I'm sorry that I can't immediately set up a test environment to test the code up but I will say that the 3.3v from the Nano could be a problem in that it is usually derived from the usb/uart chip and is normally no more than 50mA.
Can you get a repeatable situation if you completely disconnect the Nano, wait a few seconds, power it up, then wait a few more seconds then transmit a message.?