Patch design problem

[Diom1982]

Hi,
I am trying to design my first patch antenna, but I am having a lot of trouble getting the resonant frequency to where I want it (at 2.45GHz). I have a folrmula from a book "Compact and broadband microstrip antennas" by Kin-Lu Wong, which tells me that the fundamental frequency of the patch should be:

f = c/2*L*root(epsilon-r)

I am solving this for L (epsilon-r = 2.2, just taking the dielectric constant of the substrate material...is that wrong?) and I get a value of 20.62mm. But then I get a wierd S11.

I also don't know how to select the width (W), or the thickness of the substrate (T).

I was wondering if anyone here could point me to a site that just runs trough how to design a patch antenna for an arbitrary frequency? Of if someone spots any major flaws in what I have described so far please point it out. I am using HFSS.

Thanks for reading this.
D.

 

[Diom1982]

Hi...sorry about that I was searching the forum incorrectly, D'OH!
will now look through the answers on subjects like this...

If you want to answer anyways please do.

 

[VEBill]

Sweep the prototype, find the sweet spot (it should be reasonably close), and then scale your design accordingly. Repeat if required. Should be just about perfect no later than the third prototype.

Disclaimer - I've never made a patch, but this test/adjust routine is pretty normal for any antenna prototype. Almost every HF dipole or VHF Yagi is put up and taken down several times during the initial measure and adjust phase.

 

[Diom1982]

Thanks,

I'm getting some very strange problems.

I'm using a square patch of side 41.248mm on a substrate with dielctric constant = 2.2. According to my calculations that should give me a fundamental frequency of 2.45GHz.

My S11 for this shows (simulateed) that I have a resonance at about that fequency but it is very weak, with a second one at about 3.25GHz being stronger and a third resonance at 4.7GHz being the strongest (-15dB). How do I make the resonance deeper at the 2.45GHz frequency and get rid of the others?

 

[VEBill]

"...square patch..."

4.7/3.25 = 1.4 = Sqrt(2) Hmmm....interesting...


There are attendees here that know much more abot Patches than I ever will, so standby for more-informed replies.

 

[Higgler]

Diom1982,
you did the math right, that's a good size.
How thick is your material and how did you feed it? microstrip or coax. connector or coax cable from underneath?

A few things to guarantee better results.
1) most people avoid square patches unless you have two feed points and create circular polarization. Rectangular patches are more common for polarization purity reasons, square will work, but you must get your feed position perfectly along one axis.
2) the feed position should be close to the center of the patch if the thickness is small (less than 0.05 wavelengths thich) and the feed position should be close to the edge if the patch is thick (0.3 wavelengths). With a thin patch that puts the feed location near the center of the patch, there is a big question of accuracy of placement of this feed point, especially if you hand cut the circuit, plus drilling a hole for a center conductor when using a coax. feed is not very precise (+/- 0.020" maybe). Hence you may have a real tolerance headache in setting your feed location. Plus you dielectric value of rogers 5870 is just over one percent, that'll change your frequency a bit. I suspect those other frequencies, which aren't multiples of 2.45 are due to your square shape and feed placement. Make it rectangular and thicken it up if possible to move the feed probe out a bit from center. Moving the feed from the center to the outer edge is the method to improve S11 from poor to optimum. Empirically, you can do that by cutting the circuit first on one side, then on the other, I'll leave it to you to figure out the steps, hint - start with a patch size about 80mm, add a probe (coax.) from underneath, then start cutting.

3) groundplane size. If you just printed a patch as one piece of dielectric 41.248 mm square and didn't mount it on a groundplane (at least 92x92 mm size, preferably 200 mm size) then you won't get very good S11 without fine tuning, then the performance will change depending on length of cable you attach to the unit. Gain varies with groundplane size too. Need about 4 lambda square to optimize your gain if you need higher gain. typically 1 lambda ground plane = 1 dBi gain, 4 lambda groundplane = 4 dBi gain - that's ballpark rule of thumb for Er=2 patch.

add some more detail please and don't worry, it'll turn out perfectly in a one or two more tries. You won't ever forget what doesn't work - the curse and hidden blessing of Engineering. I know all my work errors vividly the past 27 years.

kch