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.

CDCE937: Errors and conflicting information in datasheet of CDCE937 and family members: CDCE913, CDCE925, CDCE949

Part Number: CDCE937
Other Parts Discussed in Thread: CDCE949, CDCE925, CDCE913

I am using CDCE937 for a new design. I want to use this part to generate 6 frequencies total, 2 frequencies per PLL using the associated dividers.
My questions are as follows:

1. Here are 2 examples of 'pairs' of frequencies (mhz).  Based on my calculations, I should be able to generate each pair using a single PLL and 2 "PDIV" dividers.

desired_freq    Actual_freq    N_parameter    M_parameter    R_parameter    Q_parameter    P_parameter    PDIV_parameter

1.561120    1.561151892    3072        483        213        25        2        110
1.807680    1.807649559    3072        483        213        25        2        95

1.602240        1.602268657    1491        335        269        17        2        75    
1.848800        1.848771527    1491        335        269        17        2        65   

The parameters listed are based on calculations of "Actual_freq" and "Actual_freq" is within 160 hz of "desired_freq" (which is acceptable)

Can you confirm that each pair can be generated using a single PLL ?

2. I am not familiar with using PLLs, can you explain why R = 0 is preferred ??

3. Can you state the true and correct limits on the parameters "P" and "R" ??

4. Can you fix the datasheet(s) for these parts?

I provide more details below.


I have downloaded the part datasheet, here is the part number, web link, and revision information


Also, I reviewed replies in the forum related to this family of parts. One particular thread
has replies by "Julian Hagedorn"

Julian Hagedorn states:
    The PLL will be in fractional mode, if R is not equal to 0.
    The datasheet examples also suggest R = 0 is a good thing.

But the datasheet has inconsistencies on the limits on the parameters "P" and "R"
    Two pages within the datasheet give conflicting limits on these parameters.

1. I am not familiar with using PLLs, can you explain why R = 0 is preferred ??

2. Can you state the true and correct limits on the parameters "P" and "R" ??

Besides these inconsistencies there are typo errors in the equations on pages 25 & 26 of the datasheet.

Note: these types of errors are also present in the other "family member" datasheets
ex: CDCE913, CDCE925, CDCE949 and also the -Q1 parts


There is an inconsistency between the limits placed on parameters "P" and "R"

Please refer to Page 25 and 26, under " PLL Frequency Planning"

Please also refer to Page 22, the footnote (5) between Table 11 and Table 12.

Page 25 and 26, list the limits as

N (1 to 4095)
16 ≤ Q ≤ 63
0 ≤ P ≤ 4
0 ≤ R ≤ 51

Page 22 lists the limits as

(5) PLL settings limits: 16 ≤ q ≤ 63, 0 ≤ p ≤ 7, 0 ≤ r ≤ 511, 0 < N < 4096

Comparing the two pages, the limits for "Q" are consistent, and for "N" they are equivalent
However, the limits of "P" and "Q" are contradictory between the two pages.


There is a typo in one of the equations:

Page 25 and 26, under " PLL Frequency Planning" show the equations and two examples
for calculating the output frequencies.

There is an error in the equation for P     P = 4 – int(log2N/M; if P < 0 then P = 0

There is a missing ")" to match "int("

If I were editing the text, I would add an extra pair of "( )" around N/M for clarification.

Note: this type of error is in the other "family member" datasheets, except CDCE925, which has a different typo.


I could suggest incorporating some information written by "Julian Hagedorn" in this forum.

For example "In order to get the best performance N and M values should be maximized."

  • Hi Lee,

    The two pairs seem reasonable to me.

    Fractional PLL introduces more spurs than integer PLL (typically). Therefore, from noise perspective, it's a good thing to have R = 0.

    The absolute limit for P and R: 0<=P<=7, 0<=R<=511. However, there should never be a chance where P>5 because the minimum value of N/M = 0.5, where input frequency = 160MHz and VCO = 80MHz. Actually, it is rare that N/M < 1 and that's possibly where 0<=P<=4 is coming from. 0<=R<=51 should just be a typo.

    Thanks for all the feedback. We'll fix these in the next datasheet rev.