Wi-Fi + Bluetooth Combo Module Design-In: 8 Decisions Before Volume Production

Most Wi-Fi and Bluetooth combo modules do not fail in the RF lab. They fail at integration: a shared bus that starves the Bluetooth endpoint, an antenna pad tucked under a metal bracket, a strap pin held high during power-on, or a reflow profile run one pass too many. This guide walks through the eight decisions that determine whether a 2.4 GHz combo module reaches volume production on schedule. The measured data comes from two shipping Realtek RTL8723 modules: the USB-based BL-R8723BT1 (RTL8723BU-CG) and the SDIO + UART BL-M8723CS1 (RTL8723CS-CG).

1. Start with the host interface, not the datasheet headline

Both modules offer IEEE 802.11b/g/n 1T1R with a PHY rate up to 150 Mbps, and both cover the same 2.4–2.4835 GHz band with 13 channels at 20 MHz. What actually differs — and what decides your schematic — is how the module talks to the host.

The USB version puts WLAN and Bluetooth on one shared USB 2.0 connection. That means one host port, one set of differential pairs, and no second controller to enable. The trade-off is that both functions live on the same port and share its interrupt and bandwidth budget. In a dash cam, that single-cable arrangement is usually the difference between a two-layer board and a four-layer board.

The SDIO version separates the two functions: WLAN runs on SDIO 2.0 (the same pins can be configured as GSPI), while Bluetooth runs on an independent high-speed UART with hardware flow control. Nothing is shared, so Wi-Fi throughput is not competing with Bluetooth traffic at the bus level — but you need spare SDIO and UART resources on the host SoC, plus roughly 30 more pins on the module footprint (44 versus 14).

A practical rule: if your SoC has a spare USB host port and no free SDIO controller, take the USB module; if SDIO and a UART are already broken out, take the SDIO module and keep the two paths independent.

USB: two functions share one bus. SDIO + UART: each function gets its own bus.

2. Choose the Bluetooth profile your product actually ships

Bluetooth is where the two modules diverge most, and where specifications are most often misread.

BL-R8723BT1 is dual-mode: Bluetooth Classic v2.1 + EDR together with Low Energy v4.0, with Classic operating in Class 1 and LE in Class 1.5. Measured transmit power is 6 dBm typical (8 dBm maximum) for BR/EDR and LE alike, and BR/EDR receive sensitivity is -90 dBm at 0.1% BER. If your product streams audio to a speaker, pairs with a legacy phone accessory, or needs an SPP-style data channel, this is the variant to use.

BL-M8723CS1 is BLE-only, Core Specification v4.1, Class 1.5, with 3 dBm typical and 6 dBm maximum transmit power, and -90 dBm typical LE receive sensitivity at 5% PER. There is no Classic stack and no audio path at all. That is not a limitation to work around — it is a cost and power decision. For provisioning, OTA triggers, sensor telemetry, or a status link to a phone app, BLE 4.1 is sufficient, and you avoid paying for a Classic stack you will never use.

Do not specify a Bluetooth feature set from a marketing overview alone. Version strings get copied between documents; contact our team to obtain the official specification document and RF test data for validation.

3. Budget the power tree around peak current, not average current

Both modules run from a 3.3 V ±0.2 V main rail. BL-R8723BT1 draws a maximum of 400 mA from that single rail; BL-M8723CS1 draws up to 450 mA on the main rail and adds a separate I/O rail (VDIO) that can be either 3.3 V ±0.2 V or 1.8 V ±0.1 V. The separate I/O rail is genuinely useful: you can keep a 3.3 V RF supply while level-matching directly to a 1.8 V SoC, without external translators on six SDIO and UART lines.

Plan for transmit peaks, not specification averages:

Source: current-consumption tables from the official module specification documents, available upon request. Values are typical / maximum at the stated test conditions; RF-test figures, not application-level averages.

Two observations worth designing around. First, the SDIO module draws noticeably more in 802.11b transmit than the USB module (328 mA versus 255 mA peak) because it runs at 17 dBm across all 11b rates, whereas the USB module is calibrated at 15–17 dBm depending on rate. Second, the USB module sits at 130–135 mA in receive, roughly double the SDIO module, because the USB transceiver stays active while the radio listens. On a battery-powered handheld, that difference is not academic.

Layout guidance: place a 47–100 µF bulk capacitor and a 10 µF plus 0.1 µF pair within a few millimeters of VDD33, keep the 3.3 V feed on its own trace or plane, and never share the branch with motors, relays or the backlight driver. A 328 mA step on a shared rail will show up as an RF desense problem that looks like a layout fault.

4. Treat the antenna pad as the highest-risk item on the board

Neither module includes an antenna. Both connect to an external antenna through a half-hole (castellated) pad, with a specified VSWR of 1.7 typical and 2.0 maximum for BL-R8723BT1, 2.1 maximum for BL-M8723CS1. A VSWR of 2.0 already reflects roughly 11% of the delivered power; a poorly tuned antenna turns 17 dBm of conducted power into noticeably less radiated power and shortens range.

The practical checklist:

• Route the antenna feed as a controlled 50 Ω line, as short as the mechanical design allows, with no stubs, no vias in the RF path, and ground stitching along both sides.

• Keep the antenna clear of the battery, speaker magnet, LCD flex, and any grounded metal. A 2.4 GHz antenna placed against a metal bracket loses more range than any driver setting can recover.

• If the enclosure is metal, move the antenna outside it and use an IPEX-to-SMA pigtail — then account for the cable and connector loss when you calculate link budget.

• Validate with the actual enclosure, not with the module on a bench. Radiated testing with the final plastics, battery and cable routing is the only measurement that matters.

Conducted transmit power is calibrated per rate, so the effective radiated performance depends on the antenna as much as on the module. Expect 17 dBm on 802.11b and 802.11g 6 Mbps, and 14–15 dBm on the high-throughput MCS7 rates where EVM limits the usable output:

Transmit power at the antenna pad, from the official specification. Sensitivity for reference: -92 dBm (11b 1 Mbps) and -66 dBm (HT40 MCS7) on BL-R8723BT1; -93 dBm and -66 dBm on BL-M8723CS1.

Keep the feed short and clear of metal, battery and shield cans; add pigtail loss to the link budget when the antenna sits outside a metal enclosure.

5. Design coexistence into the application, not into the hope

A 2.4 GHz combo module has one radio shared between WLAN and Bluetooth. The driver time-slices the radio between the two, which works well but is not free: under concurrent load, aggregate WLAN throughput drops and Bluetooth latency rises. That is normal and expected behavior for this architecture, not a defect.

What to do about it: define the traffic profile early and test against it. For an IPC streaming 1080p while a phone provisions over BLE, measure frame loss with both running, not one at a time. For a POS terminal, test a payment transaction while a BLE scanner is connected. If audio is part of the product, keep the audio link and the WLAN channel plan aware of each other — channel 1, 6 and 11 with a 20 MHz width leave more room than HT40.

Two architecture choices reduce the problem before firmware is involved: use BLE rather than Classic where the use case allows (shorter, scheduled connection events), and avoid placing the Wi-Fi and Bluetooth antennas in the same keep-out if the mechanical design gives you a choice.

6. Get the control and strap pins right before layout freeze

This is where SDIO-based designs are most often re-spun. On BL-M8723CS1, several pins carry dual functions and one carries a power-on strap:

• CHIP_EN (pin 35) is an active-low reset with a 100 kΩ internal pull-up to VDD33. Leave it accessible or drive it from a host GPIO with a defined power-on delay; do not let it float near a switching rail.

• WL_DIS# (pin 12) pulls the WLAN function down and disables the SDIO interface when driven low — useful for a hard radio-off state. BT_DIS# (pin 34) does the same for Bluetooth. Note the interaction the datasheet flags: it also describes a whole-chip reset state when both pins are deasserted, so drive these pins from the host instead of tying either of them permanently.

• GPIO1 (pin 25) is a strap pin, internally pulled low by 100 kΩ, and must not be pulled high during power-on. If your host drives this net, keep it high-Z until the module has booted — a level shifter or an external pull-up here will produce an intermittent boot failure that is very hard to trace.

• LPO (pin 24) can take an external 32 kHz or RTC clock for accurate low-power timing; it is shared with GPIO6 and the high-speed UART RTS output, so decide early which function you need.

• Pin 41 is internally connected to other ground pins, so it can be left unconnected or used for UART_CTS in your application circuit without affecting normal operation — handy when a 44-pin footprint has to fit an existing pad array.

The USB module is simpler: a 32.768 kHz low-power clock input on pin 4, plus the DEVICE_WAKE_HOST and HOST_WAKE_DEVICE pair used for remote wake. If your product relies on host-initiated wake from suspend, wire both and confirm the polarity expected by your driver.

Rails first, then release CHIP_EN, then let the host enumerate. Keep GPIO1 high-Z through power-on.

7. Match the operating range to the enclosure, not to the room

BL-M8723CS1 is rated -20 °C to +70 °C; BL-R8723BT1 is rated -10 °C to +70 °C. Both are specified for 10–95% RH non-condensing. The 10 °C difference is the one specification that most often disqualifies an otherwise correct choice: a dash cam mounted on a windshield in a cold-climate market will see temperatures below -10 °C, and the USB variant is outside its rated range there — a robot vacuum or an indoor POS terminal never will be.

The upper limit deserves equal respect. A sealed IP camera housing in direct sun, or a set-top box stacked under a TV, can push local ambient well above 40 °C. Measure the air temperature at the module, not the room temperature, and derate your thermal design against the 70 °C ceiling with margin rather than designing to it.

8. Lock down manufacturing, storage and certification scope

Three items that appear late in the specification and cause late surprises:

Reflow. Both modules specify a peak of 250 °C maximum and no more than two reflow passes. A third pass — common when a board is reworked or a second assembly step is added — is outside the specification, and the failure mode appears as a marginal RF or BLE link months later, not at test.

Storage and handling. BL-R8723BT1 defines a -45 °C to +85 °C storage range, a 12-month window from vacuum packing, and a 72-hour window after the bag is opened. If the humidity indicator has changed color, bake at 60 °C for 24 hours once before soldering. Both modules are ESD-sensitive devices and need standard handling precautions on the line.

Certification. RF and EMC scope depends on the market you ship into, and on the host enclosure — a module certificate does not automatically cover a finished product. Contact us to obtain the current test reports for the specific model and market, and match the report to the exact firmware and antenna configuration used in production. Domestic and export requirements can differ.

Side-by-side selection table

Summary from our official specification documents, available upon request.

Driver and platform coverage

Both modules are built on Realtek silicon with vendor driver support for Linux and Android, which is what most IPC, POS and appliance designs actually ship. The USB variant additionally carries Windows support, useful for peripheral and dongle-style products. Before you commit a kernel version, confirm the exact driver package with our FAE team — driver readiness, not RF performance, is usually what moves a schedule.

What support you can expect before and after the design win

• Standard modules in sample quantities, plus official specification documents, pin definitions and reference schematics on request.

• A layout review of your SDIO or USB routing, antenna keep-out and power tree before you release the board.

• Custom module development on either chip: pin-out modification, footprint resizing, antenna options (on-board ceramic, IPEX connector, half-pad), and firmware parameter tuning.

• Documentation support for the certification path you are planning.

If your project is still at the architecture stage, our RTL8723BU-CG and RTL8723CS-CG platform overview covers the chip-level differences, mass-production application cases and custom development scope; the 2.4 GHz Wi-Fi module portfolio lists the other pin-compatible options in this class.

FAQ

Next step

Send us your schematic page and mechanical constraints — antenna location, enclosure material, host interface, and target operating range — and our application team will return a fit assessment for either module, including the pin-level changes needed for your layout. Request samples, datasheet and a layout review.

All performance data in this article is typical values for reference. Full specification documents, test reports and design resources are available upon request.