RF24::begin(void) always returns false

[wmarkow]

I have noticed that the RF24::begin(void) always returns false which may confuse a developer who is using this library for the first time. I think commit with SHA b1b8cad comments out a little to much in the RF24::begin(void) function.

I can provide a very simple fix for that.

[Avamander]

It should be tested before merging by people with different hardware.

[TMRh20]

The mess of code in begin() is due to some suggested changes to detect if setup was successful, but it never worked quite right, so I think removing that code is fine.

[wmarkow]

Thanks for merging my pull request. So for now we have the following piece of code:

uint8_t setup=0;
...
if( setDataRate( RF24_250KBPS ) )
{
  p_variant = true ;
}
setup = read_register(RF_SETUP);
...
// if setup is 0 or ff then there was no response from module
return ( setup != 0 && setup != 0xff );

My source code static analysis and datasheets for both RF24L01 and RF24L01+ shows that after hardware power up (when the power supply is attached) begin() method will return true (for both RF24L01 and RF24L01+ chips) if modules are wired correctly and are not damaged.
It should be ok even if someone will call begin() method multiple times (and sets by the way the power level RF_PWR to '00'). For both chips the setup value should be different than 0x00 or 0xff. Some issues me come up when we start touching the LNA_HCURR bit in RF24L01 (but for now we do not do that).

Another question pops up, related to this code:

// if setup is 0 or ff then there was no response from module
return ( setup != 0 && setup != 0xff );

So there is an assumption: when read a 0x00 or 0xff value over SPI it may be a clue that there was no response from the module. I have noticed on Arduino: when the module is not wired at all the SPI reads 0xff (like the MISO line was pulled up). However my NodeMCU sometimes reads 0x00 (like MISO was pulled-down) or sometimes returns 0xff (like MISO was pulled-up). What do you think?

[wmarkow]

I think I can close this issue now. Now there is a new API method available that is able to check if the nRF24 module is connected, correctly powered up in responsive.