I have designed a GPS board using Eagle using the Jupiter SE880 chip from Telit (sourced from Digi-key). Being my first high-frequency board, I would appreciate it if someone who knows more about RF than me could verify the general layout, specifically the grounding/RF trace routing. I have read a little on general layout guidelines for RF, and I am pretty sure that everything looks okay, but is there anything that is obviously wrong (not considering wrong connections, etc.) that I have missed? I Thanks in advance.
Check your antenna datasheet. You may have too much ground plane around it.
I checked the antenna datasheet (specifically P/N W3010 from Pulse), and the ground keepout on my board is very slightly larger than what they suggest (hoping to improve performance slightly).
You have the active signal leading away from the antenna thru the area that's supposed to be clear. Can you bring that trace "down" from the antenna before heading to the right?
Good point. I hadn't thought about that.
I'm a little concerned about the length of the track between the antenna and the LNA. You are effectively altering the length of the antenna, and detuning it.
Also the geometry of the track between the LNA and the SE880 should be designed to maintain the correct characteristic impedance (50Ω?).
Designing microwave circuitry is an art, that can't be picked up in a couple of hours.
JohnLincoln:
Also the geometry of the track between the LNA and the SE880 should be designed to maintain the correct characteristic impedance (50Ω?).
I did a little research on the width of a 50Ω trace on a 2 layer FR4 PCB with this calculator, and I came up with 117 mils wide. The only problem is that a track this size takes up a lot of space, and many of the component pads are much less than 0.1" wide. I imagine this is a common problem for many RF circuits, so is there a way to work around it (like converting it to a higher impedance and then converting it back again using pi matching)? Do I just try to make it as wide and short as possible?
I have tried to optimize the placement of the RF components for the shortest distance to the GPS chip (roughly 17mm), while making the RF trace as wide as practically possible (I have also changed the supply voltage trace as a result of wrong schematic wiring).
(like converting it to a higher impedance and then converting it back again using pi matching)
Possible but needs expensive test equipment.
CWashburn:
on my board is very slightly larger than what they suggest (hoping to improve performance slightly).
If the circuit you have has been optimised then any modifications you make at all are likely to degrade performance.
I figured that
Boardburner2:
If the circuit you have has been optimised then any modifications you make at all are likely to degrade performance.
I'm guessing that the recommended antenna keepout area was made as small as possible without degrading performance, so that the manufacturer can say that it only takes X square inches of keepout. Since I am willing to sacrifice more board space for better reception, I figured that enlarging the keepout area a little (<0.5mm) would slightly improve performance, but I may just play it safe and go with what they suggest.
Another way of phrasing my previous question: what do 'professionals' do when it is impractical to use a 100-mil trace that is relatively short?
I also added an LDO in an 'unused' section of the PCB.
