Medical Wearable Remediation: IEC 60601 Compliance with nRF52840 and STM32L496VG

We identified and resolved critical gaps during the 5-month engagement including areas of isolation, EMC, and safety. This ultimately enabled the device to be able to successfully complete and meet IEC 60601-1 and IEC 60601-1-2 with zero major non-conformances.

Client: Stealth-mode Medical Device Company

Type of Device: Wearable, Battery Powered, BLE Connected Class II Therapeutic Stimulator (TENS-like but with closed-loop control)

Role of Project: Fractional Senior Hardware Engineer – Compliance & Design-Review Advisor

Duration: 5 months (20 weeks) | Effort: ~15 hours/week average

Outcome at Completion of Project: Passed formal IEC 60601-1 (3rd Ed + Amd 1) and IEC 60601-1-2 (4th Ed) with zero major non-conformances.

An immediate serious problem has been found: The nRF52840 has its ground connected directly to the return path for an output stimulation. This lack of functional isolation is a clear violation of IEC 60601-1 Clause 8.5 regarding the isolation of applied parts.

1. Isolation & MOPP (60601-1, Clause 8.5)

Problem:Patient-applied electrodes shared a common ground with:

• nRF52840 BLE antenna ground
• USB charging port (VBUS: up to 5.25V, but transient condition allows 240V line-to-ground fault)

Measured creepage:

Between stimulation return and USB VBUS: 2.1 mm (on PCB).

Required for 2 MOPP at 250V working voltage: 8.0 mm (per IEC 60601-1 Table 12).

Impact at notified body test: Would have failed hi-pot (dielectric strength) at 1,500VAC within seconds. Arc-over across PCB surface.

Remediation implemented:

• Remediating work done includes adding an ADuM1441 isolation chip between the STM32L4 and DRV8876 circuits
• Adding a new DC-DC isolation converter to power patient-side circuitry from the isolated 3.3V rail
• Moving all VBUS traces 12mm back from any copper on the patient side
• Routing a slot (3mm wide) in the PCB under the isolation barrier as per 60601-1 Figure 9.

The final creepage distance after remediating work was 10.2mm which exceeds the 2MOPP requirement.

Pre-compliance scan (first attempt, open-frame PCB):

Root cause analysis:

The following is a summary of the root cause analysis and remediation of the EMI issue:

480 MHz:

Boost converter inside TPS63900 switching at 2.2 MHz → 480 MHz is the 218th harmonic. There is no EMI filter being used on the input side of the boost converter.

2.45 GHz:

The output matching network of the internal PA of the nRF52840 is not designed for harmonic attenuation. In addition, the antenna ground plane resonated due to insufficient stitching vias.

122 MHz:

The STM32L4 HSE has an 8 MHz crystal that is multiplied by 15.25 via the internal PLL. This has resulted in spurious emissions from the unshielded programming header (SWD).

Remediation:

To remediate these issues, we:

• Added a ferrite bead (BLM18HK102SN1D) and 100pF capacitors to the input side of the TPS63900
• Reworked the nRF52840 antenna matching network according to Nordic's reference design (added a 2.2pF shunt capacitor to the front of the PA to eliminate potential resonance).
• Added a 47pF capacitor to the SWD clock & ground near the programming header.
• Changed the stackup to move all high-frequency traces between layers (ground and supply layers).

Second pre-compliance (enclosed, with shielding):

Original Design:

BQ25601 charger designed to charge at 1C (500mA) with no temperature monitoring during fast charge; as demonstrated in a laboratory test, peak temperature of the cell during the charge was 68 degrees Celsius when a charging FET was shorted.

Requirement (clause 11.3.2):

Parts of a patient-worn device that are accessible to the patient can not be more than 48 degrees Celsius (applied part) or 43 degrees Celsius (non-applied part) and in contact with the patient.

Remediation:

• Reduced charge current from 1C (500 mA) to 0.5 C (250 mA) and thus increased charge time from 2 hours to 4 hours (within acceptable limits for intended use).
• Added 10K ohm NTC Thermistor (Murata NCP15XH103J03RC) to TS pin of BQ25601 and software cutoff set to 45 degrees Celsius.
• Redundant hardware over temperature protection provided via TPS25944 eFuse, which provides independent power cutoff to charger at temperatures greater than 50 degrees Celsius from any other source.

Measured in accordance with a single fault condition (shorted FET) following remediation: Peak temperature measured 44 degrees Celsius → meets applicable limits (48 degree Celsius limit for applied parts).

Problem observed by client: BLE connection drops when stimulation output is active (50mA, 80V compliance). Packet error rate jumps from 0.1% to 15%.

Root cause: The DRV8876 H-bridge switching at 250 kHz induced ground bounce on the shared digital ground. The nRF52840's 32 kHz RC oscillator (used for sleep timing) was being modulated by this noise, causing clock drift >50 ppm (spec is ±20 ppm).

Remediation:

• Split digital ground into two domains: GND_LOGIC (STM32L4 + nRF52840) and GND_PATIENT (DRV8876 + DAC8562)
• Connected the two grounds at a single point (star point) under the battery negative terminal using a 0Ω resistor (later replaced with ferrite bead BLM18HG601SN1D)
• Rerouted the nRF52840's 32 kHz crystal (not RC oscillator) external crystal (ECS-.327-12.5-34B) to improve clock stability

Result after remediation: PER back to 0.2% during worst-case stimulation. No BLE disconnections in 72-hour soak test.

The client's CM had no medical test fixtures. I designed and documented three mandatory tests for their production line:

CM training: I wrote a 12-page test fixture design document including pneumatic bed-of-nails for high-volume (10k units/month). The CM built two fixtures within 6 weeks.

Pre-submission Meeting (Week 16): Reviewer (TÜV SÜD) asked for clarification re 2 MOPP isolation explaining (i) digital isolator (ADuM1441) rated 2.5kVrms and (ii) IEC 60601-1 requires 4kV based on 2 MOPP with working voltage of 250VAC.

My advance preparation:

• Provided calculations showing that for a working voltage of 250VAC → 2 x MOPP requires 4kV dielectric (@250VAC) however, reinforced insulation would be compliant to IEC 60601-1 by virtue of 2 completely independent insulation systems.
• 1st = ADuM1441 internal transformer isolation; 2nd = physical PCB routed slot + 97m creepage.
• Submitted test report from 3rd party testing laboratory showing pass at 4kV for 1 minute (not just at 2.5kV).

Result: Within 3 days, reviewer accepted all of the above. No formal non-conformance issued.

Phil, UK

Certified PCB Design & Layout Engineer with 10+ years of experience, IPC-certified expertise in EMC/EMI-compliant, high-speed, mixed-signal PCB design, signal integrity, DFM, and hardware development.