I’m trying to run a Beagleboard Black from solar power, and I set it up per the attached schematic.
The analog inputs are used to measure the BB system voltage, the battery voltage, and the current from the solar panel by measuring the voltage either side of a one ohm resistor in the ground lead (500mA from the panel should give 500mV, less than the 1.8V maximum)
I’d run it on USB power from a PC for several days, then from a car 12V USB adapter on the 12V battery, with no problem, with a proto board cape to measure
the battery voltage. Also with a 5V supply via the coax power connector.
Then I connected the solar panel, nominal 4W 12V, and left it outside while the sun rose in the sky. The BB was OK for a while, then died. It appears to be completely defunct - no flashing LEDs, on whatever supply (battery, USB etc.)
When I checked voltages, it did seem that there was a bad connection to the battery terminals. The no-load output voltage of the panel is in excess of the nominal 12V, so the obvious suggestion is that the input voltage overpowered the BB. But with the BB and battery disconnected, the USB adapter was still producing 5.25V output with an input voltage of some 20V, and I assume (though I have no specs) that this is normal - a car supply voltage can rise to at least 14V under charge, so a car adapter should be able to handle more than 12V.
That, and the “some 20V” open voltage is very close to the 22.5v maximum voltage on the AIN5 devider.
Also the 5.25V supply is awfully close to the maximum of 5.25V (5V +/-.25V)
On a side note: Car battery USB converters are mainly intended for charging USB devices, I found the power output of several of the ones in my possession to be not that stable.
LP
It would not be unheard of for an unloaded panel to go above twice the nominal output voltage. Some of our panels can go above 30v open circuit, but our panels are around 200W each or slightly better.
So I’m no electronics engineer, but I remember sometime back our system having problems with powering voltage / current meters off the same power supply. Powered directly off our battery bank. The symptom was we blew out our voltage meters, and the cure was to isolate the meters. Current still is powered via the battery bank directly, but the voltage meter uses a walwart power supply coming off our inverter AC lines.
The voltage sequencing mentioned by Gerald is big and is probably the real failure mode. In this design, the AIN pins will see voltage long before the CPU/PMIC (Power Management IC) is up and running. Basically, when a voltage is applied to the analog pins (or any IO pins for that matter) before the IO rail is up, the chip tries to power itself through the pin clamping diodes and can really cause havoc on the silicon. The chip just wasn’t made to be powered this way.
Did you check that the 1Ohm resistor didn’t blow or that the voltage across it didn’t exceed the limit for the ADC inputs? If it does fail open for some reason, the voltage would far exceed the max voltage on the Analog pins. Might want to add some protection there in case that resistor ever does blow, or the voltage across it exceeds the max analog voltage.
Also note that the 1Ohm resistor isn’t just powering the BBB, it’s charging the battery! Chances are the solar panel was putting too much juice into the battery and caused the 1Ohm resistor to have a much larger voltage across it.
You might move the 1Ohm resistor to the ground leg between the battery and the solar panel, instead of between the solar panel and the battery.
This was a big reason why I chose to use an I2C external ADC for measuring raw battery voltage in a project of mine. Gating that voltage from hitting the pins prior to PMIC startup was way more complicated than hooking up a 4 channel ADC on the I2C bus and using that powered by the 3.3V rail. It’s easy to overlook that need to keep inputs isolated until power is fully on for the BBB.
1. The voltage sequencing mentioned by Gerald is big and is probably the real failure mode. In this design, the AIN pins will see voltage
long before the CPU/PMIC (Power Management IC) is up and running. Basically, when a voltage is applied to the analog pins (or any IO
pins for that matter) before the IO rail is up, the chip tries to power itself through the pin clamping diodes and can really cause havoc
on the silicon. The chip just wasn't made to be powered this way.
2. Did you check that the 1Ohm resistor didn't blow or that the voltage across it didn't exceed the limit for the ADC inputs? If it does
fail open for some reason, the voltage would far exceed the max voltage on the Analog pins. Might want to add some protection there in
case that resistor ever does blow, or the voltage across it exceeds the max analog voltage.
3. Also note that the 1Ohm resistor isn't just powering the BBB, it's charging the battery! Chances are the solar panel was putting too
much juice into the battery and caused the 1Ohm resistor to have a much larger voltage across it.
4. You might move the 1Ohm resistor to the ground leg between the battery and the solar panel, instead of between the solar panel and the
battery.
This is a 4W panel, and a fairly hefty 1ohm resistor - 5W or something. It can't have blown.
Mmph, I think I've been stupid. The way I originally drew the circuit, it looked like AIN1 would be at +200mV, but now I think it would be at -200mV. I didn't have any more 150K in my box and thought I could put the resistor in the ground leg instead of the +ve leg.
With the ground connection to the battery loose, AIN5 could have been overpowered at about 3V, but limited to less than 200uA by the 150K resistor, while AIN1 had no current limit.
I'd have expected a problem on an AIN pin to have just blown that input, rather than the whole board, though.