Considerations when choosing between a wireless module and a wireless SoC
There are two main options to consider when adding Bluetooth to your application:
Option 1: Use a wireless system-on-a-chip (SoC) on the product printed circuit board (PCB). It's small and cheaper than a wireless module. But designing with it may be costly.
Option 2: Use a wireless module with Option 1’s SoC inside. A majority of the design is already done including a fully-characterized PCB with RF optimization and antenna layout, shielding, timing components (crystals), external bill of materials (BOM), regulatory approvals, and standards certifications. But they are generally more expensive and larger than the SoC.
Silicon Labs Announces Blue Gecko Bluetooth® Smart Module and SoC
- Blue Gecko Bluetooth Smart BMG113 module pricing = $3.07 in 100,000 unit quantities
- Blue Gecko Bluetooth Smart SoC pricing = $0.99 in 100,000 unit quantities
Example of a Wireless Module and a Wireless SoC Layouts
So, what is the easiest and most cost effective option? That changes depending on the product, the designer, time to market, and so on. Further, the best option changes with volume.
Modules cost more than their SoC equivalent, but companies use them widely. Why? And what's the breakeven volume for when to change from one option to the other?
|Cost Category (for a single product)||Wireless Module||Wireless SoC|
|Board design effort (antenna, layout, match, PCB, debug)||Low||High|
|Resource and lab equipment costs||Low||High|
|Regulatory certifications costs||Low||High|
|Standards certifications costs||Low||Med|
|Time to Market risks||Low||High|
|100K pricing (in our intro / example above)||$3.07 each||$0.99 each|
1. Flat $3.07 wireless module pricing between 10K-300K annual volumes;
2. Flat $0.99 wireless SoC pricing between 10K-300K annual volumes;
3. Flat $0.50 SoC bill of materials (BOM) pricing;- Module price includes the BOM. SoC does not.
4. Gross Margin = $5.12 or 40% above module price. Assume both SoC and module use this for the sales price;
5. SoC requires 3 months of extra development time due to more complexity in design, certification, and regulatory approvals.
Given the above, the annual breakeven volume falls between 200K and 300K.
So why is a breakeven on this so complicated? Because modules remove unknown risks of designing with a wireless SoC, and unknown risks are, by definition, hard to quantify in dollars or weeks.
Hidden Cost #1: RF Engineers and Design
An RF engineer is required for an SoC design. Or, at a minimum, access to RF engineering expertise from the SoC supplier. RF engineers can be expensive. The Glassdoor.com RF Engineer salary is $80-152K/year, unloaded, which does not account for overhead (office space, benefits, etc.). In the US, this typically adds about 33% on top of the salary.
Hiring an RF Engineer = $80K-152K/year + 33 % overhead = $100K-200K/year.
RF Application Notes - Not Always as Easy as 1,2,3
SoC suppliers provide application notes (AN) like Silicon Labs AN930 to help RF layout. These include recommended antennas, traces, board recommendations, and matching networks to maximize performance while minimizing cost and footprint.
However, since every design is different, the recommendations are always-always- hard to implement. In fact, industry experts will attest that is is very common for product designers to follow an application note's recommendations "exactly" and still have performance issues compared to the datasheet specifications and/or product exptectations.
Module companies charge more for their products partly because they are already RF-optimized within a small footprint and low BOM. The whole "system" can be placed on the product board in a matter of hours if not minutes.
Of course, it is "never always" easy. But in the base case, putting a module on the board is measurably easier than putting down an SoC. See the table below for some issues that affect RF performance.
|RF Performance Factor||Potential RF Impact|
|Antenna type, supplier, and placement||Antenna placement, type, material composition, manufacturer, and cost can change signal gain to the matching network resulting in mismatch and poor performance.|
|Antenna trace shape and length||Minor variations in length and shape can change the expected signal energy and therefore the recommended matching network.|
|Board manufacturer||Differing distances or insulation material between layers, PCB via materials, trace widths, screw holes, etc. can have effects.|
|Component suppliers||Different suppliers’ components can behave differently and result in different performance. This can result when designers use “the ones they have on the shelf” versus the recommended supplier, or save a few pennies with a cheaper alternative|
|Component types||Different component technologies can affect received power and voltage (e.g., wire-wound resistors vs. thin-film).|
|Plastics and screw location||Screw placement can have coupling effects for both radiated and received energy.|
|Battery location||Battery location and technology can affect signal power. A charging battery can also be an unknown player.|
|Display location||Like batteries, displays can create interference on the antenna.|
Hidden Cost #2: Lab Equipment and Facilities
RF engineering requires special equipment, software, and facilities to debug RF designs.
|Lab Equipment||Cost to Own||Cost to Rent/Day|
|Calibrated traceable gain horn antenna||~$2,500||Included in a single day rental at test facilities. This is generally $1,000-$3,000/day.|
|Wireless testing software with desired modulation||~$1,500|
|RF isolated, anechoic room (5m x 5m)||~$20,000|
|Wireless standard emulator, sniffer, and debug||~$20,000
Hidden Cost #3: PCB Layout and Antenna Selection
How hard can it be? Many engineers believe it should be easy to follow an application note for layout. While that can be true in some cases, antenna application notes are often complex.
AN930, the Silicon Labs Blue Gecko Bluetooth Smart (BLE) 2.4 GHz antenna application layout guidelines, provides some good examples of the nuances involved. It is designed to provide detailed RF help so customers can get close to a "perfect" layout on their early tries.
Image from AN930 on Blue Gecko Bluetooth Smart antenna matching
But there is still a good chance the PCB will need tweaks to optimize antenna performance. These take time—a few days to determine what needs to be tweaked and a week to turn the board at a local PCB manufacturer. Two weeks adds up when a typical development can take 16 to 20 weeks. As mentioned before, wireless modules can generally be successfully placed on a product board with very simple guidelines. It is still necessary to test a design’s RF performance, but it will likely be much more unpredictable.
Hidden Cost #4: Regulatory Approvals and Wireless Standard Certifications
Products that operate in the unlicensed frequency bands require regulatory "type approvals." Many also require a wireless standard certification (e.g., Bluetooth).Some wireless modules come pre-certified for type approval and wireless standards. Adding them to a product brings these approvals and certifications along, although the product designer must apply for membership in the standards bodies and conduct some product-level regulatory testing. Wireless SoCs do not carry product type approvals or pre-certifications.
Hidden Cost #5: Reduced Product Revenue from TTM Delays
One of the biggest hidden "costs" in using a wireless SoC versus a module is the risk of missing the market window due to incremental time to design it in, test it, debug it, type-approve it, and certify it.Every day the product is not on the market is a day of lost revenue. This can range from a few weeks to a few months. As we saw above with the iPhone 6, removing risk of time to market is a key reason why some very large volume companies still use modules even though they cost more.
Hidden Cost #6: Supply Management and Assurance
For companies with low-volume production runs, modules can mitigate supply risk. A module supplier bargains for SoC supply in their modules on behalf of its entire customer base. Therefore, they consolidate demand and insulate small companies from potential line-down if there is a shortage of SoCs. Sourcing a single module is also simpler than sourcing all the components to put an SoC on the board.
Moving from Wireless Modules to Wireless SoCs
When a company using modules decides to move to wireless SoCs, the question becomes how to reuse the software they have developed with the module. Module companies generally provide a unique software application programming interface (API) for their modules. This serves their customers with an easy-to-use API that allows them to transition between different modules for different SoC versions and/or wireless standards.It also helps the module company retain the module customer as a result of their software investment; the customer won’t want to port their code from the proven, hardened, and mature wireless module to a new, unproven, and unfamiliar wireless SoC.
Single Source for Wireless Modules and Wireless SoCs
Some suppliers sell both modules and SoCs. As such they may support software migration between modules and SoCs.Silicon Labs is one example of such a company. The company has a 20-year legacy of pioneering wireless SoCs, and a long history of working with module companies. Recently, the company acquired two strategic module providers: BlueGiga, a company specialized in designing, certifying, supporting, and manufacturing Bluetooth and Wi-Fi modules, and Telegesis, a leading provider of ZigBee and Thread modules.Silicon Labs has become a one-stop-shop for both wireless SoCs and wireless modules, delivering common software, stacks, support, and development tools.
The answer of whether or not to use a wireless module or a wireless SoC has a high degree of associated complexity that depends on volume, time to market urgency, risk tolerance, and available resources. By choosing a single supplier who can deliver both modules and SoCs while protecting software investment, the migration from module to SoC is simplified if and when the breakeven analysis warrants the move.
Appendix 1: Breakeven Calculations
|SoC Costs||Module Costs|
|Wireless Standards Body Certification (Single)||$ (4,000)||$ -|
|Wireless Memeberships (Single)||$ (4,000)||$ (4,000)|
|Reg. Approvals (US, IC, EU, Korea, Japan)||$ (36,800)||$ (5,000)|
|Lab Equip or Rental for Development
(3 months prorated for SoC dev)
|$ (15,000)||$ (5,000)|
|RF Expertise (3 months for SoC design and debug)||$ (20,000)||$ -|
|RF, Dev, Cert & Govt Approval Cost||$ (79,800)||$ (14,000)|
|SoC TTM Difference vs. Module (3-6 months)||3||-|
|Module Cost (BGM113)||$ 3.07|
|SoC Cost (EFR32 32 QFN w/256 Flash)||$ 0.99|
|SoC BOM Cost||$ 0.50|
|SoC Total Cost||$ 1.49|
|SoC Savings vs. Module||$ 1.58|
|Product Retail ASP Wireless Premium @ 40% GM on Module Price||$ 5.12|
|Total Revenue||$ 51,167||$ 127,917||$ 255,833||$ 511,667||$ 1,023,333||$ 1,535,000|
|Module Dev Costs||$ (14,000)||$ (14,000)||$ (14,000)||$ (14,000)||$ (14,000)||$ (14,000)|
|Module Costs||$ (30,700)||$ (76,750)||$ (153,500)||$ (307,000)||$ (614,000)||$ (921,000)|
|Module Net Profit||$ 6,467||$ 37,167||$ 88,333||$ 190,667||$ 395,333||$ 600,000|
|SoC Dev Costs||$ (79,800)||$ (79,800)||$ (79,800)||$ (79,800)||$ (79,800)||$ (79,800)|
|SoC TTM Lost Revenue||$ (12,792)||$ (31,979)||$ (63,958)||$ (127,917)||$ (255,833)||$ (383,750)|
|SoC + BOM Costs||$ (14,900)||$ (37,250)||$ (74,500)||$ (149,000)||$ (298,000)||$ (447,000)|
|SoC Net Profit||$ (56,325)||$ (21,113)||$ 37,575||$ 154,950||$ 389,700||$ 624,450|
|Breakeven (SoC Profit - Module Profit)||$ (62,792)||$ (58,758)||$ (50,758)||$ (35,717)||$ (5,633)||$ 24,450|
Appendix 2: Costs of Designing a Wireless SoC onto a Product Board and Going to Production
|Cost Category||Module Cost||Confidence Level (%)? Comment?||SoC Cost||Confidence Level (%)? Comment?|
|Selecting antenna||Zero||100%||Med||50% - The supplier likely has a list of recommended antennas. Even so, picking one with confidence can require careful analysis.|
|Laying out antenna||Low||90% - As an all-in-one system, a module is hard to mess up. However, there may be restraints on module placement and "keep-out zones" that could get messed up. It's also likely that the module package probably includes shielding to account for these eventualities and so the probability of these issues is really low.||High||90% - Very high likelihood of trial-error-tweak-repeat cycle. Even highly experienced RF engineers spend weeks optimizing antennas for Rx/Tx performance and low BOM cost. This includes rigorous attention to keep-out zones, effectively isolated inductive loops, component selection and placement, etc...Also likely to need RF expertise, labe equipment and an RF-isolated testing environment.|
|Optimizing antenna layout||Low||High|
|Reducing interference to antenna inputs||Low||High||90% - Very high likelihood of coupled noise into RF fron end from unanticipated and/or unintentional radiators.|
|Reducing interference to antenna output power||Low||Med||50% - Very likely that suboptimal layouts will degrade output performance; unintentional interferors will also degrade output power.|
|Pinout complexity||Standard||Module companies mask SoC pinout changes by accommodating them in an unchanging module footprint.||Standard||SoC pinouts may change between alpha and production silicon. Likewise they may change with subsequent releases.|
|Software complexity||Low/Med||50% - It's likely that module companies have an "soC abstraction layer" development software and API. It varies from spplier to supplier.||Low/Med||50% - Depending on the SoC company's design philosophy, their software APIs may be super easy or super hard. It varies from supplier to supplier.|
|Regulatory certifications||None/Low||100% - Modules can come pre-certified for various regions and wireless standards. There may be some product-level certs required that the module supplier cannot provide, driving some incremental cost here.||High||100% - Each product must be certified in each desired regulatory region and for each supported wireless standard. This is a time consuming and expensive task, and not always successful the first time resulting in "redos."|
Appendix 3: Regulatory and Wireless Standards Certification Cost Estimates
|Certification Body||Link/Comment||Estimated Costs||Module Applies (Yes/No)|
|US FCC, Parts 15B and 15C for unlicensed radios||https://en.wikipedia.org/wiki/Title_47_CFR_Part_15
This page provides an easy-to-read guide.
|~$7,900||Yes for Part 15B|
|Industrie Canada (IC)||http://www.ic.gc.ca/eic/site/smt-gst.nsf/eng/sf01698.html
Guidelines for testing: http://www.ic.gc.ca/eic/site/smt-gst.nsf/eng/sf01130.html
FCC Part 15 certification applies for IC certifications with written reports stating cross-country compliance.
|Europe ETSI and CE||Covers Europe, Africa, Middle East and parts of Asia.
EN 300 328, EN 301 489, and EN 60950 are all important for unlicensed radios.
|Sub-GHz/Proprietary wireless||Generally only require regulatory compliance.||Yes|
|Membership fees||https://www.bluetooth.org/en-us/members/membership-benefits||$0-$8,000 (or higher)||No|
|Qualification fees overview||https://www.bluetooth.org/en-us/test-qualification/qualification-overview/fees||$4,000-$8,000||Yes|
|$4,000-$9,000 (or higher)||No|
|Qualification fees overview||Per test house||~$4,000||Yes|