Capacitive Touch Control Panel Development Using TI MSP430

Client: A Medium Sized Manufacturer of Residential Appliances (Confidential)

Problem: Create a durable, waterproof control panel for a high-end washing machine.

Solution: Custom circuit board design and prototype using the Texas Instruments MSP430FR2633 Microcontroller.

Results: Completed a working prototype in 12 weeks that is 40% less mechanical wear parts and has IPX7 Water Resistance.

The main requirement was an interface that could endure frequent cleanings and high humidity without having a mechanical button fail (as is often the case). We chose the MSP430FR2633 from Texas Instruments due to its built-in CapTIvate™ Technology.

Why Choose MSP430FR2633:

• Cap touch (proximity sensing without having to use a physical button).
• Low Power consumption (<1µA in low power/suspend mode to fulfill Energy Star requirements).
• High Integration: Use of high levels of Analog integration gives us a way to reduce BOM costs by eliminating external op-amps and comarators.

To achieve the best result, we had to follow TI guidelines for designing mixed signal applications with our PCB. Close to the digital signal lines are many sources of noise (UART and LEDs), so it was crucial for us to focus on the layout.

A. Noisy Ground Plane

We created a 4-layer PCB:

• Top Layer (Signal):

  Capacitive touch electrodes and sensitive analog routing.

• PCB Inner Layer 1 (GND):

  Solid ground plane that provides a low-impedance return path for all high-frequency signals.

• PCB Inner Layer 2 (Power):

  Hatched copper pour that controls parasitic capacitance on touch pins.

• Bottom Layer (Control):

  I2C communication traces, LED drivers, and Power distribution.

B. Minimizing Parasitic Capacitance

In capacitive sensing, changes in capacitance measured in femtofarads are the basis for a detection system. To reduce false-positives, "Hatch Ground Fills" (grid-like copper pours) have been utilized under the electrodes instead of solid copper.

This has reduced Cp by approximately 30% and has provided substantial improvement in sensitivity for touch detection.

C. Handling High-Speed Interfaces

The board included a USB-to-Serial bridge (CP2102N) for debugging. The design guidelines required careful routing of the USB differential pair (D+/D-) with 90-ohm differential impedance and matched trace lengths to ensure signal integrity and avoid JTAG interference during programming.

Challenges Encountered: LED Noise Interference

A large discovery that occurred during the layout of the design was that the control signals for the high-power LEDs caused the adjacent touch sensors to produce noise during operation. Specifically, when the LEDs switched ON/OFF, the microcontroller interpreted this as a "touch" event.

Implementation of Proposed Solution:

As per TI application notes, we isolated the digital and analog signal lines by a distance of 4mm. Further, the LED traces were routed on the bottom side of the board and crossed the capacitive sense traces only at 90-degree angles for maximum isolation.

Hardware Validation

Upon receiving the bare PCB from the fabrication house, we performed continuity tests and power-on verification. Using the TI Code Composer Studio (CCS) IDE, we connected the MSP-FET debug probe to the 14-pin JTAG header.

Initial Boot Failure

The first prototype failed to enumerate on the JTAG. Debugging with an oscilloscope revealed that the TCK (Test Clock) line was held at ~1.5V instead of a clean 3.3V logic level.

• Root Cause:

  The UART Tx/Rx pins (used for serial communication) were shared with the JTAG programming pins. The USB-to-Serial transceiver was driving the lines during programming attempts, causing contention.

• Resolution:

  The board required a hardware revision with 0-ohm resistor jumpers to isolate the serial transceiver during programming.

Software Bring-up

With programming stabilized, we used the CapTIvate Design Center (a graphical software tool from TI) to auto-generate the sensor libraries. This reduced manual coding time by 60%. The tool allowed us to configure the 5 buttons and 1 slider wheel in real-time, tuning the touch thresholds to respond accurately through a 2mm glass overlay.

The design team took their project from concept to final prototype through three rounds of testing; these included three phases of validation:

Environmental Testing:

The design had been subjected to water splashes. The panel design was based on ensuring parameters of the layout (avoiding long floating traces that act as antennas) and therefore the MSP430's baseline capacitance remained stable.

EMC/EMI Testing:

Used the MSP430's internal 16-bit timers to allow for PWM dimming of LEDs and eliminate the need for CPU intervention; this produced less digital noise radiation.

Thermal Testing:

The MCU was tested within its operating temperature range of -40 to 85 degrees C without any external heat sinks due to thermal vias connected to an internal ground plane providing thermal conduction.

Through the case study presented here we have shown that successful integration of TI MCUs has as much to do with the physics of the PCB design as it does with the software/code. Our design process employed best mixed-signal layout techniques (ground separation, trace isolation, decoupling) and utilized the integrated analog capabilities of the MSP430 to create a rugged, manufacturing-ready PCB.

Key Takeaways:

• Read the Errata:

  The JTAG vs. UART conflict is a common trap verified in TI documentation.

• Hatch, Don't Solid:

  For capacitive touch, use hatched ground planes under sensors.

• Simulate Early:

  Use TI's reference designs (like TIDA-01237) as a template for power delivery layouts to avoid re-engineering the wheel.