design_coupler - for designing directional couplers (part of the atlc package)

design_coupler [-C][-d][-e][-H height][-L length][-q]

[s fstep][-Z Zo] CF fmin fmax

This man page is not a complete set of documentation - the complexity of the atlc project makes man pages not an ideal way to document it, although out of completeness, man pages are produced. The best documentation that was current at the time this version was produced should be found on your hard drive, usually at

/usr/local/share/atlc/docs/html-docs/index.html

although it might be elsewhere if your system administrator chose to install the package elsewhere. Sometimes, errors are corrected in the documentation and placed at http://atlc.sourceforge.net/ before a new release of atlc is released. Please, if you notice a problem with the documentation - even spelling errors and typos, please let me know.

design_coupler is used to design directional couplers. It it not used to analyse couplers for which you know the dimensions. Instead, it is used but when you require a coupler to have specific properties, but don't know the required odd and even mode impedances or the required physical dimensions that will achieve those required properties.

As a minimum the user must specify the coupling factor CF in dB, the minimum frequency fmin in MHz and the maximum frequency fmax in MHz. With this information, the design_coupler will

a) Tell you the required odd and even mode impedances Zodd and Zeven assuming the coupler is for 50 Ohms and assuming the coupler is is a quarter wave long, which might be an impractical length. There a numerous ways of making a coupler having those impedances and design_coupler does not (without the addition of options mentioned later), tell you how to make such a coupler. b) Given you the frequency response of the coupler, making the assumptions about the 50 Ohm impedance and quarter-wave length. The frequency response is calculated at 5 points in the range specified by fmin and fmax.

By use of the -Z 'Zo' and -L 'length' and -f 'fstep' options it it posible to specify different a different characteristic impedance, length and different frequency steps to display the frequency response.

The computed values of Zodd and Zeven required are valid no matter how the coupler is design physically. So no matter whether it's implemented on a PCB, air spaced or whatever, the above impedances are correct and the frequency response is correct.

The -d option causes design_coupler to not only report the required odd and even modem impedances but also the physical dimensions of a coupler that achieves these properties! Currently, the only stucture for which it is possible to compute the physical dimentions is two wide edge-coupled striplines between two wide plates like this:

----------------------------------------------------- ^

| | |

| Er | |

| | |

| ----------- ----------- | H

| <----w----><--s--><----w----> | |

| | |

| | |

| | |

----------------------------------------------------- v

<-------------------------W------------------------->

The width W must be much greater than the height of the coupler and generally it is assumed that this width will at least 2*w+s*5*H, otherwise the calculations will be incorrect. In order to calculate these dimenisions an analytical method is used, which is only valid if the width W is infinity, but should be resonably good assuming W is at least 2*w+s+5*H.

It is later intended to enable design coupler to use other structures, which migth be more suitable for construction, such as microstrip couplers on PCBs, but for now at least, it is only possible to compute the physical dimensions of the coupler using the above stucture. For strong coupling (less than 20 dB or so), the dimenions calculated might be impractical, as the spacing s will be so small. However, for weak coupling, the physcical dimensions are practical.

-C

print copyright, licensing and copying information.

-d

Design a coupler, using two edgle-coupled stiplines inside a wide 4-sided rectangular enclosure.

-e

Priont an example of how to use design_coupler

-H height

Specify the height of the enclosure in some convenient unit. By default, a height of 1 unit is assumed, but by use of this option it is possible to specify any height you want. Since its the ratio of dimensions that is important, not the absolute values, this just scales all the other dimensions by the specified height. It is just a conveneince for the user.

-L length

Specifies the coupler length in metres. By default the coupler is assumed to be a quarter-wave, but this allow any length you want. Don't chose a length that is a multiple of a half-wave though, as this will make it impossible to couple any power out. -q

This is the 'quite' switch and causes design_coupler to print out less information. One can use -qq to cause the even less output.

-s fstep Causes design_couler to print out the frequency response at different steps from the default 5 values. fstep must be in MHz. The default value of fstep is obviously (fmax-fman)/5.

Z Zo

Causes design_coupler to compute properties of an impedance Zo (shecified in Ohms). The default value for Zo is 50 Ohms.

Run design_coupler gives examples of its use. However, here are those same examples.

Here are a examples of how to use design_coupler In the examples, the % sign is used in front of anything you must type which is what you will probably see when using the csh or tcsh as a shell. It would probably be a $ sign if using the sh or bash shell.

To find the odd and even mode impedances and frequency response of a 50 Ohm coupler, covering 130 to 170 MHz, with a coupling coefficient of 30 dB:

% design_coupler 30 130 170

Note the frequency response is symmetrical about the centre frequency at 0.192 dB below that wanted. You may wish to redesign this for a coupling coefficient of about 29.9 dB, so the maximum deviation from the ideal 30.0 dB never exceeds 0.1 dB Note the length suggested is 0.5 m (nearly 20") is a quarter wave at the centre frequency of 150 MHz. You might find this a bit too long, so let's specify a length of 0.25 m.

% design_coupler -L 0.25 30 130 170

What you may notice is that while the coupling to the coupled port is exactly 30 dB below the input power at the centre frequency (150 MHz) it is no longer symmetrical about the centre frequency. Also, deviations from the ideal 30 dB are now much larger, with a maximum error of 1.012 dB Unlike the case when the length is the default quarter wave, there is not much you can do about this, since the deviations occur in both directions.

Now assume you are reasonably happy with the response when the length is 250 mm but would like to see the response at every 2.5 MHz. This can be done using the -s option to design_coupler.

% design_coupler -L 0.25 -s 2.5 30 130 170

Assuming the performance is acceptable, the dimensions of the coupler can be determined by adding the -d option. This will design a coupler that must look like the structure below. The two inner conductors, which are spaced equally between the top and bottom edges of the outer conductor, must be very thin. These are placed along the length of a box of width W, height H and of a length L determined by the user, which in this case is 250 mm.

----------------------------------------------------- ^

| | |

| Er | |

| | |

| ----------- ----------- | H

| <----w----><--s--><----w----> | |

| | |

| | |

| | |

----------------------------------------------------- v

<-------------------------W------------------------->

The program reports: H = 1.0, ; w = 1.44 ; s = 0.44 The height of the box H must be small compared to the length L, (perhaps no more than 7% of the length), or 17.5 mm in this case, with a length of 250 mm, otherwise fringing effects will be significant. The width of the structure W should be as large as possible. The program suggests making this 5*H+2*w+s. The 7% and 5*H+2*w+s are educated guesses, rather than exact figures. There is no problem in making the width larger than 5*H+2*w+s. The length L must be kept at 250 mm. The RATIO of the dimensions H, w and s (but not L or W must be kept constant. W just needs to be sufficiently large - it is uncritical.

If you happened to have some 15 mm square brass available, then using that for the side-walls would require that H becomes 15*1.0 = 15 mm, w = 15*1.44 = 21.6 mm and s = 15*0.44 = 6.6 mm

There is no need to compute the above scaling with a calculator, as using The -H option allows one to specify the height H. The program then reports the exact dimensions for the length L, height H, w, s and suggests a minimum width for W.

In summary we have:

30 dB coupler +1.02 dB / -0.78 dB for 130 to 170 MHz

Length L = 250 mm, height H = 15 mm, stripline spacing s = 6.3 mm

stripline width w = 21.6 mm enclosure width W >= 124 mm

By default, design_coupler prints a lot of information to the screen. This can be reduced by the -q option or reduced to only one line with -qq Other options include -Z to change the impedance from the default 50 Ohms and -C to see the fully copyright, Licensing and distribution information

No files are created at all.

atlc(1)

create_bmp_for_circ_in_circ(1) create_bmp_for_circ_in_rect(1) create_bmp_for_microstrip_coupler(1) create_bmp_for_rect_cen_in_rect(1) create_bmp_for_rect_cen_in_rect_coupler(1) create_bmp_for_rect_in_circ(1) create_bmp_for_rect_in_rect(1) create_bmp_for_stripline_coupler(1) create_bmp_for_symmetrical_stripline(1) find_optimal_dimensions_for_microstrip_coupler(1)

readbin(1)

http://atlc.sourceforge.net - Home page

http://sourceforge.net/projects/atlc - Download area

atlc-X.Y.Z/docs/html-docs/index.html - HTML docs

atlc-X.Y.Z/docs/qex-december-1996/design_coupler.pdf - theory paper

atlc-X.Y.Z/examples - examples