indigou:
From what I see, seems like the biggest problem is that the nrf is not able to receive simultaneously and if I use the master/slave with identifiers should work well. However I am not sure how much delay could work, I need it to be as low as possible, I am recording data at 50Hz and I feel like this approach (although I don't think there is other option) is going to miss some data.
An nRF24 message plus acknowledgement takes a few millisecs so the master could possibly deal with about 200 per second.
However if there is no co-ordination of the sending by the two slaves the more frequently you send messages the more likely there will be a data collision. That then requires BOTH slaves to re-transmit their messages after a suitable interval with a scheme that makes it much less likely there will be a second collision over the same message. You should be able to envisage how the time taken to deal with collisions will have a huge impact on the total throughput of messages.
As a separate issue (and as I mentioned earlier) including an identifier within each message is a very simple way to enable the master to know who sent the message.
...R