Issue - Summary:

Under certain circumstances, when a high voltage (>6V) is applied to the solar input (e.g. plugging in an external 12V battery), voltage shoot-through can occur, resulting in instant irreparable damage to the BSoM unit.

This is not an issue for normal solar panels, not an issue with high rise time inputs (power supply which is turned on after electrical connection).

We do not expect this errata to cause an issue with normal solar panel operation

Issue - Details:

Observed Failure:

Board: v1.9

Failure was observed when a 12V battery was plugged into the solar input (to use as an additional power source). After this was done, the Particle status lights would either not light, or would produce unexpected outputs. It was evident that at least the BSoM was dead, presumably from an overvoltage condition.

Investigation:

The conditions were replicated (plugging in of 12V SLA battery) and the VUSB bus was observed, as that seemed to be the most likely source of over-voltage to the BSoM.

When this was done it was seen that a significant (~12V), but short lived (<2ms) spike was present on the VUSB rail when the battery was simply plugged in. See capture below.

TEK00010

It became clear that this was the source of the failure, but the question remained as to why, as this should not exceed the operating conditions of the system and the system had been tested with a 12V input.

Next the system was confirmed with the previous testing condition: 12V applied via power supply and switched using the soft-switch on the power supply (Rigol DP832). The results are this are seen below. The key aspect being that no voltage transient is observed (as expected), but also that the turn on time is relatively slow (>10ms rise time) compared with a mechanical contact.

TEK00011

Then we wished to confirm that the source of the error was, indeed, the rise time and not the input impedance of the battery vs the power supply. To test this the power supply (set to 12V, 3A limit) was connected to the solar input with a mechanical switch in series, then triggering that mechanical switch. The results are seen below, and we see the same voltage transient. This would seem to confirm the source of issue is the 'rise time' of the plug in action.

TEK00012

At this point, we also wished to confirm that the action of 'hot plugging' a solar panel in was not of concern. To test this we set the power supply to 7.2V (the expected max voltage of the solar panel output) and performed the same switch action. The results of this are shown below. We observed a less significant spike (as expected) which is within the absolute limits of the USB bus, so we confirm this action is not of concern.

TEK00017

Now that the cause of the issue has been isolated, we must look at cases when this might be mitigated, as this could potentially provide a workaround.

Conditions to test:

System switched on (VSYS energized)
USB plugged in (VUSB energized, simulated with power supply connected to USB input with 5.2V, 0.5A limit)

These tests are shown below, but in both cases the condition fails to mitigate the voltage shoot-through

TEK00014 TEK00015

With existing conditions failing to work around the problem, we now investigate modifications to the system to fix the problem.

The first attempted fix was the addition of a zener clamping diode between the VUSB bus and GND. A 5.6V zener diode was used for this test, and the diode was soldered as close as possible to the junction of VPRIME and VUSB. The result are shown below, but this had almost no effect on holding the rail within spec.

TEK00018

Next, we tested replacing the MOSFET used for switching (Q2) with a low voltage, high power, schottky diode (specifically: CMSH3-40MA). This (along with the addition of a 10kΩ resistor between VUSB and GND to combat reverse leakage currents in extenuating situations). This worked well and resulted in a voltage peak well below the limits for the USB bus.

See v1.9 circuit and modified circuit below

circuit circuit (1)

The results of testing with this circuit are shown below.

TEK00020

Circuit Modification (Patch):

circuit (1)

The patch used in testing is effective, but is not ideal is it reduces efficiency in the USB charging situation (which is why this design was initially rejected). As a result it was important to test the thermal properties of the new system and insure the power dissipation was not too high.

To test this, the system was connected to a set of moderately discharged batteries, such that a high current would be used for charging. The USB input was sourced from a power supply (RIgol DP832) and case temperature of the diode was measured by contact with a K-Type thermocouple, measured by a high precision DMM (Keithley DMM6500). This test was performed in open air in a lab at ~20°C air temperature. The test was run for approximately 10 minutes, after which time it was evident that equilibrium had been reached. The results of this are shown below, with a max temperature of ~60°C (+40°C from baseline).

TempPlot

While this system stays within operational limits at room temperature, there is concern that this could exceed operational conditions if charging is done when the logger is in the field and heated by the environment before additional charger heating occurs.

Because of this, this modification is acceptable, but not recommended unless logger is required to be powered by external 12V battery. In other cases, the v1.9 board should remain as is.

Design Update:

TBD

GEMS-sensing / Project-Kestrel

Solar Input High Voltage Failure #20

Issue - Summary:

Issue - Details:

Observed Failure:

Investigation:

Circuit Modification (Patch):

Design Update:

Board modification details

Pad Diagram

Component Diagram

Test Point Diagram