I have designed the below double sided PCB which uses 3 separate SIM900 modules with separate SIM cards (different operators, same band) and GSM antennas. The PCB is housed inside a plastic enclosure which in turn is housed inside an enclosed steel cabinet. The cabinet also houses some other components like an AC-DC adapter (the type with shielding) and an 20x4 character LCD. There is an AC line running through a hole on the cabinet to the adapter. Cabinet is not connected to common ground.

The antenna pin of each module is connected to an SMD GSM antenna connector via a short trace on the PCB (about 10mm). Each connector is then connected to a commercially available GSM antenna (please see photo and link for specs), running though holes on the steel cabinet and attached (by glue) to the outside of the cabinet.

I am having unstable signals with this design. While on my telephone the signal is full, I am reading from 5% to 60% signal strength on these modules, and it is also kind of fluctuating.

I have seen some commercial SIM900 modules which also run the antenna trace directly to the antenna connector just like my design. But I have also seen some others which use a 1K resistor in series with the antenna trace, and some vias around the trace.

My question is that what should be causing the signal fluctuation, and is there anything I should do to improve the signal quality and stability?

Thank you so much in advance for any experience shared!

External antenna specs: https://www.aliexpress.com/item/2dbi-3dbi-gsm-824-960Mhz-1710-1990Mhz-GSM-antenna-SMA-plug-male-connector-Aerial-1-5M/32228127548.html?spm=2114.13010608.0.0.eIRczU

enter image description here

enter image description here

UPDATE: I based my design upon the Small Breakout for SIM900 GSM Module by Open Electronics (photo below). Now I can see the only difference is that in their layout, the SMD connector is place as near as possible to the ANT pin of the SIM900 module (I think they are just <1mm apart). Should that be what causes my problem? enter image description here

UPDATE 2: The (painted) steel cabinet has a lid (also steel) at the back of it. I notice that once I remove the lid (so there is a large opening at the back of the cabinet), the signal strength improves immediately by around 30-40%. And when I put the lid back on the signal drops immediately. So could it be the case that this has something to do with interference?

  • 1
    \$\begingroup\$ Are those antenna intended to be used on a metal cabinet? Perhaps try different antenna: the stubby ones with a magnetic base are definitely intended to be placed on a metal cabinet. \$\endgroup\$ – Steve G Nov 1 '16 at 8:58
  • 1
    \$\begingroup\$ @Dave - golden rule in EE design = never use a component without a recognizable and plausible data sheet. \$\endgroup\$ – Andy aka Nov 1 '16 at 9:33
  • 1
    \$\begingroup\$ If you have the antenna just hanging 300mm from any metal does it offer a decent signal. If so then sticking it to a metal surface is causing part of your problem. I have seen numerous GSM antennas (probably made for and) used on metal cabinets but they tend to be NOT-flat. These flat antennas may be intended for sticking to plastic or glass. \$\endgroup\$ – KalleMP Nov 1 '16 at 10:27
  • 1
    \$\begingroup\$ Also keep the three antennas as far from each other as possible. They WILL interfere with operation occasionally even if only one is in active use, heartbeats and negotiations occur sporadically on idle connections. \$\endgroup\$ – KalleMP Nov 1 '16 at 10:30
  • 1
    \$\begingroup\$ Also make sure the antennas are for the correct band. If you have a spare one you can crack it open and measure/inspect the elements. If it is a single band antenna and you are using modules in quad band mode it will have horrible performance when it is connected at the other main band. \$\endgroup\$ – KalleMP Nov 1 '16 at 10:32

I was going to say "check the power supply", since those things have very spiky current draw and demand proper decoupling and low ripple. They might also be subject to crosstalk through the ground plane.

Do you get better results if you only power up one of the modules?

However, the fact that taking the back off the Faraday cage improves reception greatly suggests that the antennas are doing very little work. If you've stuck the antennas to the metal surface, they're probably too close to it to be able to work properly. Normally you want antennas perpendicular to the ground plane.

What sort of signal strength do you get with the lid on and same antenna, but hanging the antenna in free space > 30cm from the box?

Is the box grounded? To the same ground as the PCB?

Are the track impedances correct? Do you need a cutout in the ground plane around the SMA connector?

| improve this answer | |
  • \$\begingroup\$ yes, I have been doing research and what you say is right, I think its the power supply. well antenna should be doing very little work because most of the time the GSM module is idle. I have tried grounding the box (PCB GND) and signal improves a little. I will try other things you suggest and I believe you are pointing me in the right direction, thanks! \$\endgroup\$ – Dave Nov 2 '16 at 4:09
  • \$\begingroup\$ I have done some experiment with antenna attached and not attached to the steel box. This time, I solder a second test PCB which performs a little better. With back lid closed and all antenna attached, the left & right modules report a signal of about 70% while the middle one just 0%-30%. When I move the antennas 5cm from the box, the left & right modules got full signal white the middle one got 70%. So while I am sure there are still issues with the PCB layout, the problem lies more in the part of antennas. Would you have any further comments/advice? \$\endgroup\$ – Dave Nov 2 '16 at 11:39
  • \$\begingroup\$ Antenna design isn't really my area, but when I used GSM antennas they came with datasheets specifying the minimum distance to a ground plane. I suspect you'd get better results with "stubby" antennas, positioned as far apart as possible and perpendicular to the plane of the box. \$\endgroup\$ – pjc50 Nov 2 '16 at 11:42

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.