GPS Linear polarization antenna reads... nothing (HELP)

Wireless & IoT
1

Hello all,

after having hit my head a good hundred times with this one, my last hope is that some RF-knowledgeable forum member would be so kind as to cast some light over this issue…

I’ve designed a vehicle positioning device as my first RF project (you know, GPS + GSM/GPRS). I chose the so-called “chip antennas” for the design. Here is a photo of the finished thing :

http://paylet.billingpronto.com/100_4433.JPG

( excuse the big size, I did so intentionally in order we to be able to “see” the small details later)

it’s a double-layer design, being the GPS/GSM IC on the bottom layer. You can clearly see the blue chip antennas, being the one at the right for GSM (900Mhz) and the small one at the left for GPS (L1 freq, 1575Mhz).

Notice the big vias just besides the feed pad of both antennas. That’s the point where I wanted to pick up the signal from.

My GPS/GSM IC uses ultra-miniature coaxial connectors, which in turn can be mated to the appropiate ultra-miniature coaxial cables.

Here you can see those mini-coaxials ( notice the female-to-female configuration)

http://paylet.billingpronto.com/mini-coaxial.jpg

The GPS IC has a male receptacle. The idea is that you buy another male receptacle, put it on your PCB, and then mate the GPS IC’s antenna input with your antenna’s output by using the female-female coaxial. Thus you don’t have to make a microstrip on your pcb. Until now, no problem.

Now comes the time when I though that I was smarter than I really am, only for discovering that I have no idea about RF (life’s sad, yes :wink: ).

In order to cut costs (those minicoaxial are really expensive, not to mention the male receptacles), I decided to cut the female-female coaxial and make two cables from one. Then, I could solder the cut end of each cable directly to the antenna via on the pcb, and plug the female ends on the GPS IC’s male receptacles.

Thus I was saving a whole coaxial and two receptacles for each PCB. So far, so good…

pity that it won’t work. :oops:

For one, the GSM part works flawlessly. I get a consistent RSSI of 20-24 from a maximum of 31, GPRS transmission is flawlessly, and I can even reprogram my PICs on the fly sending the code via GPRS, which proves that the link quality is ok.

But the GPS part… oh man, it won’t take a single read. The NMEA data I get from the IC is the same as if I were to try with absolutely no antenna (the sirfIII gps chip has a “test nmea pattern” which is output when it has no antenna signal. You can easily recognize that pattern since the GSV sentences will show the satellite azimut and power to all ‘000’ ).

I need some advice from you guys, regarding the stupid things I will have - for sure - done here in order this not to work. Whithin my RF inexpertise, I though that removing a few milimeters of coaxial external shielding and soldering the inner conductor to my vias, wouldn’t harm the signal integrity.

Alas, and it is so for the GSM part, which works flawlessly. But the GPS won’t work . Is that due to the GPS much higher working frequency, thus being more sensible to these “experiments” ?

Am I really destroying the GPS signal by cutting&soldering the coax ?

thank you very much for your time !

2

I’ve seen people do basically the same as what you’ve done for WiFi (2.45 GHz), so I think the principle is sound. The main thing about GPS is just that it receives much weaker signals than most other radios.

Does your GPS module use an external active antenna?

3

Hi, thanks for replying

as you have seen this “approach” (so to call it) is passive. I have the option of using an active antenna, since the GPS module has a “bias/feeding” pin in order to feed the active antenna.

I tried it in the past, and of course it worked flawlessly. But two things here :

  1. this SMD pcb is “closed”, I mean, since it’s designed with passive antenna in mind, the bias/feed pin is right now difficult to reach. I could try, though.

  2. active antennas are too big for this design… I really will be needing the patch one to work.

I agree with you in the weakness of the GPS signal. I wonder if integrating a LNA on the pcb would solve the thing.

regards,

4

I could suggest to use an integrated GPS/GSM modem from Telit, but I suppose now it’s too late to do that. Even so, your antenna problem would still remain.

5

Those chip antennas are loners at heart. They don’t like being anywhere near other components, and no ground plane should be beneath the chip. If there is not sufficient groundplane on the board away from the antenna, the receive sensitivity will suffer, and a LNA in series with these chips is almost a must due to the reduced gain. Other than that, it should work fine!

Ron

6

well, the PCB bottom cannot be seen in the photo I uploaded, but I can swear that both antennas have no GND plane beneath them. Man, GSM (the big one) has even a couple of GND traces passing beneath it, and it works flawlessly (now talk about logic ! :lol: )

And the GPS (small one) has not even a single GND trace beneath, and it will not work.

Only thing I could think of, is that 16F886 right besides the GPS antenna…

jmv,

EDIT: excuse me, the 16F886 is not on that photo, it is on the newest PCB version. What you see in that photo is a 3.58Mhz ceramic resonator. Ok, imagine that the resonator is a QFN-28 16F886 (same place on the pcb more or less).

7

I am using the Fractus GPS antenna, and a CEL LNA for GPS. I also have a GPS filter between the two, and am getting positive reception in my cement basement. The filter will help with the GSM antenna so close, and also inductive noises picked up by the RF input line.

8

do you mean a SAW gps filter between the two ? I should try one of those, if only for seeing whether I get some improvement.

I have some samples from a SMD LNA , i’ll give them a try in the PCB.

Anyway, regardless of saw filters, LNAs, etc… one thing is true : even with the antenna alone, if it’s on exterios, it should read something.

but thing is that it is read nothing, absolutely zero.

So something’s really goofed here…

jmv,

9

EFCH1575TCB1 from Panasonic @ Digikey.

10

Hi jmv,

Was wondering if you had discovered the problem with the passive GPS antenna, or had better luck in newer versions of your board?

I’m about to start a project with internal GPS and GSM antennas and would appreciate any and all information I can glean from others! My board will be based around the GE863 module.

Thanks,

Brett

11

RonnyM,

I was thinking about using the Fractus GPS antenna or the passive antenna SFE started carrying recently. It sounds like you’re getting excellent reception with your setup. What GPS receiver module are you using? And would you be willing post a schematic or describe your LNA and GPS filter setup in greater detail? I’m not strong with RF design so it’s black magic to me.

Thanks in advance,

Brett

12

Brett,

pay close attention to RonnyM’s advice, since it’s really accurate: the key to these antennas is proper placement, and I was failing miserably on that one.

In fact, my antenna’s clearances were faulty. Any metalisation close to the antenna will disrupt the signal, as I (painly) learned.

I’ve found also the SAW not to be critical : it’s just OK to use it, but if you don’t , there’s no perceptible signal degradation. Of course, once again, depending on your particular layout this could prove just untrue. YMMV.

With a properly placed and laid combo of chip antenna + LNA (unconditionally stable, mind you, I had to match input network to 50ohm), I was able to get indoor cold starts as low as 20 seconds. Quite an impressive mark, if you ask me 8)

All in all, RF is quite a different animal than “standard” electronics. This is particularly true with medium-high frequencies (as L1 GPS is). You can get 8 GSV readings at around 40dB (an excellent fix by all means), then you de-sold the antenna cable, sold it again, and you have completely lost the signal.

Why? oh my, that second time you made a somewhat-cold join, and where a low-freq signal will survive, a hi-freq one will get destroyed. So delicate, so beautiful is RF design.

In fact, I’ve found that GSM 900 is way more robust than L1. GSM900 antenna will stand cold joins and still provide you with a reliable signal. Mind you, 900Mhz nearly halves L1, and that does make a BIG difference, you can bet.

OTOH, make a little, small, insignificant misstep with the L1 joint, and you’re dead.

funny, this RF stuff, isn’t it.

13

Thanks for the suggestions, jmv. Any way you could share your schematic (just GPS + GSM) and a photo of your new layout? Also, what chip antennas are you using?

14

The GSM units work with tens of millivolt signals…GPS is in the microvolts, and varying atmospheric conditions. It will be inherently more difficult to work with. The USglobalsat ET-301, available at Sparkfun, is one of the best GPS receivers I’ve worked with. The current unit I am working with is proprietary and I can’t share any info on it.

Ron

15

very true, RonnyM.

I thought that I could make my way without using a U.FL SMT coax receptacle, but instead soldering directly a cut coax into the signal feed pad.

With the GSM thing, no matter how bad your solding attempt was, you got always a decent signal.

With the GPS part, you sold a first time, and - by chance - you do it right and get a signal. Lo and Behold. Then you de-sold, sold again, piss it off a little, and you got absolutely zero signal.

It’s still not microwave-esque stuff, yeah, for sure… but L1 is damn delicate.

Finally, I realized the whole picture: med-to-hi freqs are not to be messed with.

jmv

16

OK, after digesting several different posts, I’m thinking a SAW filter and LNA are the way to go for connecting a passive antenna to the GE863-GPS module. Here’s what I have:

http://brett.hackleman.net/temp/gps_ant … tic-v1.GIF

That’s a Fractus chip antenna (FR05-S1-E-0-103), GPS SAW filter (EFCH1575TCB1), and GPS LNA (UPC8215TU).

A few questions:

  1. Any obvious mistakes?

  2. The GE863 assumes an active antenna on startup, and will supply Vbatt on the GPS antenna line. Will this cause problems with my schematic above? Is there any way for me to drive VCC using the antenna line, or should I use a direct connection to Vbatt as shown above?

17

OK, a second read thru the datasheet reveals the LNA VCC supply can be 2.7-3.3v, so it can’t be connected to Vbatt (2.7-4.2v) directly. Besides, this would result in a 4-10mA load even when the GPS was shut down.

I’m going to tie VCC to the Telit VAUX1 line, which will provide 2.9v when the Telit module is powered on.