autonym wrote:Does this mean you can power the Grand Central with a regulated 3.3V source to the output of the GC's 3.3V regulator?
Yes, though it's a case of using an external regulator instead of the onboard one.
autonym wrote:I'll be using 4x 12V cordless drill batteries (Milwaulkee M12's) in parallel to power the project, but they're dumb batteries that if I run too low/hot, they'll break, so I'm adding voltage, current (INA219), and temperature (DS18B20) monitoring to them, and also the ability to shut each battery in the array off with an N-channel power MOSFET (IRLB8721PBF) if they get too low. I plan to be able to hot-swap individual batteries out, or even round-robin them a bit to share the load.
The only part of that list that doesn't sound like a good idea is putting the batteries in parallel.
All power sources can be modeled as a perfect voltage source (capable of holding the same voltage no matter how much current it takes) connected to the load through a resistor. The source's ability to maintain its output voltage under load depends on the value of that theoretical resistor.. coin cells get up near 100 Ohms when you draw more than a couple of milliamps from them, while LiPos can stay down in the milliohm range when delivering several amps.
So when you put two power sources in parallel, you have the equivalent of two perfect voltage sources connected through two resistors.. low-value resistors if you're working with LiPos. That's fine if both power sources have exactly the same voltage at all times, but that never happens in reality.. there's always some small difference. And since LiPos have such low internal resistance, even a few millivolts of difference can make the higher-voltage LiPo try to push a large amount of current backwards through the lower-voltage one.
At first glance that looks like the higher-voltage LiPo is just recharging the lower-voltage one, but LiPos are picky about their charging process. Running normally, they skate along the line between 'energy storage' and 'incendiary device'. There's no clearly observable change between a LiPo that's charging and one that's overcharged to the point where it's about to catch fire, so their charging has to be managed by a circuit that maintains tightly controlled conditions.. a charger whose voltage sensor is off by a few millivolts can wreck your day (and insurance premiums).
You can put circuits between the LiPos that manage them properly, including step-down voltage converters. Even then, putting step-down converters in parallel can cause weird problems if they aren't designed to work that way. Round-robin sequential sharing or straight switchover when one LiPo is mostly discharged are good strategies.
autonym wrote:The problem, naturally, is how to I kick-start the board to boot-up if all the batteries in the array are off because the MOSFETs don't have power to trigger them on.
That's a logic problem: control the first battery's connection with a NAND gate that reads all the other battery connections. If no other source has a connection to the Grand Central, connect the first one. If any other battery is connected, disconnect the first one.
You can use the gate voltages of the mosfets instead of the battery voltages themselves.
autonym wrote:I have a PowerBoost 500 Charger coming that I hope can solve the problem by allowing me to have it charge an attached L/Ion battery from the 12V array
Don't do that.. the PowerBoost doesn't know how to handle battery voltages higher than 5V. Higher voltages equal property damage.
You could use a Verter instead:
https://www.adafruit.com/product/2190
because that one knows how to switch between being a step-up converter and a step-down converter, and can handle 12V input.
autonym wrote:I thought I could connect the 5.2V from the PowerBoost to the USB power line on the Grand Central but I'm worried about back-pushing voltage from the PowerBoost to my computer's USB when I connect it to write updates to my code
The Grand Central has a circuit that isolates the USB-5V line from the 5V power bus if external power is connected, but it's still a good idea to use a 3.3V external supply.
autonym wrote:If I send more than 3.2A of current through a stock INA219 with the 0.1Ω resistor, will it hurt the sensor, or simply max out the value reported?
It would max out the readings, but wouldn't damage the INA219.
autonym wrote:Oh, one more question - the Grand Central power regulator is mentioned at 7-9V, but I've seen people mentioning it can take 12V.
Do I need to buck my 12V down to 9V before giving it to Grand Central?
'Need' is a strong word. 'Greatly prefer' covers the situation better.
The underlying problem is that the energy from any voltage you don't use has to go somewhere. If you run it through a linear regulator like the NCP1117 connected to the Grand Central's Vin pin, it gets burned off as heat in the regulator.. linear regulators are basically smart resistors that adjust their own resistance to keep the voltage at their output pin stable.
If you send 12V in and get 5V out, 60% of the power from your battery is lost making the NCP1117 hot before it ever gets to the microcontroller. Then you lose another 35% going through the 3.3V regulator. Only about 27% of the power that leaves the battery reaches the SAMD51. That isn't fatal, but the efficiency sucks.
Switching regulators can convert voltage to current and vice versa by temporarily storing the energy in a magnetic field. Converters that reduce voltage and increase current are easier to design and optimize than ones that reduce current and increase voltage, to the point where most step-down converters are better than 95% efficient these days.
So instead of drawing 100mA @ 12V (1.2W) and burning off 700mW of it as heat to get 100mA @ 5V (500mW) from a linear regulator, a 95% efficient switcher will draw about 44mA @ 12V (525mW) and give you the same 100mA @ 5V (500mW) output. To get 100mA @ 3.3V (330mW), the switcher would draw about 29mA @ 12V (347mW) from the LiPo.
With good step-down converters being so easy to find and use these days, the balance of design choices has shifted to having to justify the decision to use a linear regulator for any large difference between the input and output voltages. If you're only using 10mA, you can say, "enh, what the heck", but if you get much higher than that, it's worth weighing the pros and cons of a switcher.