Safety Concern: P-MOSFET Orientation in Power Path on Challenger LoRa Board

[anton]

Hello,

While testing several Challenger LoRa boards, I encountered a hardware issue that appears to be related to the orientation of the P-channel MOSFET (Q11) in the power path. The problem occurs when the board is connected to a fully discharged battery or a large supercapacitor on VBAT, and VBUS (USB power) is applied.

Under these conditions, the power section of the board overheats and emits a burnt smell, indicating likely damage to Q11 or surrounding components.

Upon reviewing the schematic, I noticed that Q11 is drawn with its source connected to VBAT, which is the typical orientation for a high-side P-MOSFET. However, in this specific application — where the goal is to disconnect the battery when VBUS is present, and enable it when VBUS is absent — this orientation causes a problem.

Here's what I believe is happening:
When VBUS is applied and VBAT is at 0V (discharged), the MOSFET's body diode becomes forward biased (VBUS pulls drain high, VBAT holds source low), causing uncontrolled inrush current into the empty battery or capacitor through the body diode. This is especially dangerous because there’s no current limiting in this path.

I compared this design to the Adafruit Feather RP2040 with RFM95 LoRa, which uses a similar setup — but their P-MOSFET is flipped, with drain connected to VBAT and source toward the load. While this goes against traditional high-side P-MOSFET usage, it cleverly ensures that:

- When VBUS is present, the external diode between mosfet's gate and source creates a voltage drop, turning the MOSFET off, and blocking current from VBAT.

- When VBUS is absent, but VBAT is charged, the body diode pulls the source high, and the gate is pulled to ground via a resistor, turning the MOSFET on, allowing current to flow from battery to system.

This reversed orientation solves the body diode inrush issue and prevents damage during power transitions.

Recommendation:
Given the current design, I believe the MOSFET Q11 should be reoriented with its drain to VBAT, or alternatively, a more robust protection scheme should be implemented to prevent damage when VBUS is applied with a discharged battery.

As it stands, this behavior poses a real risk in production environments, particularly in battery-operated systems that may frequently experience deep discharge.

Could you confirm if this issue is acknowledged, and whether a revision or fix is planned?

Thank you for your time, and looking forward to your response.

Best regards, Anton

[Pontus]

Hi Anton,
Thanks for getting in touch with us.

This is a remnant from the 0.1 schematics which still seem to be available on our gitlab repo. This was caught early during the review process and never made it into production. The products that we are selling have the mosfet correctly oriented.

We will make sure to update the repo with the correct schematics.

/Pontus

[anton]

Hi Pontus,

Thank you very much for your reply — I appreciate the clarification, and it's great to hear that Q11 is oriented correctly in production units.

However, the issue I described still persists on boards we recently purchased.
- Boards purchased: from ilabs.se
- Purchase date: March 2025
- SKU: 98-00195-1
- Example serial: 040141

I performed the following test:
Connected a bench power supply to the USB and GND pins (5V DC, current limit set to 3A).
While holding a finger on the power section of the PCB, I briefly shorted the BAT pin to GND, then released. This test imitates a battery at 0V.

Observed behavior:
- Current immediately jumped to 3A
- Voltage dropped to ~2.5V
- The microcontroller rebooted
- The power section immediately became burning hot

I have now repeated this test on three different boards, all with the same result.

This clearly shows that current into the battery is not limited, which could pose a risk under certain conditions, e.g. cold-start conditions, devices stored or shipped with depleted batteries, or systems using supercapacitors as backup power.

Could you please confirm this behavior on your side?

Thanks again for your support, and looking forward to your feedback.

Best regards, Anton

[Pontus]

Anton,

I will investigate to see what could be happening. In the mean time i pushed the long overdue 0.2 version of the schematic to the repo.

/P