A lot of work has gone into this. They are using a trinket m0 board. Custom designed for the layout and what not on the board. Multi rounds of power solutions.
Praise from the master themself, thanks buddy!
Pinout is as follows, but I had planned to modify your code instead of asking you to do the work so don't feel obligated in any way.
LED is a 1mm x 1mm RGB LED (bugger to hand solder) and is common anode, so lines must be pulled low to turn on and high to turn off.
D7 = Red
D8 = Green
D2 = Blue
I have D0 (the + button) accepting 3.3v through S1 whenever pressed which also turns on LDO to wake up the MCU. The idea is whenever the MCU starts, it checks to see if D0 is HIGH, and if so it increments the payload number, writes it to eeprom, skips injection, blinks blue to indicate the number of the payload selected then goes to sleep. The idea is to hold the button until the LED stops flashing.
Also have A0 reading battery voltage through a voltage divider. R1 is 2.5MΩ and R2 is 1MΩ, which converts 0-3.85v from battery to 0-1.1v at A0 to match the internal 1.1v reference of the ATSAMD21. The plan is to have a low battery LED indication of 0.5 seconds on boot. I have yet to map out the voltages I want to use for battery status, but colors would be green for healthy, yellow for warning and red for critical battery level. On red I would skip injection and just flash red a few times before putting MCU to sleep. Reason being, injection can fail if voltage is too low, which causes the Switch to freeze, requiring a 20sec power button shutdown. Advanced users who want to risk it for a few more injections can modify the firmware.
Schemmy attached. I've been told my schematics look awful, so sorry about that.
void battery_check() {
pinMode (ONBOARD_LED, OUTPUT);
pinMode (REDLED, OUTPUT);
pinMode (GREENLED, OUTPUT);
pinMode (BLUELED, OUTPUT);
delayMicroseconds(500000);// allow to stabilise
battvalue = analogRead (BATTERY_LEVEL_CHECK);
if (battvalue >= 900) {
digitalWrite(GREENLED, LOW);
flatbatt = false;
} else if (battvalue >= 860 && battvalue <= 899) {
digitalWrite(REDLED, LOW);
digitalWrite(GREENLED, LOW);
flatbatt = false;
} else {
digitalWrite(REDLED, HIGH);
flatbatt = true;
}
delayMicroseconds(500000);
digitalWrite (REDLED, HIGH);
digitalWrite (GREENLED, HIGH);
digitalWrite (BLUELED, HIGH);
if (!flatbatt) {
return;
} else sleep (-1);
Right... Been working on the battery code.
You are using a potential divider to chop the 3v to 1100mV, correct?
We have 10 bit resolution on the analogue pin, so it is approx
battvolt / 3 = 1000 - this is under10241023 our maximum
2.7v / 3 = 900 - amber warning?
2.580v / 3 = 860 - red warning / shutdown?
Obviously needs testing...
EDIT: Something simple like this...
Code:void battery_check() { pinMode (ONBOARD_LED, OUTPUT); pinMode (REDLED, OUTPUT); pinMode (GREENLED, OUTPUT); pinMode (BLUELED, OUTPUT); delayMicroseconds(500000);// allow to stabilise battvalue = analogRead (BATTERY_LEVEL_CHECK); if (battvalue >= 900) { digitalWrite(GREENLED, LOW); flatbatt = false; } else if (battvalue >= 860 && battvalue <= 899) { digitalWrite(REDLED, LOW); digitalWrite(GREENLED, LOW); flatbatt = false; } else { digitalWrite(REDLED, HIGH); flatbatt = true; } delayMicroseconds(500000); digitalWrite (REDLED, HIGH); digitalWrite (GREENLED, HIGH); digitalWrite (BLUELED, HIGH); if (!flatbatt) { return; } else sleep (-1);
instead of delayMicroseconds, going to use a
void pause (int pausetime){ }
with digitalReads on D0 pin
routine so we can detect if D0 is held and action accordingly.
Hope I am understandable
Unfortunately I don't have the money for injection molding right now, as the DI would need 5 molds which would be over $10000. Right now I'm printing on a pair of Anycubic Photon DLP printers. The resin is quite strong once cured. The downside is that it's quite brittle and can't take as much flex as plastic can before breaking.
Looks pretty easy to read at first glance, I'll poke at it tonight once I get home from work.
Thanks so much for your help!
Unfortunately I don't have the money for injection molding right now, as the DI would need 5 molds which would be over $10000. Right now I'm printing on a pair of Anycubic Photon DLP printers. The resin is quite strong once cured. The downside is that it's quite brittle and can't take as much flex as plastic can before breaking.
I look like I'm treating it delicately on the video because I have a camera in my face and can't see very well what I'm doing.
Looks pretty easy to read at first glance, I'll poke at it tonight once I get home from work.
Thanks so much for your help!
By my calculations D+ can supply about 2mA (full speed uses 1.5k pull up at ~3v) when set as USB slave, which is more than enough to enable the LDO. It's tested and working great as you can see from my videos.
The AP7331 is indeed momentary. My thinking for that is, I don't want the user to have to press any buttons to inject a payload. You just plug it in, D+ enables the LDO, and boom, injection. RCM stops, D+ goes low, DI shuts itself off. This greatly helped with battery life.
D+ actually has more than enough current capability to turn on the LDO.
It might be because I work automation by day and have seen some pretty shady code controlling life safety equipment, but I don't really like the idea of giving the MCU control of something as important as the LDO EN pin. Any programming error or abnormal condition that causes the MCU to freeze or keep that pin high means the potential for a dead battery and a very unhappy user with a Switch stuck in RCM. I know technically that should never happen, but I prefer to do it with passives.
Keeping the MCU on for D0 isn't really a big deal since injection shouldn't happen if it's pressed, so USB host would be a big battery drain for nothing. (11mA draw!) Instead use can just hold down the button until they get visual feedback of which payload is selected. Bootloader runs for well under a second so this shouldn't be inconvenient, and gives the user time to let go of the button if they only pressed it by mistake, instead of having to cycle back to the payload they were using (wasting even more battery).
Last advantage of using D+ is this: to get into firmware update mode, the user can hold + button (D0) and hit reset twice, and the MCU will then keep itself on with D+. Plug it in and update firmware, and once done, DI shuts itself off because no more USB slave mode.
It's all in the name of maximum battery life and minimum complexity for the user who just wants to kill some goombas.
Anyways, those are my thoughts and self-rationalizations, if I'm amiss in my thinking by all means throw some sense at me.
If it was me, I'd have a rapid blinking led and five the user the blinking window to save it.Thanks! Thats' exactly right.
About the only thing I was debating on is when to save the new payload value to EEPROM; write before LED's flash and it might get changed unintentionally, but write after LED's flash and it might not write before user releases button. I think the latter is the best way as you suggested. Writing to EEPROM is quick, and I have a cap between EN and GND of the LDO that will keep it on a little bit longer (400ms ish).
Agreed on the E18 vs. G18. I didn't see the point of an external crystal. If the internal oscillator is good enough to keep USB happy, it's good enough for whatever I'd use it for. No need to build an atomic clock into a payload injector.
MATTYTROG
