anelok: revisiting Bluetooth, Don't want no ...

Werner Almesberger werner at
Fri Nov 22 19:42:23 EST 2013

... short FIFO, man ! (With apologies to "20 Fingers")

There are more chips to look at than just TI's. Here is a very nice
treatise on doing BTLE with hardware that can't do BTLE:

(Thanks to Peter "whitequark" Zotov for finding this !)

He uses a Nordic nRF24L01+ [1] which looks like a nice enough chip
at first sight, putting the "micro" back in "microelectronics" with
a 20-QFN 4x4 mm package, can do 1 Mbps, has GFSK, uses a frequency
deviation of +/- 160 kHz which is probably close enough, but ...

... limits packets to 32 bytes while BTLE needs 39 bytes. Darn.

There seems to be a whole class of similar chips: the Beken BK2423
[2] seems to be a direct clone and the RFM TRC104 [3] has somewhat
similar characteristics but also supports a "continuous mode" in
which the raw unmodulated bit stream is exchanged with the host.
That's a bit more low-level than what we're looking for but may be
a possibility. At USD 1.80 @ 1000 it's also a bit pricy for what it
does. The TRC104 was also mentioned as a possible choice on

(Credits: the people there, especially Mike Ryan and "mossmann",
have been extremely helpful.)

Last but not least, I found the Cypress CYRF8935 [4], another small
chip (24-QFN 4x4 mm) that may have suitable characteristcs,
including a decently sized FIFO. One interesting detail is that one
can connect an unbalanced antenna with only a very small matching
circuit, avoiding the bulky and costly balun.

It seems to be a brand-new chip, so there's a small risk of there
being issues with that. One known issue is that Digi-Key don't
sell it in small quantities (yet) - but others do.

Conclusion: RFM TRC104 and Cypress CYRF8935 go to the list of
potentials as well, pending further examination.

Next: chips in alphabetic order, beginning with A ...

- Werner


More information about the discussion mailing list