I am working with the AD6636 (DDC) and AD9251 (ADC) , both of which have a PN sequence generator, usually used for BIST purposes.

I am interested in vaildating my raw data streams from these devices when they are outputting the PN streams,

Can anyone tell me

1. the algorithm that generates these data... are they LFSR based?

2. Any post-PN generation data changes that might be occuring**

.. such that I can recreate the raw device output?

For AD9251 option, there are 2 possible generators: PN9 & PN23.

**For PN9, I notice ALL 14 bits seem to be changing, which means there is not a direct relationship between the PN generator state and the output.

Also, PN9 seqeunce period seems to be significantly less than the expected 511 samples.

Thanks for any replies.

David.

Dear David,

Thank you for your inquiry.

FOR THE AD9251:Yes the algorithm is LFSR based. There are no post generation data changes happening so you should be able to recreate this data. Please keep in mind the output format (e.g. Offset Binary, 2s Complement) will affect the actual 1s and 0s produced.

Regarding the sequence period, it will repeat every 2^9 – 1 bits, not samples. So you are correct in your observation that it repeats more often than every 511 samples. It should repeat every (511*2)/14 = 73 samples.

Both the PN9 and PN23 sequences and how to generate them are found in the ITU-T 0.150 (05/96) standard document. Below are some excerpts:

For the PN9 sequence, from Section 5.1 of the ITU document:

This sequence may be generated in a nine-stage shift register whose 5th and 9th stage output are added in a modulo-two addition stage, and the result is fed back to the input of the first stage. The sequence begins with the first ONE of 9 consecutive ONEs.

- Number of shift register stages 9

- Length of pseudo-random sequence 2^9 -1 = 511 bits

- Longest sequences of zeros 8 (non-inverted signal)

For the PN23 sequence, from Section 5.6 of the ITU document:

This sequence may be generated in a twenty-three-stage shift register whose 18th and 23rd stage outputs are added in a modulo-two addition stage, and the result is fed back to the input of the first stage.

- Number of shift register stages 23

- Length of pseudo-random sequence 2^23 -1 = 8388697 bits

- Longest sequences of zeros 23 (inverted signal)

Unfortunately I do not have the complete document to send you.

FOR THE AD6636:The PN generator in the AD6636 is a randomized PN sequence that was set up to generate broad band noise. It is not a standard PN9 or PN23 sequence. It's a 16 bit random noise sequence and has a repeat length of 2^28-1. This sequence is input ahead of a filter path and is typically used for viewing the filter shape (rather than for BIST). In this case it is not possible to reproduce the outputs. Even if you could somehow implement the algorithm, the filter would affect the values.

Please let me know if you have further questions and I’ll do my best to get the information. I hope your project goes well.

Sincerely,

Doug