Detecting the NRF24L01

Can't claim expert status since I've only used this chip on the one protocol, but I'll start things off. When I toyed with this before I used this:

Yes, it's a bit unfriendly for identification. May be there are registers which has unique read-only bits and will ignore writing to them? I will take a look at the datasheet.

How about FLUSH_TX, FLUSH_RX and reading STATUS with NOP and with regular read from STATUS register? They should match, and low nibble should read as 0xE.

Also, if you set one of the bits in EN_AA (auto-acknowledgement), EN_CRC bit in CONFIG is "forced high", so it should "resist" writing to.

I am far from my nRF24L01, so these are only results of theoretical research of datasheet plus some memories from watching SPI traces.

Yes, that code was for checking a known nRF was responding. I don't know much about the other radio chips so don't know how to distinguish among them.

The Beken/Hope clones of the nRF have a device ID in the extended register set. Haven't seen anything similar for the nRF itself.

nRF will always return the content of the status register when sending a command, regardless which command you send:
"The STATUS register is serially shifted out on the MISO pin simultaneously to the SPI command word shifting to the MOSI pin"

Datasheet mentions that sending 8 '1's = 0xff is a NOP, which is to be used to read only status register back.
Reset value of status register is 0xE (= RX FiFo Empty).

Programming a register and expecting the very same value back is not such a good proof for a specific device, this even works on most of the registers of my good ol' VIC (C64 Graphics chip)

The point of my method is interdependence of the registers - you write something into one register and expect some bits in another. It works only if the chip does exactly the same procedures, especially WRT to (hypothetical, not tested but deduced from datasheet) write resistance of a EN_CRC bit. I need to verify this, which I'll do in several hours.