John, antenna issues are probably the most challenging problem for engineers new to wireless design. To give you an example, I was working on a GSM product for the USA market. We had the hardest time getting through att certification. They kept telling us the antenna design was not good enough, even though we never really experienced trouble establishing a connection. The issue was that the antenna couldn't pick up a connection when the network signal strength was low. So while my phone had one bar, the device didn't see the network at all. Our redesign took over four months and we missed our launch deadline. it was pretty brutal.
If you're serious about building a connected device, I strongly encourage you to hire an antenna design firm. These guys know know what their doing. Or at the very least hire an experienced antenna design engineer. It will save you tremendous time / money in the end.
Also, AT&T has put up a good white paper on antenna design: http://developer.att.com/devcentral/tools_technologies/network/docs/Antenna_Fundamentals_v1.0.pdf
Talk with the experts: www.ethertronics.com
They have a very high level of profissionalism apart from local support infrastructures (chambers to measure the antenna performance) and a really great technology called IMD (isolated magnect dipole) which provides smaller size, higher isolation and greater efficience.
Ps. I was at Cinterion when replied this message first and I had so many good experiences that I joined Ethertronics! The company is becoming well kown in the M2M market.
- Corporate Brochure.pdf (313.4 K)
One thing we do when we design a device is make sure there is room to modify the RF signal path even after the PCB has been designed. Depending on the module being used; we recommend not using a transition cable based design. We've found during our design cycle that it is easier to use an RF trace and implement a passive antenna matching circuit. We do this for both cellular and GPS RF signal paths. Also; add ESD protection whenever possible to RF signal paths. Another thing to do is to utilize a co-planar waveguide PCB (see attachment.) This is important when it comes to the PCB material itself and ensuring proper routing of the RF signal path. This alone can cause failures if not done properly.
I've attached a schematic PDF of our open source design (ModelONEt) for the Telit GE865-QUAD module, and also an example RF matching circuit for GPS (this one is for a Trimble module.) You can see an example circuit you can use. A good antenna engineer can "tune in" your antenna. Taoglas is a good company to work with and they have outstanding antenna engineers. We use them in most of our designs. More or less; these antenna matching circuits don't necessarily need to be populated but are there to ensure that if you need to tune in the antenna after the PCB is designed, you can. The last thing you want to do is spend money building circuit boards that are paperweights and door stops because of poor RF performance. These passive circuits do offer some tuning ability after the fact. Leave them as "do not populate" and let the antenna engineer take over during testing.
We've found that with proper PCB layout and circuit implementations; along with the right antenna vendor its not that difficult to pass PTCRB. I recommend that once you have the first version of your hardware ready that you do a preliminary PTCRB "pre-test" that takes measurements at 90 deg. just to see where you are at. The only difference between the pre-test and actual PTCRB testing is that PTCRB is done at 30 deg. patterns. For a couple thousand dollars you can do a pretest with several of the PTCRB authorized labs and save yourself a headache. Also have the antenna vendor and/or module vendors assist you in a pretest.
Lastly; use EMI filters in the design, all over the place. Usually at minimum in front of the power supplies to the modules and core components. EMI on everything (except analog). What I typically do is install the filters and then remove them later for testing internally. If you don't encounter EMI/spurious issues, then remove them for production, else leave them in. It's much easier to design filters in and not install them than it is to add them in after the fact. Murata makes several filters and they are cheap. Better to prepare for a headache than to deal with one after the fact.
Principal Electrical & Systems Engineer
Systems on Chip LLC.