I have designed a vehicle tracker for two-wheelers which is installed in 500 bikes. I am facing an issue with GPS performance (lock percentage is quite low for more than 75% bikes). This makes me question 2 possible things:

  1. GPS antenna quality from the vendor
  2. PCB design of RF part of GPS module

For the first part, I have asked the vendor to share the RF performance and quality check report with me. In the meantime, I want to make sure that the RF design part on PCB is not impacting the performance.

Following is the snap of my GPS RF design part:

enter image description here

As shown in the design:

  • I have tried to put as many vias as possible around the RF part
  • RF trace width is calculated using microstrip width calculator
  • I have avoided putting any power trace near the RF antenna part and separated the ground plane.

I am using an external GPS antenna (through U.FL to SMA converter) which is placed under the front panel of bike (which is made of plastic).

Based on this, I have a few questions which might impact the performance:

  1. U.FL connector is placed at the edge of PCB. Is it creating any kind of issue?
  2. I have GSM antenna and module placed very close to GPS, which might hamper the performance (though I have tried my best to separate out the GND planes of these modules)
  3. Any other inputs/thoughts based on your RF experience will help me a lot.
  • 3
    \$\begingroup\$ I would first like to "take a step back", 25% of the trackers appear to work OK so I could argue that the design is OK (at least under some circumstances). I wonder if maybe more is going on. Maybe it is the bike (causing interference) or small differences in how the tracker is mounted. Can you swap a tracker between a working setup and a non-working one. Which one then stops working, the same tracker (the tracker is the problem) or the tracker that used to work before (the bike might be causing problems). I would try to pinpoint the issue before doing design changes! \$\endgroup\$ – Bimpelrekkie Apr 11 '19 at 10:04
  • \$\begingroup\$ Pinout looks like u-blox NEO. You are using a model with internal SAW, right? \$\endgroup\$ – Andreas Apr 11 '19 at 19:19
  • \$\begingroup\$ I seriously would NOT have carved up the ground plane like that, and please tell me that that is not a 2 layer board... Also, the decoupling is crap, draw the current loops for C28 - C32 to see what I mean. Now the fact that some units work does as @Bimpelrekkie says provide somewhere you might wish to investigate first, but really, 4 layers, buried ground plane (with NO splits) would have been my starting position when designing this thing. \$\endgroup\$ – Dan Mills Apr 11 '19 at 23:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.