This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

lmx2594: Optimising fractional spurs when adjusting output phase

Part Number: LMX2594
Other Parts Discussed in Thread: TIDA-01346

Hello. 

I am trying to reduce the fractional spurs that I see when adjusting the output phase of the LMX2594 using the MASH_SEED.  I was hoping that Ti might have some tips. 

I have been reading TIDA-01346 which states:

"Programming non-zero seed values can actually produce fractional spurs with a numerator of zero. Therefore, optimal spurs are obtained when MASH_SEED is a multiple of IncludedChannelDivide and the fractional denominator is set to k × IncludedChannelDivide, where k is a positive integer that has no factors of 2 or 3. Sub-fractional spurs can be avoided using this strategy"

To this end I have been using a fractional denominator of 127, such that k=127 which contains no factors of 2 or 3.  I'm incrementing MASH_SEED by 1 which is a multiple of IncludedChannelDivide.  Please see the setup below:

I'm getting quite bad fractional spurs when I increment the MASH_SEED in steps of 1.  For instance, with a MASH_SEED of 1 I get:

I believe these are fractional spurs at multiples of 200MHz / 127 = 1.5748MHz.

Would Ti have any tips on how I can adjust the output phase of the LMX2594 whilst minimising SPURS?   I have found that some MASH_SEED values are better than others, but I've not found a way to predict the behavour.  I think PLLatinum Sim does not work with a numerator of zero. 

Thnks for the help 

  • Variac,

    PLLatinum Sim does model the impact of MASH_SEED on spurs, but not for a numerator of 0 as you say. The impact of MASH_SEED on spurs does not have a simple closed formula, but rather involves finding the sequence and calculating the Fourier Series as described in my PLL book at (www.ti.com/.../pll_book).

    If the numerator is zero and the seed is nonzero, you do get a sequence, so therefore the fourier series and spurs can be calcualted. I tried this on PLLatinum Sim and it's not working. So in summary, when we release the next PLLatinum Sim, it will handle this feature, but this will probably be in a few months.

    Regards,
    Dean
  • Hi Dean. 

    Thanks for the help and the link to the book.  I'm not very familiar with the LMX2594 sigma-delta architecture so I'll have a study of your book and hopefully be able to figure out how I can optimise these spurs. 

    For now I think a brute for search might be the way to go to figure out what denominators and MASH_SEEDs give the best performance in my application.