I have a Pro Mini setup, communicating via a Sim800L and reading/writing digital high and lows and an analogue voltage read, so no sensors at all. My rather extensive code runs perfectly stable on a 5V Pro Mini. For power saving reasons and to communicate with the Sim800L at safer voltage levels, I need it to run on a 3.3V Pro Mini.
When installed on a 3.3V Pro Mini, with a 16850 battery connected to RAW, the program goes haywire. I crashes at any random opportunity, not only when communicating with the SIM800L, but also within the normal loop. I don't have an oscilloscope, but my multimeter shows a stable supply voltage and a stable 3.3V output on Vcc. As soon as I switch to supplying 4.2V on VCC, everything goes back to its stable normal. The Sim800L is supplied directly from the battery in all setups.
That's because the regulator input voltage need to be 1V higher than 3.3V in order to meet the regulation specifications otherwise operation will be unpredictable.
Since the SIM800L is running on direct battery power, you should do the same with the Pro Mini. Feed the battery into the Vcc pin instead of the RAW pin. What's probably happening on RAW is sagging voltage under any significant load because you aren't meeting the dropout requirements of the regulator on the Pro Mini. And that would only get worse as the battery discharges over time.
You might even consider removing the regulator if it's not needed. That would save a few 10s of microamps, and extend battery life.
Which board are we talking about?
The Sparkfun ProMini 3.3volt has a MIC5205 regulator, with a dropout of ~0.1volt.
A single 3.7volt LiPo should not be a problem.
I'm guessing that poor decoupling, flimsy wiring and/or a fake battery are the culprit.
Note that a SIM800 draws 2Amp peak during transmit. That has to come from somewhere.
Leo..
The board is a chinese knock-off ("The Simple"?), just as the 5V version. I tried a 470uF capacitor on the SIM800L power supply, no change. Vcc reads 3.3V stable on a multimeter when feeding 4.2V into RAW, so the regulator seems to be working. Also, I can't tie the crashes with code calls from the Arduino. It is possible that they happen with normal background network communication, but I feel that they happen too often for that. I tried a 26650 for power supply, which normally supplies 4x 5V Pro-minis with 4x Sim800Ls simultanously on the bench. Same unstable result. For power supply between the 18650 charger, battery and SIM800L I used 18AWG, and all the wire runs are short, but also squeezed together in a very small space. The same physical setup works when I supply directly on VCC. But yes, it must be a regulator issue, as everything returns to normal when I do this.
Does the clone have a voltage regulator with three pins and a tab, or a tiny 5-pin regulator.
What does VCC read with 3.4volt on RAW (what is the regulator's true dropout voltage).
Leo..