Cellular and GNSS Antennas

Hi there,

I’m designing a PCB with an LTE cellular module that has built-in GNSS and I have everything done except the antenna design and I’m no RF expert. I have attached a picture of the recommended schematics for active and passive GPS antenna as well as my current unfinished PCB layout. The GPS antenna is at the bottom and cellular antenna is at the top. Currently I am using this GPS chip antenna, which supports GPS, GNSS, and Beidou, which is what the cellular module supports per its datasheet.

Since I have quite a bit of extra space at the top I really want to design a PCB trace antenna for the cellular, but I have no idea how to do that and I couldn’t find any reference designs online. Also, I am really not sure which off-the-shelf surface-mount antenna would suit me. The LTE module operates at LTE-FDD B2/B4/B12/B13 and LTE-M1 frequencies.

Now to the design:

  • The datasheet for the cellular module says “…the trace impedance must be controlled to 50 Ohms.” Since the traces are so short in my layout (distance from the cellular module pin to the antenna pad is about 6mm), do I have to worry about this? For 1.6mm-thick double-layer FR4 substrate, traces need to be about 120mils wide (really, really fat) to be 50 Ohms, according to this online calculator.

  • The cellular module datasheet says to use a low-noise amplifier (LNA) for a passive antenna, but I’m trying to skip this because RF LNA’s like this one are insanely small (yea, like 0.7x1.1mm with 6 pads). Please tell me if I shouldn’t do this.

  • The chip antenna gives a few mounting configurations, and it looks like mine is the first “End Mount” configuration with the trace going straight out to the antenna. The datasheet says to keep 1.5mm separation distance between the chip antenna’s pad and the GND plane. I’ve done this. However, are the GND planes large enough on either side?

  • Will the capacitors (1210 packages) interfere with the GNSS antenna? I heard that big barrel objects aren’t good near RF antennas. However, those large caps are needed to prevent large voltage drops on the input and there’s really nowhere else to put them (I don’t want those two sore thumbs on the bottom of the board, and they have to be close to the power pins on the module).

  • I’m considering switching to this chip antenna instead, since it has higher gain (peak 3.7dBi vs 1dBi) and is much smaller and also requires a much smaller GND clearance area. Is there any real difference or should I go for this new chip?

I know I threw out a lot there, but any suggestions? Thanks!

androidfanboy:
Hi there,

I'm designing a PCB with an LTE cellular module that has built-in GNSS and I have everything done except the antenna design and I'm no RF expert. I have attached a picture of the recommended schematics for active and passive GPS antenna as well as my current unfinished PCB layout. The GPS antenna is at the bottom and cellular antenna is at the top. Currently I am using this GPS chip antenna, which supports GPS, GNSS, and Beidou, which is what the cellular module supports per its datasheet.

Since I have quite a bit of extra space at the top I really want to design a PCB trace antenna for the cellular, but I have no idea how to do that and I couldn't find any reference designs online. Also, I am really not sure which off-the-shelf surface-mount antenna would suit me. The LTE module operates at LTE-FDD B2/B4/B12/B13 and LTE-M1 frequencies.

You need one that supports all the bands, it has to resonate on each.

Now to the design:

  • The datasheet for the cellular module says "...the trace impedance must be controlled to 50 Ohms." Since the traces are so short in my layout (distance from the cellular module pin to the antenna pad is about 6mm), do I have to worry about this? For 1.6mm-thick double-layer FR4 substrate, traces need to be about 120mils wide (really, really fat) to be 50 Ohms, according to this online calculator.

Absolutely yes, 50 ohms it needs to be all the way from the antenna to the next device on the signal path, and you need enough clearance on either side to prevent coupling to nearby components. Every mm matters at
microwave frequencies, which this is.

  • The cellular module datasheet says to use a low-noise amplifier (LNA) for a passive antenna, but I'm trying to skip this because RF LNA's like this one are insanely small (yea, like 0.7x1.1mm with 6 pads). Please tell me if I shouldn't do this.

Any 50 ohm MMIC LNA that covers the band(s) of interests will probably do, yes they are available in sensible
packages.

  • The chip antenna gives a few mounting configurations, and it looks like mine is the first "End Mount" configuration with the trace going straight out to the antenna. The datasheet says to keep 1.5mm separation distance between the chip antenna's pad and the GND plane. I've done this. However, are the GND planes large enough on either side?

which datasheet is this? The GPS one?

  • Will the capacitors (1210 packages) interfere with the GNSS antenna? I heard that big barrel objects aren't good near RF antennas. However, those large caps are needed to prevent large voltage drops on the input and there's really nowhere else to put them (I don't want those two sore thumbs on the bottom of the board, and they have to be close to the power pins on the module).

Antenna's best nowhere near the power components, that's a very noisy part of a PCB.

  • I'm considering switching to this chip antenna instead, since it has higher gain (peak 3.7dBi vs 1dBi) and is much smaller and also requires a much smaller GND clearance area. Is there any real difference or should I go for this new chip?

I know I threw out a lot there, but any suggestions? Thanks!

Where's the image of the PCB?

I found an interesting variant GPS antenna which goes over ground plane, so allows stuff on the other
side of the PCB rather than forcing it to be empty: ONBOARD GPS Mini Antenna 2.4GHz

Thanks for the response!

  • Sounds like for the cellular antenna I need to just find the right one. Making my own would be a challenge lol.

  • The chip antenna diagram I was referring to is page 2 of the datasheet but I don’t think I’m going to use this chip antenna anymore (it’s too long).

  • I think I am going to switch to this GPS chip antenna because it’s much smaller/shorter and has a higher gain than the other one.

  • So if the chip antenna specifically states that its impedance is already 50 Ohms, I still need to worry about making the feed line 50 Ohms? I thought the impedance matching is between the cellular module and the antenna, and if the antenna isn’t 50 Ohms then that 50 Ohms has to be from the trace that leads to the antenna (feed line).

Thanks!

Here’s what I have using the new chip antenna.

The fat trace (feed line) you see in the attached image is only 50 mils though… do I have to get it to 120 somehow??

Also, is using a single 33pF capacitor in series with the feed line OK? Do I need a full matching circuit? At this point I’m just trying to get it to work, then fine tune later, but I need general advice on this design.

Also, will this GSM Quad-band antenna work? My cellular module operates at the following frequencies:

  • GSM frequencies: EGSM900MHz and 1800MHz
  • LTE FDD-B2
  • LTE FDD-B4
  • LTE FDD-B12
  • LTE FDD-B13
  • LTE-M1

That’s a heck of a lot of frequencies to radiate all at once!

Thanks!

I changed the trace to 125mils wide and this is how it looks lol. Only problem is that I can’t make the trace from the cellular module to the 33uF cap that thick, but it’s really really short so I hope that ok?

Those GPS ceramic chip antennas are fairly poor, first fix times without some form of external assist can be 15 minutes or more.

There is a comparision of signal strengths of ceramic chip, 1/4 wave wire, 3/4 wave wire, small ceraminc patch and large ceramic patch GPSs in this doucument

GPS Antennas and their effect on LoRa Tracker power consumption V2

androidfanboy:
I changed the trace to 125mils wide and this is how it looks lol. Only problem is that I can't make the trace from the cellular module to the 33uF cap that thick, but it's really really short so I hope that ok?

You have no clearance around the trace, so it will be less than 50 ohms. That microstrip trace width is assuming
a ground plane on the opposite side of the board, and clear space around it top-side.

Normally these things go on 10- to 16-layer boards so that the traces are much thinner for 50 ohms.

This site shows what 50 ohm microstrip looks like on 2-layer 1.6mm pcb, note all the clearance either
side of the strip: http://www.kaislab.com/

Can you go to 1mm or thinner FR4? Or even an impedance-controller substrate (more predictable
impedance, lower RF losses).

MarkT:
You have no clearance around the trace, so it will be less than 50 ohms. That microstrip trace width is assuming
a ground plane on the opposite side of the board, and clear space around it top-side.

So if my impedance is less than 50 ohms, can I just decrease the width? I know it’s probably harder to calculate, but wouldn’t that theoretically work?

Also, the spec sheet for the cellular module says to use large caps in parallel (summing up to 300uF if the power supply can handle up to 2A spikes) to prevent voltage drops on the power supply, and the three VBAT pins are right next to each other but sandwiched in between the cellular antenna and the GNSS antenna. This means I can’t really get any clearance unless I remove the large caps. How do you suggest I do this?

Another thing: I’m using the SIM7000 series modules from SIMCOM (spec sheets attached below) and the website and hardware spec sheet all say that there is GNSS built in to the module, but there is no mention of GNSS or GPS on the AT command manual. What is going on here? Should I just use one of their GNSS modules with antennas?

EDIT: I guess I’ll assume that the GPS AT commands are the same as their other products like this one.

So if my impedance is less than 50 ohms, can I just decrease the width? I know it's probably harder to calculate, but wouldn't that theoretically work?

This is the sort of problem a VNA is needed to solve.... RF test equipment for microwave frequencies is not
cheap.

Some means of evualating the trasmitted power can help tune up a transmitting antenna, but
for reception direct into a chip you only have whats on the chip to help diagnose...

OK, so I guess it gets unpredictable with that sort of layout. If I put the caps on the other side of the PCB and lay out the traces like the attached picture with 120 mils, would that be more controlled at 50 Ohms?