3
\$\begingroup\$

I am designing a wearable product that uses GPS. In prototypes up to now we have used a patch antenna, and it worked great, however in this revision to cut down on size, we put in a chip antenna, along with a circuit to do antenna tuning.

When I run the GPS I get very consistent results, but they are quite far away from where I actually am. They tend to be about 0.014 degrees south and 0.139 degrees east of where I am.

I have a feeling that antenna tuning is causing the problem. My circuit schematic and layout is below. I believe that I have followed all of the antenna manufacturers guidelines in regards to layout around the antenna.

I do not have / have access to a network analyzer to help me tune the antenna, so my method of tuning will likely be guess and check. (Update: I can afford the low end Rigol spectrum analyzer)

How do you use a Pi circuit and spectrum analyzer to tune an antenna? What values of parts would I need to try and such?

The antenna I am using is: JTI 1575AT43A0040

The chip I am connecting it to is: SIM868

The keepout square around the antenna is 18.8mm x 10.4mm

I am working with standard US GPS frequency.

PCB layout Schematic

\$\endgroup\$
  • \$\begingroup\$ Why not call up Keysight Technology, if you are a student, and give them the opportunity to loan the equipment for a week? \$\endgroup\$ – analogsystemsrf Jun 10 '17 at 3:28
  • \$\begingroup\$ Unfortunately, I graduated in 2016 with my undergrad in business administration. I can afford anything under $1000 though. But I don't think they make what I need for that low of price. \$\endgroup\$ – Reid Jun 10 '17 at 4:15
  • \$\begingroup\$ I think the low end Rigol spectrum analyzer could work, but I have never used one and so I do not know what the process would be for trying to find proper values for the matching circuit. So I guess my question becomes more of how do you do impedance matching with a PI circuit. \$\endgroup\$ – Reid Jun 10 '17 at 5:37
  • 6
    \$\begingroup\$ If you're getting a 10 or 20 mile error, that's a software problem - wrong datum, conversion error, lost digits - not an antenna problem! Antenna tuning really requires a network analyser, but if you have lots of time and less money, you could get quite far with a spectrum analyser or SDR stick, and an ebay noise source. \$\endgroup\$ – tomnexus Jun 10 '17 at 6:44
  • 1
    \$\begingroup\$ I do have a ground plane underneath, controlled impedance 50ohm trace. I ended up going with a spectrum analyzer. Got it tuned perfectly, and then found out it was a firmware error on the side of the GPS chip manufacturer. All is fixed as of about 9/1/17. \$\endgroup\$ – Reid Oct 14 '17 at 16:04
2
\$\begingroup\$

An antenna that is not correctly tuned will degrade your input signal. Ie you will get lower signal levels and less SNR. This will degrade your position solution, but not consistently so. Ie if you have a consistent offset in your position, this is more likely due to reflections (aka multipath) of buildings/trees/objects in the vicinity and not due to your antenna. Multipath is worse with chip antennas than with patch antennas as the latter are circular polarized antennas, that dampen reflected signals to some extend, while the former are linear polarized antennas that accept direct and reflected signals with the same strength.

The book "GPS/GNSS Antennas" by Rao explains the effects and how to deal with them in details.

Though, you are saying that you are 0.1degree off, which is roughly 8km. That is a lot, even for multipath (that usually causes a few dozen meters to a few 100m). It is more likely that your GPS receiver has problems, either with the signal levels or a genuine firmware bug.

\$\endgroup\$

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.