• State TI Thinks Resolved
  • Locked Locked
  • Replies 6 replies
  • Answers 3 answers
  • Subscribers 21 subscribers
  • Views 1397 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

LMX2572: Undesired Spurious Frequency

Richard Elmquist
Mastermind 45250 points
Part Number: LMX2572
Other Parts Discussed in Thread: LMX2594, LMX2582

My customer is seeing the following issue with the LMX2572:

It is a correct carrier frequency, but there are undesired spurious frequency (offset to 20KHz) in the pictures that i have sent. It has 100MHz oscillator with  LMX2572's PreR = 9, Multiplier = 7, N = 41.14311 in evaluation kit, so that the phase detect frequency is 77.7777777MHz, and integer boundary frequency is 11.13111MHz offset to carrier. I wonder that the frequency is 20KHz offset to carrier why it appeared, and how can i reduce them.



Can you explain why he is seeing this?

Thanks for your help with this!

Richard Elmquist

  • 0
    TI__Mastermind 27911 points
    Richard,
    Our PLLatinum Sim tool is a very useful software to understand the spur levels.

    In this case, it looks like you are using the programmable multiplier to try to avoid spurs.
    Note that 3200.02 MHz is not near a multiple of the phase detector frequency, so this is good for spurs.

    However 3200.02 is close to a multiple of the input frequency of 100 MHz. PLLatinum sim calls it the Fout%Fosc spur and Fvco%Fosc spur. Note taht 3200.02 % 100 = 20 kHz.

    The programmable input multiplier for this device is effective in reducing the spur at 20 kHz, but it does not completely eliminate it.

    Regards,
    Dean
  • 0 in reply to Dean Banerjee
    Prodigy 70 points
    the fout%fosc was only 30dbc to carrier in the picture that i sent, and there were more spurs which is multiple of 20kHz offset to carrier. in most of purpose, i do the best to avoid pll's integer boundary spurs, and it seems useless if i need a fine step frequency generator by a fixed oscillator.
    what's more, there is no fout%fosc spurs appeared when i make up my system by hittite(now is ADI) pll chip, eg, hmc833. so, is fout%fosc special within LMX chips? how do i do to reduce them?
  • 0 in reply to Dean Banerjee
    Prodigy 70 points
    the fout%fosc is only -30dBc to carrier, and there are more spurs which are multiple of 20KHz offset to carrier. in most of purpose, i do the best to reduce pll's integer boundary spurs, and it seems useless if i need a fine step frequency generator with a fixed oscillator in the system.
    what's more, there are no fout%fosc spurs appeared when i make up my system by using hittite(now is ADI) pll chip, eg, hmc833. so, Is fout%fosc spurs is special within LMX chips? how do i do to reduce them?
  • 0 in reply to Neal·J
    TI__Mastermind 27911 points
    Richard,

    I typically see the Fout%Fosc spurs and Fvco%Fosc spurs more in the -60 range, so it seems that -30 seems a little high and maybe something is off. The main thing is that you might find that the nature of the Fosc signal impacts the spur. For instance, you may find that a low amplitude, high slew rate signal like LVDS improves these spurs.

    If it's not an issue with the OSCin signal, also ensure that there is not a noise coupling path between the oscillator and the outputs. For instance, if you have an XO at 100 MHz directly connected to the same Vcc that is for the LMX2572 with no series resistance/inductance, this could be compling spur energy. Also, the LMX2572 output should have no pull-up and I wanted to confirm this because some of the other "pinout compatable" sister parts like the LMX2582 and LMX2594 require this pull-up. If you have this pull-up resistor on the outputs, it should be removed for the LMX2572.

    Whenever you integrate things together on the same die, you will see a crosstalk product spurs. The HMC833 has good spurs because it is a multi-chip package, but has many other drawbacks (current, size, cost ...).

    Regards,
    Dean
  • 0 in reply to Dean Banerjee
    Prodigy 70 points
    i had noticed that lmx2572 didn't need external pull-up components while lmx2594 needs them. I fellow the tips you noticed above, find what the spurs reduced much more than 30db when i configured the oscillator work as LVDS(AC couple) , but they became worse when oscillator work as LVPCEL(AC couple). so, are there some points should be cared when i design the reference oscillator circuits for lmx2572?
    truly that, I design my new circuits by using lmx2572 but not HMC833 for HMC833's larger size and higher power consumption. finally, the spurs specification is much more important than power consumption or package that it means to me.
  • 0 in reply to Neal·J
    TI__Guru* 98290 points
    Hi Richard,

    Did you mean your 100MHz reference clock is configurable between LVDS and LVPECL?
    I tend to agree that LVDS input may produce less fout%fosc spurs because the amplitude of fosc is smaller with LVDS.
    You may also try, configure LMX2572 input as single-ended (register R5[12]). Just like the default EVM configuration as shown in the EVM user's guide. My observation is, usually, single-ended input returns smaller spurs.