PIC reset problem

[POPE19]

It is connected to vdd (5v) (output from a 5v Voltage regulator) through 47k ohm resistor in series.

[Ttelmah]

The serial needing 1.9 seconds before it starts working, is 'odd'. Normally perhaps 20 or 30mSec, and certainly not over 1 second max. This says to me there is something strange happening.

How is the serial connection actually being done?. Are you using real serial or perhaps a USB-serial adapter?.

If the latter, are you unplugging this when you attempt to power the board down?.

I'm starting to suspect that your board is actually being powered by the adapter.

[POPE19]

I am using ch340g usb to serial (TTL) module for serial connection. This is connected to usb port on my computer and board side connections are as follows:

[Ttelmah]

What is happening, is that the RX line coming into the PIC is trying to go up to just under 5v, and this via the protection diode in the PIC is pulling the supply rail up.
On either chip it is a dangerous design. On this one it is pulling the supply rail high enough that the chip is not completely powering down. This chip works down to 1.8v. The F917, 2v. I'd guess the supply rail is keeping up to perhaps about 1.7v. Problem is this risks damaging the chip in the CH340. It is driving this line into a near short, and long term could easily be damaged by this....
The other possibility is the 917 board is drawing a little more power, which will also allow it to power down. Either way the USB driver will be being overloaded...

Add a series resistor in the receive data line. Enough so that the line is not overloaded, and no longer pulls the supply rail up to the point where the chip doesn't think it has gone off....

[POPE19]

Yes again. It works. I have used 1k ohm series resistor and it worked. It is driving vdd at 1.6 volts now well below 1.8v.

And this solves that delay serial print problem too. Now with 300 msec delay it is serial printing "started " without skipping s letter.

Thanks for the explanation.