Patch antenna design

[bmnet]

Hi guys,
I was recently given a project to design a broadband patch antenna with a center frequency of 900MHz. Here are the specs that I used for my desing..

1. Center Frequency Fc=900Mhz
2. Er (Dielectric Constant)=2.2 (PCB Board)
3. Substrate Height = 1.5mm
Calculated dimensions were w=110mm & l=97mm.
Did some simulation using Ensemble Design
After fabrication, I plugged it into a network analyser and was just surprised that the only place where I got a decent return loss (-12dB)was at 1500MHz! I've looked through everything and I can't seem to figure this one out. Any thoughts would be highly appreciated.
Meant to use parasitic elements to broadband it.

 

[Higgler]

You dimensions sound ok for 900 Mhz.
Is this a standard half wave resonant patch antenna?
Your coax. probe from underneath? It should be located just off the center line, probably 5% out from the center towards the edge. Did you use a connector on the bottom? and cut away some copper on the base so the center conductor doesn't short to the bottom ground plane?

What size ground plane did you put underneath it?

kch
PS: What's your bandwidth? did know know that 2x thickness of substrate = 2x the bandwidth of the antenna (thicker substrate needs the coax. closer to the outside edge of the patch too).

 

[bmnet]

Yes, Higgler, its 1/2 wave resonant antenna. I used the simulator to find the 50Ohm point in the patch for the coax probe, lacated about 15% off the center.. Got a BJC (gold coated) connector at the bottom, then even checked for shorts...
The ground plane dimensions are 300x300mm. The substrate thickness was about 2mm.
Was aiming for a 80% bandwidth, but with the aid of the parasitic elements.

Thanks
Bmnet

 

[Higgler]

Sounds ok except for the 80% bandwidth part. I've done alot of patch work, manufactured about 6 different designs the past few years and have 100 articles on them from various sources. Bandwidths in the 2-30% are common, but I don't think I've seen 80% bandwidth before, maybe a max. of 50%. Your thickness substrate (60 mils) seems way too thin to achieve large bandwidth, I'd expect 10-20 Mhz bandwidth on your antenna. Did you mean 80 Mhz bandwidth for the antenna?
Three bandwidth examples of mine;
1) at 3.5 times your (0.2 inches) and 8 Ghz frequency (11x frequency) we achieved 20% bandwidth for a thick patch which required the probe to be attached on the edge of the antenna.
2) at 1.33x your thickness for GPS which is 1.575 Ghz, we achieved only 30 Mhz bandwidth (1.9%)
3) 433 Mhz quarter wave patch (2"x4" size) using 60 mil FR4 had 20 Mhz bandwidth (2:1 VSWR point).

Antenna bandwidth depends on the volume of your antenna per the laws of physics (except for Marketing, Sales and Management which break the laws of physics daily), i.e. LengthxWidthxHeight sets the bandwidth of the antenna, and there is no getting around that unless you own negative dielectric constant maybe.

kchiggins

http://www.toyon.com

 

[savedadogs]

Maybe your Ensemble simulation is incorrect? If you have Ansoft HFSS you can simulate your patch there. Here is a tutorial on a patch antenna using HFSS:

http://www.emtalk.com/tut_1.htm

and the measured versus hfss results are really good:

http://www.emtalk.com/forums/viewtopic.php?t=7

Enjoy.

 

[bmnet]

Thanks Higgler. Sorry about the bandwidth, needed it to be as wide as possible ( about 200MHz at the least). I guess what puzzles me the most is the failure of the antenna to operate at its design frequency...
As for the simulation Savedadogs, it did agree with theoretical design calculations I did initially.
Currently am suspecting the coax connector I've used is 75ohm.
Will check and keep you posted. Unfortunately I don't have another simulation software..

 

[Higgler]

For that bandwidth, your patch needs to be 2" thick with an air dielectric spacer to get your bandwidth and the probe needs to be very near the outside edge. You'll need a coax. feed pin with the outer coax. running 3/4 of the way up the 2" thickness to match it.

I don't think the 75 ohm impedance would be the reason for no resonance at 900 Mhz. Make sure your dielectric is correct.

If all else fails, build one extra long with your probe in the middle, then start hand trimming each edge for match and frequency.

kch

 

[bmnet]

Thanks Higgelr... Your last point's looking interesting by the minute (even though its crude)...

 

[Higgler]

That trimming works well and you can get perfect VSWR.

A point of interest when trimming, I made a 433 Mhz patch that had fingers on the end that were used for frequency trimming. I found that I didn't have to remove the fingers of copper to achieve a frequency shift, only making a cut in them was necessary. Each finger was 4 Mhz of tuning, and the difference between just a cut and removing the entire copper finger was maybe 0.1 Mhz. So if you trim all the way across the patch to remove metal and you get really close to your frequency, you can just cut a corner to raise the frequency slightly instead of removing a sliver of copper all the way across the patch.
The really thick patch may have some trimming challenges, so good luck if you try that one.

kch

http://www.toyon.com

 

[bmnet]

Got up this morning and had a crazy idea... Before going on the trimming idea, I decided to shift the feed point... After about 15 holes of trial and error, I got the frequency down from 1.5GHz to 1GHz!! Think I should be happy with that (considering I didn't even callibrate the network analyser!)
It brings up another question though: Did I manage to change the dielectric constant of the material or the 15holes (x 1.3mm) have got no effect?

Will still try the trimming just for future reference.
Thanks heaps Higgler.

Bmnet.

 

[Higgler]

Poking holes in a dielectric lowers the dielectric constant and should raise the frequency. I read that 2.2 dielectric can be lowered to 1.2 that way. So there's something odd.
If things don't make sense, often just starting from scratch can shed some light on the subject. Hence, build another one and it might work fine.
I just calculated 97 mm gives 1.54 Ghz half wave frequency in air, maybe just a coincidence. Your rtn loss should be at least -10 dB at resonance. Suggest making your patch length x width about a 2:1 ratio just for safety sake if you retry. More rectangular patches also have better crosspole purity which is often desired.

kch

http://www.toyon.com

 

[bmnet]

Higgler, the dielectric constant in air makes the difference... I did start again but made L:W ratio about 1:1.5.. Not bad now. Just got 50MHz off target.

Thanks guys for all the help.

Bmnet

 

[Higgler]

Now you're a full size half wave patch antenna guru.

If you want it twice as short, go to the exact center of the patch, cut it in half, (just the top part, leave the ground full size) and short the top of the patch to the bottom of the patch all the way across it with metal tape or solder metal across (you can also just drill holes and put screws from top to bottom, use 5 or more evenly spaced). The frequency will stay approx. the same, you're size is halved, but your bandwidth is cut in half and the patterns become slightly asymmetrical (higher gain towards the open end) but now you'll have a quarter wave shorted patch antenna.
Many receivers are sensitive to static blowing them out. Having your center conductor of the antenna shorted prevents any static from getting into your receiver.

kchiggins

http://www.toyon.com