Designing Your First Cellular-Enabled IoT Device - Part Two

With the functional, market, regulatory and carrier requirements identified, the next step in our journey is to select a suitable cellular module, antenna solution and carrier partner.

To select a suitable cellular module, a common approach is to first identify some potential candidates by searching module vendors’ websites, screening by suitable technologies (e.g. LTE Cat-1), interfaces (e.g. USB, UART, PCIe) and form factors (e.g. land-grid array / LGA module). There are many cellular module vendors in the market today, with some examples including Telit, Nordic Semiconductor, Quectel, u-blox and Sierra Wireless. In most cases, you will be able to identify suitable modules from multiple vendors. When selecting your module, in addition to evaluating the technical and regulatory suitability, it is also suggested to consider the quality of available documentation, and the helpfulness and timeliness of engineering support. Of course, price, availability and lead time must also be considered.

Once one or more candidate modules have been identified, it is prudent to contact the vendor or their representative to begin a dialogue about your product needs. They will be able to help guide your selection and support your design-in process. They will also be able to provide you with technical documentation, and in most cases, an evaluation kit. Necessary regulatory compliance documents should be provided as well for verification. Also, with consideration to the ongoing semiconductor shortage, they will be able to update you on the latest availability and lead time information.

With a module selected, it is time to design it onto your PCB.

Common module form factors are land-grid array (LGA), edge-castillated modules, or card-type (mini-PCIe or M.2). In each case, follow the manufacturer’s recommended guidelines whenever possible, using datasheets, reference designs and manufacturer field application engineers (FAEs) as resources. Good engineering practices should always be followed - building in margin, designing for electromagnetic compliance (EMC), designing for manufacturing (DFM), etc. Check out our blog for entries on these items.

While module integration is manufacturer and part specific, there are some common elements. A solid ground plane underneath the module, well connected to the module ground pins and well connected to the other ground plane layers with stitching vias is important to ensure good RF and signal integrity performance. Similarly, a low-impedance (wide, short copper tracks or pours), well-decoupled power path for the module VCC or VBAT input (generally 3.3 or 3.8 V) is also important to ensure a clean voltage supply. The RF trace(s) to the antenna(s) need perhaps the most care when routing to ensure 50 Ω single-ended trace impedance is maintained, the transmission losses are sufficiently low, and the matching network and any other inline elements do not cause any major reflections or impedance mismatches. A common structure to use for the RF trace(s) is the coplanar waveguide (CPW) due to its low losses and ease of implementation on a typical four or more layer PCB.

The placement of a cellular module in your design is often chosen to minimize the routing length needed for the RF trace(s), power input and other communication buses running to the module (roughly in the order of descending priority).

When it comes to selecting an antenna, in general, using the largest antenna your product can tolerate will result in the best RF performance. There’s no beating physics, and a larger antenna will almost always outperform a smaller one when working in the cellular frequency ranges. That being said, with today’s ever shrinking devices, compromises usually need to be made to attain form factor or size targets, and smaller antennas often need to be considered.

If your product can support it, using an external, multi-band, whip-type antenna is often the simplest choice that offers good performance and flexibility. Most cellular modules are certified with this style of antenna, so as long as an antenna with maximum gain below what is allowed in the FCC modular certification, there should be many options for antenna choices.

Figure 1: Linx Technologies ANT-LTE-MON-SMA

For smaller form factor antennas intended to be embedded into a device, there are two general categories for off-the-shelf choices: PCB-mount and flexible-patch type. PCB-mount antennas are soldered onto a PCB, usually near a board edge, and require both a copper-keepout area and GND plane to function properly. Reducing the size of either the keepout area or GND plane will reduce the antenna performance, and at some point, will lead to failures in certification testing and/or acceptable real-world performance. Bear in mind that many PCB-mount antennas will require at least a quarter wavelength’s worth of ground plane distance in at least one axis, at the lowest frequency of operation. For LTE Band 12 operation (700 MHz), this corresponds to a length of around 11 cm, which can be quite impractical or impossible to achieve in small devices. Finding the balance is an engineering trade off, and is an area that NeuronicWorks has significant experience in.

Figure 2: Ignion TRIO mXTEND

Some antenna companies, such as Ignion, provide value-added services to help you integrate an antenna into your design. Such services include recommending an antenna model, performing pre-analysis for RF performance, calculating matching network component values, and more.

At this point, we have selected a module and antenna solution and designed them into the first prototype. It is a good idea to have the module vendor review the design to check for errors as well as suggest improvements. They will have seen many designs and should be able to provide valuable feedback in a reasonable timeframe (usually two weeks or less). With their blessing, it is time to secure some prototype inventory quantities of the modem and antenna and build some units.

To release a cellular-enabled product in the North American market, wireless testing and certifications will need to be performed and passed. We touched on this in step three of part one. At this point, with your regulatory requirements identified, it is prudent to begin discussions with test labs in order to receive both quotations and information on how to submit your product for testing.

Generally, the test lab will want to receive one or more functional prototypes, representative of the final product, in order to perform the testing. Special firmware and/or RF ports (e.g. u.FL connectors inline with the signal path where measurements may be made) may also be needed in order to facilitate the testing.

Certification costs will vary, sometimes greatly, depending on the cellular technologies used, the type of product, the test labs engaged, etc., but in general, you can expect to pay the following for testing and certification:

• FCC Part 15 Subpart B (Unintentional Radiator): $5,000 to $10,000 USD
• TRP/TIS pre-testing: $5,000 to $10,000 USD
• PTCRB: $7,500 to $30,000 USD
• Carrier: $0 to $10,000 USD

For FCC testing - which also generally covers ISED requirements - NeuronicWorks has partnered with TUV SUD and Nemko, located in the Greater Toronto Area (GTA) and Kitchener-Waterloo (KW) respectively. These partnerships allow NeuronicWorks to efficiently and effectively test products and obtain certifications with a short lead time, and provide onsite engineering support when needed.

Some test labs that offer TRP/TIS and PTCRB testing include 7Layers, CETECOM, Intertek and BluFlux. All of these laboratories are located in the continental US.

With your product design complete and certifications done, it is time to begin manufacturing. Launching a product and ramping up production volumes is a whole topic in itself, so we will instead focus on cellular-related considerations when doing so.

In step four of part one, we touched on mobile virtual network operators (MVNOs). If you haven’t already secured your SIM solution, now is the time to do so. Partnering with an MVNO is a great option if you are looking to deploy in multiple geographical locations and wish to bypass the process of negotiating directly with local carriers. Some examples of MVNOs that we have worked with at NeuronicWorks previously include Soracom, KORE Wireless and Hologram.

You will need to manage your SIM (or embedded SIM: eSIM) activations. SIMs / eSIMs typically have recurring monthly fees when activated, as well as an allotment of monthly data. This data is sometimes pooled with other devices in the plan. You’ll generally only want to activate a SIM when the product is being shipped to a customer, as it makes little sense to pay monthly fees on a SIM if it’s sitting unused in a warehouse. You’ll also have to manage customer subscriptions, if your business model uses them, to deactivate SIMs when a customer discontinues services (or doesn’t pay their bill).

Like any other wireless technology, there are going to be hiccups with certain deployments. Think of Wi-Fi, which is over two decades old but still gives us grief from time to time. You’ll need to be able to assess the situation when a customer calls and says their device isn’t getting connectivity. Sometimes this is related to an unactivated SIM card. Sometimes this is related to an area of poor cellular coverage. You’ll need to be able to run through common troubleshooting items to support customers in the field.

Most carriers will also require you to have a system in place to push remote updates to your products, specifically the cellular modules, but sometimes other programmable devices (e.g. microcontrollers) as well if they affect the behaviour of the cellular connectivity. Nearly all cellular modules will support this, but you will need to build the “back-end” services to provide the updates in a secure and controlled manner.

Also important is a system of monitoring and alerting for high data usage. This can indicate abnormal behaviour (e.g. a firmware bug), or heavier-than-normal usage, both of which should be investigated to avoid data overage charges.

Navigating the landscape of designing, certifying, manufacturing and selling a cellular-enabled IoT product can be a daunting task. This article has covered some of the major steps, but as always, the devil is in the details, and no two situations are identical. Reach out to us at NeuronicWorks to work with a team of experts with experience launching dozens of cellular-enabled devices.