Using CC1101 with 8 byte sync word and Manchester Encoding

Note that CC1101 need 2-4byte of preamble to settle the AGC etc.  

Sverre said:

One thought is to set up the CC1101 in NRZ mode (not Manchester) with bit rate = Manchester encoded symbol rate . Each symbol in the transmitted Manchester encoded signal will then be treated as a bit. The Manchester encoded preamble (0xAF) will thus look like 01100110 01010101, corresponding to a 16 bit sync word of 0x6655

Enable sync word detection and set the sync word as 0x6655 (i.e. 16/16 sync word bits detected).

The nibbles in the FIFO should then be 0x5, 6, 9 or A. The data in the FIFO should be decoded as shown in the table below. Assuming 0-1 corresponds to a logic 1 and 1-0 corresponds to a logic 0 the decoding (in SW) from nibble to 2 bits will be as follows:

0x5 11
0x6 10
0x9 01
0xA 00

This is a clever solution Sverre. I am also in the same situation as with the first poster. 1 byte preamble and 1 byte sync word. What I did before is used synchronous mode. But sometimes I fail to catch the sync word, maybe because I am just having 1 byte preamble and it should be 2 or more as what TER has pointed out.

I tried this solution earlier, and just check the GDO0 (0x06 toggle when sync word is received) via SmartRF04EB.  Although I haven't checked the RX FIFO and I experience some false triggers sometimes (maybe because of the construction of my syncword?), but it is much better than missing some data.   I would also like to eliminate the clock and data toggling on the GDO pins due to synchronous mode as I am afraid that this may cause sensitivity degradation on my design.

Thanks

That looks promising, thanks for the pointers!

Hello,

I would like to confirm that Sverre's solution works perfectly. I did a test and from an old device that sends 1 byte preamble, and 1 byte sync word. Whenever I read the FIFO, it is in this format:

data[0]= PKTLEN - 1

data [1,,,n] = my data

then after that, I do the manchester decoding manually and I can see that the data are all valid.

Thanks Sverre. This is a brilliant solution. :)

Need to correct a mistake in previous post: CC1101 Manchester encoding: 0-1 corresponds to a logic 0 and 1-0 corresponds to a logic 1.  The decoding (in SW) from nibble to 2 bits will be as follows:

0x5 00
0x6 01
0x9 10
0xA 11

hi Sverre :

        we have a application is similar with your describtion, now have questions: 

        1、  in your  thought , the preamble  is  01100110, but the cc1101 datasheet show the preamble must be 101010……

         2、 why not use manchester encoded, if i want  to use manchester ,what should we config the regiters ?

user4269738: The original post had a protocol using only one byte sync. Using NRZ is a way to expand this to 16 bit.

If you could post a description of your protocol we will help you find a way to support this with CC1101.

It's generally better to start a new thread than post on a 2 year old thread.