Pcb strip feed to pcb mounted ceramic patch (GPS)

[robroy1]

Hello;

I am a new dog sniffing around for a scent of where to go on this subject....yes, this is in the line of duty rather than pure speculative fun..!

I am using a ceramic patch antenna, nice Q at 1575.42 L1 GPS frequency, return loss of 12-15dB (coaxially fitted for test), and I desire to make it mounted on a pcb, with the pcb transporting a coaxial termination from the edge (U.FL) of the pcb to the feed point of the ceramic patch. The patch antenna 15mm x 15mm is a purchased antenna which has a metallized surface on the bottom which I am soldering to the pcb, and a metallic feed rod that extends from the bottom of the antenna ~2mm (and connects to the element on the top of the ceramic antenna) which is plenty to poke through a fr4 pcb and be soldered to a microstrip comprised of pcb traces. Why would I want to do this crazy thing????, well I need a ground plane and the only way is to create one...I can use a pcb or I can use laser trimmed metal and pcbs are much cheaper and easier to redesign...or so I think at this moment.

Knowing little, and as I heard a rapper say "low budget high stakes gamble" (I dont listen to rap, but it was on when my wife was spinning the radio dial....or button) I am concerned that while I may be able to calculate the correct impedance for the pcb trace so that the impedance is what I think it should be (of which my ignorance rears its ugly head when I cast the ubiquitous 50 Ohms from my mouth and heart) it is unlikely that I will be able to correctly design this strip without additional information. First I really do not know what the "real" impedance of the antenna is and I dont have the equipment to measure it (assuming that I would be able to configure the measurement setup properly even if I had the equipment). I have asked the manufacturer for the information but I am waiting on the sales representative for that information. Secondly, in my imagination I think that I should design the pcb microstrip with tuning elements at least at the antenna end to provide a means to match the antenna. Can this be done with a simple PI network and can the components be pcb traces if the pcb manufacturing tolerances are achievable? I know that etching is a problem with pcb microstrips but I dont know what or where the limits are...of course in my dreams this was not a problem....! And would a PI network be the worst choice for a narrowband antenna?

Some additional details...Once coaxial, it then terminates on the GPS engine which on my inspection looks like a matching network feeding a Agilent ALM1106 amplifier. Coaxial feedlength expected to be no longer than three inches tops.

To sum it up from the ramblings:

PCB microstrip with tuning elements comprised of pcb features in copper and absense of copper, is this something that is doable and realistic?

What issues will I present myself by terminating coaxial transmission line to the pcb which then is soldered to the ceramic patch?

Am I headed of a cliff without a hanglider attached?

Cheers.

Rob

 

[Comcokid]

FR4 materials are frequently used on RF for cost sensitive products. I've designed 900 and 2.4 GHz transceivers on double sided FR4 - sometimes a challenge to keep one side fully ground plane and route everything single sided and impedance matched on one side with a tight footprint.

I've seen websites for antenna manufacturers where they've got a combo antenna - GPS and WiFi or Cellular - where a ceramic dielectric GPS antenna was located on (what appeared to be) FR4 board with the antenna-etch antenna pattern for the other frequencies.

To do a 50 Ohm microstrip on 0.060 FR4, the trace width will be about 105 mils. For 0.030 FR4 width will be about 52 mils. I say 'about'. There are lots of microstrip calculators available on the internet, and they will all vary slightly in what they calculate. This variation will be less that what is possible from the dielectric constant tolerances of the FR4 itself.

FR4 is not controlled for impedance - it can vary lot to lot and manufacturer to manufacturer. To tie this down some, specify a manufacturer for the PCB material. Check the manufacturers data sheet, and get their dielectric constant spec - at a frequency near the GPS. Most manufacturers specify at 1 MHz - don't use it for your calculations. Find or request the spec for 2 or 2.4 GHz because the dielectric constant will be less at the higher frequency.

 

[robroy1]

Thanks Comcokid!

Right off I need to locate the specific dialectric constant for the vendor that I will be using. Very good direction to start me down the road to something workable..

Does the transmission velocity change dramatically with frequency?

Regarding the challenge to route everything single sided and impedance matched is another challenge here:

I need to pick up the feed from the patch on the opposite side from the patch antenna, then transfer the trace to the top side at an outer edge so that I can use a U.FL connector. The transitioning of the stripline from one side to the other I have not found any information to help guide me....any ideas where I should be poking my nose for answers?

I really appreciate the people that share their knowledge here, I think they are a higher caliber thinkers..!

 

[Higgler]

The real antenna impedance will be cery close to 50 ohms in your feeding configuration.

The inductance of that feed probe can mistune your antenna if it's too long, i.e. if your circuit board is too thick your antenna efficiency/VSWR might degrade due to an added input inductance of that center conductor.

You also have to etch back a ground plane side hole in your FR4, don't just drill a thru hole in your FR4 and stick the probe thru it. Remove some copper around your ground plane so the metal on the patch antenna sets/(is responsible for setting) the antenna connection outer conductor diameter.

I think overall it won't be too bad. It's done alot just the way you plan, without much added fanfare.

Most antennas like to be on a ground plane half wave square or larger. The patch can shift frequency with smaller ground planes.

Good luck,

kch

 

[robroy1]

Thanks Higgler!

>> The patch can shift frequency with smaller ground planes.

I dont understand the mechanism that would cause the detuning, and which way would the patch detune?

I thought that the ground plane essentially provides the beamwidth/directivity of the antenna not the tuning. The patch by itself has metallization on the bottom of the ceramic which I use to solder the antenna to the pcb. I would have guessed that perhaps other types of capacitive loading in the front hemisphere of the antenna would detune the antenna but I would not have guessed that the ground plane reduced area would detune it...

What about then irregular shaped ground plane around the antenna? Is it possible to then retune the patch by using ground plane smooth curve shapes rather than rectangular ground plane?

Cheers...Rob

 

[Higgler]

>> The patch can shift frequency with smaller ground planes,

The shifting is often small, so it depends on your bandwidth of the antenna and groundplane size as to whether it affects results. Wide band antennas are friendlier to changing surroundings.

The mechanism is; antennas push current that runs along the groundplane, current/electrons get to the edge of the ground plane, turn the corner and go underneath the ground plane. The antenna ground plane edge becomes another antenna radiation source since when electrons change direction, they become a radiating antenna. If you ever see pattern ripple from antennas measured on a ground plane, the ripple is always from the edge of the groundplane adding in phase and out of phase with the antenna, the edge is hence an additional antenna in real life usage. Hence if your ground plane is tiny, it reflects energy back from the edge to your patch and shifts it's frequency slightly.

Once the groundplane is so large, say half wave square, the reflection from the edge that gets back to the patch is reduced. Using a large ground plane has minor frequency shift. If they recommend a large ground plane and you use a tinyone, the frequency will shift, not sure if high or low, but i'd guess lower, it may depend on what you place your antenna on. If you put a single antenna into an array of antennas, it shifts lower in frequency and you have to make the antennas smaller to reshift your antenna array frequency.

Hence, most of the frequency is set by the antenna, but it can change based on it's surroundings.
kch

 

[robroy1]

Thanks Higgler!

That was written so that even I could understand it!

I recall seeing an antenna (might have been called a reference antenna) that had circular elements around it which it was claimed reduced or eliminated the effect of multipath. I want to think that there is some kind of feature that could be done in pcb layout to emulate this effect or reduce the return from the edge of the ground plane.

Is it possible that in my groundplane design that I could use a pattern (I am thinking in terms of trying to fold the groundplane) to reduce the overall horizontal area. Since this is a circularly polarized antenna (My stomach rumbles when I look at it and think that it is really well polarized!) is it possible that a spiral groundplane might have something like the effect of the four ground plane extenders for a half wave vertical? Or even if I can mimic the resonant dimensions on the ceramic patch itself?

Just to make sure I am not miscommunicating, the antenna I am working with is ~15mm x 15mm x 5mm and has a rectangular metallization on the top of the ceramic, with the feed point ~33% of the width but not located on the diagonal, rather it is located center line along the shorter dimension. I am repeating this because I have mentioned doing a spiral pattern in the above paragraph, but it is not necessarily a real spiral, perhaps a rectangle morphing into spirals (no sharp edges to cause more reflections).

I am working to get a measurement setup to use, no VNA available, but I think I can get power meter, detector, bridge, and sig gen. I am under the impression that it does not matter which way you feed the signal to measure the antenna..is this true in real life? (havent got much real life antenna experience at this point in the journey)

Cheers.
Rob

 

[Higgler]

Reduction in multipath is due to the antenna being circularly polarized. Signal from space comes to you right hand circularly polarized, but if it bounces once off a building (multipath), that bounced signal comes to you left hand circularly polarized and gets to your connector 15-25 dB weaker since it's cross polarized. That's the method to reduce multipath, not any ground plane shape.

Just stick with a standard ground plane, as large as you can handle it, you get more gain upward. Very few people try anything but a standard groundplane.

KCH