Schematic Review & Critical Additions
Schematic Review — Critical Additions Before Assembly
Section titled “Schematic Review — Critical Additions Before Assembly”Date: 2026-04-23 Status: Client confirmed 3 capacitors are on board. This review identifies additional critical components.
Client schematic
Section titled “Client schematic”
Original breadboard-style schematic shared by client. Visible elements: servo (top-left), fuse (yellow rectangle), XIAO (large gray square), RTC and Notecard on the I2C bus (SDA/SCL), battery connector (black + red wires, top-right).
Summary of Findings
Section titled “Summary of Findings”✅ Good:
- I2C topology with RTC and Notecard on same bus — clean
- Fuse on input — basic protection present
- Notecard uses built-in GNSS (no separate GPS chip)
- VMODEM_P fed directly from battery (correct per Blues datasheet)
🔴 Critical gaps:
- Bulk capacitors on servo rail (separate from Notecard caps)
- P-channel MOSFET power gate for servo
- RC snubber on motor terminals
- Ferrite bead on servo power line
1. Bulk Capacitors on Notecard VMODEM_P
Section titled “1. Bulk Capacitors on Notecard VMODEM_P”Problem
Section titled “Problem”Blues Notecard LTE startup: modem attach causes 2A peak in 20–100 ms. Without local capacitor, battery’s internal resistance causes voltage sag from 4.0V → 3.0–3.2V. XIAO detects brownout reset → undefined state → won’t boot.
Solution
Section titled “Solution”Add capacitor stack within 5 mm of Notecard VMODEM_P pin:
- 470 µF / 10V electrolytic (low-ESR) — bulk energy
- 10 µF / 16V ceramic X5R (1210) — medium-frequency response
- 100 nF / 25V ceramic X7R (0603) — high-frequency noise
Part numbers:
- Panasonic EEE-FK1A471P (470µ) or Nichicon UWT1A471MCL1GS
- Murata GRM32ER61C106KA12L (10µ)
- Murata GRM188R71E104KA01D (100n)
Why three values: capacitors work on different time scales — electrons move fast at nanoseconds. Three-value approach covers 1ns to 1ms bandwidth.
2. Bulk Capacitor on Servo Rail (Separate!)
Section titled “2. Bulk Capacitor on Servo Rail (Separate!)”Problem
Section titled “Problem”Servo startup draws 300–700 mA for 1–3 seconds. Without local cap on servo power rail → voltage droop on VBAT → affects Notecard and XIAO.
Solution
Section titled “Solution”330 µF electrolytic + 10 nF ceramic on servo V+ after the MOSFET, before motor.
3. P-Channel MOSFET Power Gate for Servo
Section titled “3. P-Channel MOSFET Power Gate for Servo”Problem
Section titled “Problem”Servo trickle current for “holding position” = 40 mA continuous. Over 24 hours: 960 mAh/day wasted just sitting idle. This destroys battery autonomy.
Solution
Section titled “Solution”P-channel MOSFET between battery and servo:
VBAT ──┬── Source ──┐ │ │ P-MOSFET │ ├── Drain ── Servo V+ │ │ │ ┌── Gate │ │ ├──[10 kΩ]── Gate (pull-up = MOSFET OFF by default) │ │ │ └── to MCU GPIO (LOW = turn servo ON) │ └── + 330 µF bulk cap on Servo railRecommended MOSFET: AO3401A (Alpha Omega, SOT-23, -30V/-4A, 50mΩ) or Si2305CDS
Gate pull-up: 10 kΩ to VBAT (ensures safe OFF state on power loss)
Result: Servo 0 mA when idle. ~960 mAh/day recovered → 5–10× better autonomy.
4. RC Snubber on Motor Terminals
Section titled “4. RC Snubber on Motor Terminals”Problem A: Back-EMF Spikes
Section titled “Problem A: Back-EMF Spikes”When servo stops, coil inductance generates 50–100 V spikes that can:
- Exceed MOSFET Vds max rating (-30V)
- Couple RF noise into antennas via power rails
- Cause I2C communication errors
Problem B: Broadband EMI
Section titled “Problem B: Broadband EMI”Commutator arcing generates 1–1000 MHz noise that directly degrades GPS and LTE reception.
Solution
Section titled “Solution”10 Ω resistor + 10 nF capacitor soldered directly on motor terminals:
Motor(+) ──┬── [10Ω 1W] ── [10nF/100V] ──┬── Motor(-) │ │ (one RC pair per motor)Part numbers:
- Vishay CRCW121010R0FKEA (10Ω 1W 1210)
- Murata GRM32DR72A103KW01L (10nF 100V 1210)
The resistor limits peak current, capacitor absorbs energy, together smoothing the spike from -100V to -15V (safe).
5. Ferrite Bead on Servo Power Line
Section titled “5. Ferrite Bead on Servo Power Line”Problem
Section titled “Problem”Motor noise couples onto servo power wire and radiates from the wire as an antenna, directly into GPS/LTE receivers.
Solution
Section titled “Solution”Ferrite bead on servo V+ after MOSFET Drain:
MOSFET Drain ── [Ferrite bead] ── Servo V+Part number: Murata BLM18KG471SN1D (0603, 470Ω @ 100MHz, 2.5A max)
At low frequencies (DC, 50 Hz PWM) → impedance ~0 Ω, no loss. At high frequencies (>1 MHz) → impedance 400–1000 Ω, noise absorbed as heat.
Optional: Ferrite ring on servo cable (if external) — Fair-Rite 2643167251
6. I2C Pull-Up Resistors
Section titled “6. I2C Pull-Up Resistors”Problem
Section titled “Problem”I2C is open-drain (pull only to GND, can’t actively push HIGH). Lines must be pulled to VCC by resistors.
Solution
Section titled “Solution”4.7 kΩ on SDA, 4.7 kΩ on SCL (one pair):
+3.3V ──┬─[4.7k]── SDA │ └─[4.7k]── SCLPart number: Yageo RC0603FR-074K7L
Important note: Notecard has built-in 2.2k pull-ups. If XIAO also has them, parallel 2.2k + 4.7k = 1.4k (too strong, bus latency). Check XIAO datasheet first. If XIAO has NO pull-ups → add 4.7k. If both have → don’t add external.
7. Accelerometer LIS2DH12 (Recommended, Not Critical)
Section titled “7. Accelerometer LIS2DH12 (Recommended, Not Critical)”Container stationary 99% of the time. GPS on 24/7 = 432 mAh/day waste. With accelerometer + wake-on-motion: GPS only when moving → save 400+ mAh/day.
STMicroelectronics LIS2DH12 on I2C (addr 0x18 or 0x19):
- I2C add-on (Notecard already 0x17, RTC 0x68 — no conflict)
- 2 µA in ODR 1 Hz sleep mode
- Programmable motion interrupt
Part number
Section titled “Part number”STM LIS2DH12TR (~$1.50)
8. Fuel Gauge MAX17048 (Recommended)
Section titled “8. Fuel Gauge MAX17048 (Recommended)”Battery voltage-based SOC calculation is ±5% inaccurate. MAX17048 uses impedance tracking → ±1% accuracy.
On I2C (addr 0x36), read via Notecard’s card.voltage API.
Part number
Section titled “Part number”Analog Devices MAX17048G+T10 (~$3.50)
9. Reverse Polarity Protection (Recommended)
Section titled “9. Reverse Polarity Protection (Recommended)”Problem
Section titled “Problem”If someone plugs battery backwards at manufacturing → all chips burn instantly.
Solution
Section titled “Solution”P-channel MOSFET as diode with 0 mV drop (vs. Schottky’s 300 mV):
Battery(+) ── Source ──┐ Circuit V+ │ ├── Drain ── (rest of circuit) │ Gate ── [100 kΩ pull-down to GND]Part number: Infineon IRLML6402TRPBF (~€0.40)
Checklist Before Soldering
Section titled “Checklist Before Soldering”- 2× Panasonic EEE-FK1A471P (470µ Notecard + servo)
- 4× Murata GRM32ER61C106KA12L (10µ ceramic)
- 2× AO3401A P-MOSFET (servo gate)
- 10× 10 kΩ 0603 (gate pull-ups)
- 4× Vishay CRCW121010R0FKEA (10Ω snubber)
- 4× Murata GRM32DR72A103KW01L (10nF snubber)
- 4× Murata BLM18KG471SN1D (ferrite bead)
- 4× Yageo RC0603FR-074K7L (4.7k I2C)
- 1× LIS2DH12TR (accelerometer, optional)
- 1× MAX17048G+T10 (fuel gauge, optional)
- 1× IRLML6402TRPBF (reverse polarity, optional)
Total cost: ~€15–20 per board
What to Tell Customer
Section titled “What to Tell Customer”“Good that you have 3 caps. Before soldering, add these elements: (1) separate capacitor set on servo rail, (2) P-MOSFET power gate (AO3401A) to cut servo idle current, (3) RC snubber directly on motor terminals, (4) ferrite bead on servo power line. These are critical for reliability and autonomy. Optional but highly recommended: accelerometer + fuel gauge. Total 5 additional components, cost ~€5–10.”