite difference basic
code is there in place, but has all these unwanted branches, peforming
multiple simulations, which you screwing you up.

I occasionally design weird current shunts, punched or photoetched
flat manganin or zeranin thingies, and I'd like to be able to predict
the resistance of arbitrary shapes. People who do hybrids (and
monolithic ic's, too) have similar problems.

Sounds like fun, but, alas, I have a day job. Like, roughly, 14 hours
a day.

Quote:

Piece of cake. [1]
[1] Americanism for "really easy."

Yes, we use that in the UK too. There is another four letter word that
is sometimes used to replace cake, but perhaps its only the British who
use that.

No, we refer to all Microsoft products that way.

John

John Larkin
Guest

Posted: Fri Sep 19, 2003 8:49 pm Post subject: Re: atlc - transmission line/coupler design sofware.

On Fri, 19 Sep 2003 10:15:15 +0100, "Dr. David Kirkby"
<drkirkby@ntlworld.com> wrote:

Quote:

Of course! Light goes on. Smacks head. Nature is wonderful. If you
know C' and the prop velocity, you can compute L. And the velocity
must be c.

Duh. [1]

John

[1] Yet another Americanism.

qrk
Guest

Posted: Sat Sep 20, 2003 6:03 am Post subject: Re: atlc - transmission line/coupler design sofware.

On Fri, 19 Sep 2003 10:15:15 +0100, "Dr. David Kirkby"
<drkirkby@ntlworld.com> wrote:

Quote:

and might have any (or multiples) of the following, where as above I've
used # to denote a comment.

CIRCLE 50 53 6 P # centre (50,53) radius 6, positive conductor.
CIRCLE 50 53 6 N # centre (50,53) radius 6, negative conductor.
CIRCLE 25 30 5 4.8 # centre (25,30) radius 5 and Er=4.8
RECTANGLE 30 40 5 8 P # rectangle centered (30,40) 5x8, positive
RECTANGLE 30 40 5 8 N # rectangle centered (30,40) 5x8, negative
RECTANGLE 30 40 5 8 10.2 # rectangle centered (30,40) 5x8, Er=10.2

I like this sort of thinking, parameter files that the generator
program reads. Makes life easier when doing small modifications to
your structure.

Don't forget custom colour parameters
RECTANGLE 30 40 5 8 7.2 c035a0 # rectangle centered (30,40) 5x8,
Er=7.2, colour c035a0

Mark

John Larkin
Guest

Posted: Sat Sep 20, 2003 6:42 am Post subject: Re: atlc - transmission line/coupler design sofware.

On Fri, 19 Sep 2003 19:03:54 -0700, qrk <mark@reson.DELETE.ME.com>
wrote:

Quote:

On Fri, 19 Sep 2003 10:15:15 +0100, "Dr. David Kirkby"
drkirkby@ntlworld.com> wrote:

and might have any (or multiples) of the following, where as above I've
used # to denote a comment.

CIRCLE 50 53 6 P # centre (50,53) radius 6, positive conductor.
CIRCLE 50 53 6 N # centre (50,53) radius 6, negative conductor.
CIRCLE 25 30 5 4.8 # centre (25,30) radius 5 and Er=4.8
RECTANGLE 30 40 5 8 P # rectangle centered (30,40) 5x8, positive
RECTANGLE 30 40 5 8 N # rectangle centered (30,40) 5x8, negative
RECTANGLE 30 40 5 8 10.2 # rectangle centered (30,40) 5x8, Er=10.2

I like this sort of thinking, parameter files that the generator
program reads. Makes life easier when doing small modifications to
your structure.

Don't forget custom colour parameters
RECTANGLE 30 40 5 8 7.2 c035a0 # rectangle centered (30,40) 5x8,
Er=7.2, colour c035a0

Mark

Yup, that closes the loop on dielectric constant. But what about wall
thickness? And filled/hollow structures?

It would be cool to allow color (the standard names) as alternates to
P N 4.8.

RECTANGLE 30 40 5 8 RED

Er, I think you need a command to declare the bitmap file name, too.

John

Dr. David Kirkby
Guest

Posted: Mon Sep 22, 2003 10:26 am Post subject: Re: atlc - transmission line/coupler design sofware.

qrk wrote:

Quote:

I meant to post something about custom colours, but for some reason it
did not appear. I have two ideas. I think the second is the better, but
I'm open to suggestions.

IDEA #1
---------

You create a command file, as before. I add an option for a BITMAP name
as suggested.

BITMAP_NAME new-transmission-line.bmp
CIRCLE 50 53 6 P # centre (50,53) radius 6, positive conductor.
RECTANGLE 30 40 5 8 10.2 # rectangle centered (30,40) 5x8, Er=10.2

but if you want an odd dielectrics, you just enter the dielectric you
want, without bothering to specify a colour.

RECTANGLE 30 40 5 8 134.3349873478 # Er=134.3349873478
RECTANGLE 30 40 5 8 21.33443 # Er=21.33443

then make_bmp decides on a colour to make Er=134.3349873478 and
Er=21.33443 from a list of say 10 new colours it programmed to generate.
Lets say the first three colours are 0x435334, 0x123465, 0x897656. Then
at the end it prints a 2 lines like this:

Now run the following command to get atlc to compute the properties for
you:
% atlc -d 0x435334=134.3349873478 -d 0x123465=21.33443
new-transmission-line.bmp

If you then create another file, with just one odd dielectric this time:

BITMAP_NAME microstrip.bmp
RECTANGLE 30 40 5 8 1.23456789 # Er=1.23456789

then make_bmp will print:

Now run the following command to get atlc to compute the properties for
you:
atlc -d 0x435334=1.23456789 microstrip.bmp

So the same colour (0x435334) is once used to indicate Er=134.3349873478
and another time to indicate Er=1.23456789. But make_bmp will each time
indicate the correct command line to run atlc with. There could even be
an option for make_bmp to automatically invoke atlc using the system()
command in C.

Of course, this means the user can't keep his/her own database of say 5
dielectrics and 5 colours, as the colours that are used to represent any
particular dielectric are not fixed.

IDEA #2 (the better one I think)
------------------------------

Another option would be for the user to create a file called atlc.cfg,
which is a configuration file. In that file the user creates a list of
colours and permittivities that they ever use. Each time they use a new
dielectric, they decide on a colour and add it to atlc.cfg

# this is file atlc.cfg
1.234 0x34a344
9.343 0x5a2bf3

make_bmp then looks in there for any permittivity it does not know
about, generating an error if you specify one that is neither one of the
14 or so colours on listed in the configuration file. Assuming a
permittivity is listed in the atlc.cfg file, it does not need to be
specified when invoking atlc. So both make_bmp and atlc both use the
same configuration file.

--
"The day Microsoft makes something that doesn't suck is probably
the day they start making vacuum cleaners." -Ernst Jan Plugge.

Dr. David Kirkby,
Senior Research Fellow,
Department of Medical Physics,
University College London,
11-20 Capper St, London, WC1E 6JA.
Website: http://www.medphys.ucl.ac.uk/~davek
Author of 'atlc' http://atlc.sourceforge.net/

John Larkin
Guest

Posted: Mon Sep 22, 2003 7:26 pm Post subject: Re: atlc - transmission line/coupler design sofware.

On Mon, 22 Sep 2003 07:26:58 +0100, "Dr. David Kirkby"
<see-my-signiture-for-email-address@nowhere.com> wrote:

Quote:

IDEA #2 (the better one I think)
------------------------------

Another option would be for the user to create a file called atlc.cfg,
which is a configuration file. In that file the user creates a list of
colours and permittivities that they ever use. Each time they use a new
dielectric, they decide on a colour and add it to atlc.cfg

# this is file atlc.cfg
1.234 0x34a344
9.343 0x5a2bf3

make_bmp then looks in there for any permittivity it does not know
about, generating an error if you specify one that is neither one of the
14 or so colours on listed in the configuration file. Assuming a
permittivity is listed in the atlc.cfg file, it does not need to be
specified when invoking atlc. So both make_bmp and atlc both use the
same configuration file.

I like #2. How about...

# this is file atlc.cfg
1.234 0x34a344 puce # foamed treakle
9.343 0x5a2bf3 mauve # alumina

Now allow the new shape flinger program to accept the config file
color names too. So a user gets to build up his own named dielectric
set, only deal with hex colors once, and can look at a bitmap image
and *see* what's happening. The names could actually be anything, not
just colors: "prepreg" or "sheetrock". I think something like this
would improve the "drivability" of the whole system.

And how about some snazzier program names? atlc is not too catchey,
and things like make_bmp and bitmap_name are hard to say or type. Have
consideration for us slow typists! I nominate "fling" as the new prog
name, and shorter verbs in the command file (bitmap, rect, etc.)

John

Helmut Sennewald
Guest

Posted: Mon Sep 22, 2003 11:35 pm Post subject: Re: atlc - transmission line/coupler design sofware.

"Dr. David Kirkby" <drkirkby@ntlworld.com> schrieb im Newsbeitrag
news:3F692A6B.DB254F71@ntlworld.com...

Quote:

Hello David,
thanks for this great program.
I tested three structures with the same ratio W/H.
First, I placed the conductor about in the middle. Then I placed it
closer to the left side in the second example. In the third example
I added a wall on the left side of conductive material(green).

The results:
------------
Case 2 and 3 have shown 10 Ohm differencec compared to 1.
Is the left and right side assumed to be connected to the ground when
calculating the fields/impedance?
I have read something about truncated field in the manual.
What does it mean here?
By the way, the result of case 3 was nearly the same as in case 2.

Is there any rule of thumb how far I should keep away from the left
and right side?

Best Regards
Helmut

The three configuarations
-------------------------

* is green (ground)
+ is red (signal)

1. Configuration, Er=1
Zo about 95 Ohm

400

100 +++++++++++++++++++++

0 *********************************************************
0 200 350 575

2. Configuration, Er=1
Z0 about 85 Ohm

400

100 +++++++++++++++++++++

**********************************************************
0 50 200 575

3. Configuration, Er=1, conductive wall at the left side
Z0 about 85 Ohm

400*
*
*
*
*
100* +++++++++++++++++++++
*
*
*
**********************************************************
0 50 200 575

Dr. David Kirkby
Guest

Posted: Tue Sep 23, 2003 6:01 am Post subject: Re: atlc - transmission line/coupler design sofware.

Helmut Sennewald wrote:

Quote:

"Dr. David Kirkby" <drkirkby@ntlworld.com> schrieb im Newsbeitrag
news:3F692A6B.DB254F71@ntlworld.com...
Some might be interested is some free (issued under the GNU General
Public Licencse) of mine for the anaysis and synthesis of transmission
lines and directional couplers (which are transmission lines anyway).

The source is available for those wishing to build it. The code is very
portable, so should run on any UNIX system.

Someone else has produced some Windoze binaries from the source code.

Anyway, take a look if interested.

http://atlc.sourceforge.net/

There are some nice pretty examples around. Improvements to
documentation or code are always welcolm
--

Hello David,
thanks for this great program.
I tested three structures with the same ratio W/H.
First, I placed the conductor about in the middle. Then I placed it
closer to the left side in the second example. In the third example
I added a wall on the left side of conductive material(green).

The results:
------------
Case 2 and 3 have shown 10 Ohm differencec compared to 1.
Is the left and right side assumed to be connected to the ground when
calculating the fields/impedance?

No, they are not forced to zero - there are no boundary conditions
applied.

Quote:

I have read something about truncated field in the manual.
What does it mean here?

The problem is that if the fields extend to +/- infinity, as they do in
these examples, atlc has some difficulty handling that situation. I've
not investigated this fully and perhaps there are ways to improve things
in these cases. I prefer to keep a solid wall around all sides, forcing
a set of well defined boundary conditions. If you find that moving the
position of those walls does not make much difference to the result,
then they must be sufficiently far away to be of no significance.

Quote:

By the way, the result of case 3 was nearly the same as in case 2.

Is there any rule of thumb how far I should keep away from the left
and right side?

The best advice I can give is to try it at some distance x, then 2x. If
there is little difference between the results, you are okay. In your
case you added a wall and found it made a significant difference

I would add that the amount of testing I have done in cases where the
fields are not fully contained is fairly small. One case that I have not
looked at is the case of two parallel wires (twin-wire cable). There is
an exact formula for this, which has the cosh() function in if I recall
correctly, although there is an approximate formula in many books - most
of which fail to make it clear that it is not an exact formula. I would
at some point like to investigate that in some detail.

The problem I have with atlc is testing it. As you will see, where I
have managed to test it, errors are very small (well under 1%), but
there are only a limited number of cases where I can get exact answers
to compare against. The twin-wire cable is one such case where it should
be possible to obtain an exact result with an open-structure and compare
it to atlc's results. This might give some more insight in how far to
put boundaries.

If anyone has some time and can test atlc against exact analytical
methods that I have not done, I'd appreciate it.

Quote:

Best Regards
Helmut

The three configuarations
-------------------------

* is green (ground)
+ is red (signal)

1. Configuration, Er=1
Zo about 95 Ohm

400

100 +++++++++++++++++++++

0 *********************************************************
0 200 350 575

2. Configuration, Er=1
Z0 about 85 Ohm

400

100 +++++++++++++++++++++

**********************************************************
0 50 200 575

3. Configuration, Er=1, conductive wall at the left side
Z0 about 85 Ohm

400*
*
*
*
*
100* +++++++++++++++++++++
*
*
*
**********************************************************
0 50 200 575

--
"The day Microsoft makes something that doesn't suck is probably
the day they start making vacuum cleaners." -Ernst Jan Plugge.

Dr. David Kirkby,
Senior Research Fellow,
Department of Medical Physics,
University College London,
11-20 Capper St, London, WC1E 6JA.
Website: http://www.medphys.ucl.ac.uk/~davek
Author of 'atlc' http://atlc.sourceforge.net/

Dr. David Kirkby
Guest

Posted: Tue Sep 23, 2003 6:17 am Post subject: Re: atlc - transmission line/coupler design sofware.

John Larkin wrote:

Quote:

So you are saying one can enter a lines like:
RECTANGLE 30 40 5 8 1.34 #Er=1.34
RECTANGLE 30 40 5 8 mauve # Er=1.34
RECTANGLE 30 40 5 8 milk #Er=1.34

and them all produce the same output ?? The problem is the more
combinations of data that are accepted, the more difficult (i.e. time
consuming) it gets to write a program to parse the file.

Quote:

And how about some snazzier program names? atlc is not too catchey,

The program has been around for more than 5 years, so I'm reluctant to
change its name now. If you stick "transmission line" in google, atlc is
the #2 entry. If you stick in "atlc" it is the #4 entry. I don't wish to
change its name.

Rather annoyingly, if you stick in "transmission lines" (note the extra
s), atlc is nowhere to be found

Quote:

and things like make_bmp and bitmap_name are hard to say or type. Have
consideration for us slow typists! I nominate "fling" as the new prog
name, and shorter verbs in the command file (bitmap, rect, etc.)

Why the hell fling? The letter x would be shorter still, but it does not
convey any information. At least create_bmp says what it does. I admit
the other program names are too long. But one can always create links to
other file names and call them something else if you want.

I'm happy to accept suggestions for a program to create bitmaps, but
fling does not seem very helpful to me. Perhaps cbmp.

--
"The day Microsoft makes something that doesn't suck is probably
the day they start making vacuum cleaners." -Ernst Jan Plugge.

Dr. David Kirkby,
Senior Research Fellow,
Department of Medical Physics,
University College London,
11-20 Capper St, London, WC1E 6JA.
Website: http://www.medphys.ucl.ac.uk/~davek
Author of 'atlc' http://atlc.sourceforge.net/

Dr. David Kirkby
Guest

Posted: Tue Sep 23, 2003 6:58 am Post subject: Re: atlc - transmission line/coupler design sofware.

"Dr. David Kirkby" wrote:

Quote:

I think perhaps insisting the user put a permittivity, material, hex
representation and a descriptive colour in a configureation file would
be best:

i.e. atlc.cfg looks like this:

1.34 milk 0x234334 redish-brown

with perhaps comments allowed, as the 5th and subsequent fileds are
always ignored.

The problem is then how to read get create_bmp to read it? Does it
expect to read a float (1.34), a material (milk), a hex representation
(0x234334) or a descriptive colour (redish-brown)? The more flexible I
make it, the easier it is to use, but the harder it is to program. There
are constraints on my time.

If someone wants to write a program that can parse such a file and
produce the same hex colour no matter if the user enters 1.34, milk,
0x234334 or redish-brown, I would not object to their input. If they
would write the program such that is reads the command file with each of
the following being equivalent, all the better.

RECTANGLE 30 40 5 8 1.34
RECTANGLE 30 40 5 8 milk
RECTANGLE 30 40 5 8 0x234334
RECTANGLE 30 40 5 8 redish-brown

But that is a lot of work. Better still if someone can write the program
fully, so it converts the dimensions to a best fit grid, then writes the
bitmap, all the better. I'll probably eventually get around to doing the
whole thing, but some input from othes would speed the process.

I would personally find testing atlc with open wire feeders, or any
other stucture that have not been tested but for which an analytical
result is known, of far more interest.

Should I allow them to choose any colour they want, or force them to use
one of 10 colours I predefine, but for which they can use how they want?

Another thing someone might like to do is find me a list of say 10-20
colours, giving me the hex representation, that look as dissimilar as
possible from each other and the colours currently used:

http://atlc.sourceforge.net/colours.html

I chose those colours (apart from red, green, blue and white) is a
pretty arbitrary fashion. If I do force peole to use one of a number of
pre-defined colours, I need to find some that look dissimilar on both
screen and printed paper.

I think forcing people to use any of say 10-20 colours would be a good
idea, as it will save them hunting around to find something suitable.
The chances of anyone needing any more than that are very remote.
--
"The day Microsoft makes something that doesn't suck is probably
the day they start making vacuum cleaners." -Ernst Jan Plugge.

Dr. David Kirkby,
Senior Research Fellow,
Department of Medical Physics,
University College London,
11-20 Capper St, London, WC1E 6JA.
Website: http://www.medphys.ucl.ac.uk/~davek
Author of 'atlc' http://atlc.sourceforge.net/

Display posts from previous:

All About Circuits - Newsgroup Archive Forum Index -> sci.electronics.design	All times are GMT Goto page 1, 2 Next
Page 1 of 2

You can post new topics in this forum
You can reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum

sitemap