Design of microstrip to give multiple phase delays at any f from 10 GHz to 14 GHz

I want to produce phase delay of 45 ,90,135, 180,225,270,315,360 like that at 10 GHz or 12 GHz. Sub_h=1.6, Er=4.4, Width of microstrip=0.7. I calculate the line lenght using Online Microstrip Line calculator. But it didn't work out!! i wonder what is wrong with my simulation. Anyone can guide me on this and also what element should i put to know whether phase shift is produced or not in HFSS. A sample of drawing is attached below in which each element is of l=4.32mm, W=0.7mm at 8 GHz. Looking for kind response urgently!

- - - Updated - - -

image is attached in this link!! http://obrazki.elektroda.pl/7571903200_1427850694.jpeg

- - - Updated - - -

a little modification L= 5.4 a at 8 Ghz and 4.32 is at 10 GHz.

Try to use this calculator: **broken link removed**

er=4.4
h=1.6
t=35
f=10000

output is W=2.7mm (50 ohm line)
quarterwave = 3.9mm

quarterwave is 360/4=90 deg

so your element must be half of quarterwave 3.9mm/2 = around 2 mm

Please tell about your interesting project. I am very curious why you need multiple phase delays. Is it for data transmission? Also, do you need all this signals simultaneous? In such case you can use microstrip tee.

Why did you choose 1.6mm substrate? Usually projects for 10GHz use 1mm FR4 substrate. But i personally prefer 1.6mm because it gives more radiation and bandwidth, also easier to make high impedance lines. Drawbacks is too long "vias" that may affect many things, but still some can be solved with high impedance lines.

term, you do not have a prayer of using an FR4 substrate at 10 ghz! just sayin

rizwan183 said:

Sub_h=1.6, Er=4.4, Width of microstrip=0.7.

Click to expand...

If you want 50 Ohm line impedance, 0.7mm is the wrong width.

50 Ohm line width on 1.6mm FR4 is around 2.7mm, which is too wide to be useful for most 10GHz circuit design.
And you don't want to use FR4 at 10GHz because it is very lossy. Use something like Rogers RO4003 or RO4350.
**broken link removed**

Thank You guys for your kind suggestions. i tried many cases and may be using FR4 to produce phase shift is a bad idea. i guess W of microstrip does not matter for the case to give phase shift theoretically!! Sorry i could not present more information actually L=4.3 so that each element give a phase delay of 90 degrees and i am using this calculator.
https://www.emtalk.com/mscalc.php
i tried many versions of the designs none of them worked.
@term am working on smart antennas.
@volk Is using RO4003 or RO4350 make a difference on simulation??

Could i use port at the end of each element with 0 amplitude and plot the result with give the clear idea about the phase difference between the start and end of each element in HFSS. or any otherway possible??

Usually you define ports like 1,2,3 and so on. There must be some simulation that gives S-parameters between all ports as output. S11, S12, S21, S22, S13,S31, etc. Maybe you defined wrong port type. I do not see any ports on your image. Finally you can draw graphs of S-parameter phase in degrees and see if it is correct. Also you can draw S-parameter amplitude.

rizwan183 said:

i guess W of microstrip does not matter for the case to give phase shift theoretically!!

Click to expand...

Not for the phase, but for input matching. You to use this for series feed of multiple antennas, and looking only at the line phase is enough enough to make that work.

rizwan183 said:

Could i use port at the end of each element with 0 amplitude and plot the result with give the clear idea about the phase difference between the start and end of each element in HFSS. or any otherway possible??

Click to expand...

It seems that you have placed many ports on the line. How did you design the matching for all these ports?

rizwan183 said:

@volk Is using RO4003 or RO4350 make a difference on simulation??

Click to expand...

Yes. But I think you should first read about antenna arrays with series feed, and the alternative of using parallel feed.

- - - Updated - - -

volker@muehlhaus said:

Not for the phase, but for input matching. You to use this for series feed of multiple antennas, and looking only at the line phase is enough enough to make that work.

Click to expand...

I can't edit my post, so here is the correct statement:

You want to use this for series feed of multiple antennas, and looking only at the line phase is NOT enough enough to make that work.

Dear fellows thanx alot for all you suggestions.

50 Ohm line width on 1.6mm FR4 is around 2.7mm, which is too wide to be useful for most 10GHz circuit design

Click to expand...

Line Width is around 3.55mm with RO4003 or RO4350 which is also to wide to be used at 10 GHz. decreasinh substrate height and increasing Er helps to decrease line width. should i use higher Er material and upto which i can decrease the height of substrate. and what possible side effects could be of these?

But I think you should first read about antenna arrays with series feed, and the alternative of using parallel feed.

Click to expand...

Could you suggest some good refernce material to study about the series and parallel feeding network?

@term can you suggest some good resources about microstrip tee. Also have i don't know what to plot against what to see phase difference. i tried mag(S(p2,p1)) vs ang_deg(Z(p2,p1)).but the plot was weired. probably HFSS is not good to show this. may be i need to use another software like ADS.

rizwan183 said:

Line Width is around 3.55mm with RO4003 or RO4350 which is also to wide to be used at 10 GHz. decreasinh substrate height and increasing Er helps to decrease line width. should i use higher Er material and upto which i can decrease the height of substrate. and what possible side effects could be of these?

Click to expand...

Yes, you need to use a thinner substrate for your circuit design. The large thickness is typically used for antennas.

The "best" thickness for your design depends on the line impedances that you need. If you go to very thin substrates and have couplers with small line width and narrow gaps, there might be a problem with manufacturing precision (etching tolerances, minimum allowed width/gap). A typical substrate thickness for 10GHz circuit designs is 0.5mm.

rizwan183 said:

Could you suggest some good refernce material to study about the series and parallel feeding network?

Click to expand...

You should find good references if you start with an internet search engine, with the above keywords. Antenna books will at least cover the parallel feeding networks that are typically used in antenna arrays.

rizwan183 said:

i tried mag(S(p2,p1)) vs ang_deg(Z(p2,p1)).but the plot was weired. probably HFSS is not good to show this. may be i need to use another software like ADS.

Click to expand...

I think that your setup (many ports on one line, ignoring impedance matching) isn't correct, and your plot definition isn't correct (why plot S21 magnitude vs. Z21 phase???) , so the problem is not the software. Investigate how others build antenna array feed networks.

rizwan183, give more details about what do you want from antenna. Start from easier task, draw single microstrip line and define two ports, plot s21 angle at frequency you want. Nowdays series fed patch arrays with tapered feed line is popular, as it gives low side lobes. You can divide power using loseless power divider with matching: https://www.ittc.ku.edu/~jstiles/723/handouts/section_7_2_The_T_Junction_Power_Divider_package.pdf . Parallel and series networks all can be done with simple formula. For parallel feedback you have division ratio 1. In 50 ohms system using formula from this pdf 50 ohm line divided to two 100 ohms. After quarterwave matching it gives around 70ohm. Just google for patch array images, some of them have impedance written on lines. For series fed arrays the same formula is used. But division ratio is not 1, because we do not need equal division. Some portion goes to the patch, and some remains in series feed line going to next patches. Formula produces tapered line. Or there also one type where feeding line goes "through" all patches (lambda/2 line then patch then lambda/2 line and so on). The same formula of t-junction with matching gives sqrt(2) in wilkinson power divider, if division ratio is 1. Anyways give some more information, for example you can tell similar design you want.

Terminator3, sorry for late reply!! Image in the following is somwhat similar to what i am doing.
https://obrazki.elektroda.pl/7516681700_1428219248.jpg

Samething i tried with some modification with substrate=RO4350, sub_h=1.4mm,Er=3.66, 45 deg lenght = 2.16mm length of microstrip approx. Each consecutive port is longer by 2..16 mm than thhe previous port. in this way phase difference should be 45 degree. but results are strange to me at f = 10 GHz.links are as follows
https://obrazki.elektroda.pl/5213469000_1428219249.jpg
https://obrazki.elektroda.pl/2239877100_1428219249.jpg
https://obrazki.elektroda.pl/6040710200_1428219250.jpg
https://obrazki.elektroda.pl/5059591400_1428219251.jpg

Doesn't look like phase difference plots. Isn't S(portx,porty) the transfer function from porty to portx?

rizwan183 said:

https://obrazki.elektroda.pl/5213469000_1428219249.jpg

Click to expand...

You have different length in the line segment that does the impedance transformation. That is wrong.

The length in the power splitter must be 1/4 wavelength with 50*sqrt(2) Ohm impedance. If you want to add length, that extra length must be 50 Ohm.

Try to start from simple array, where all patches have same phase:
look at this image: **broken link removed** (if you can't open it, try this webpage and choose advertisement of 2x2 patch array **broken link removed** ).
Assuming array input is 50 Ohms (wide line):
wide line = 50 Ohms
thin line = 70 Ohms - only this line must be quarterwave
very thin line = 100 Ohms
You can alter length of 50 Ohms and 100 Ohms lines as you like to get desired phase. In this case lengths seems chosen only to get proper patch distances.

Earlier i gave you a link to PDF. Look at page 4 and 5. In this case Z01=50Ohm. Division ratio a=1.
So Z02=Z01*(1+1/a)=Z01*(1+1/1)=2*Z01
Z03=Z01*(1+a)=Z01*(1+1)=2*Z01.
We get Z02=Z03=2*50=100 [Ohms]

As you can see, 50Ohm lines on the 2x2 patch array picture is first divided to two 100 Ohm lines.
Then each 100 Ohm line is matched to short portion of 50 Ohm line with 70 Ohm quarterwave line (page 7 in PDF above). Then it again divided to two 100 Ohm lines (upper part of image) and similar for the bottom part of image. Then there is no any additional matching to 50Ohm. 100Ohm lines directly goes to the patch. It seems that patch imput impedance was very high. Image you gave is very similar, but 100 Ohm lines length is altered to get some phase shifts. It is very important to know imput impedance of patch, there are some formulas at the bottom of this page: **broken link removed** Although i still in search of best way to estimate imput impedance of patch, because using different formulas and antenna calculator gived various results from 200 to 400 Ohm.

Thank you guys for helping me and i am able to design matched transmission in almost no time using quarter Wave Transformer equatio[Zin= Square(Z1)/Zload]. I found many new things and information from these posts. Infact reading is solution to everything. But to do things shortly and study what is required is best found out using Forums. Special Thanx to Terminator3 for taking so much time to reply and also volker@muehlhaus .