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: Modulator questions

Part Number: LMX2594

Hi,

This is a continuation from https://e2e.ti.com/support/clock-and-timing/f/48/t/907072

setting DEN = 10000 did improve the walking spurs but have degraded the spur performance as a result.

can you please assist with the following questions.

1. is the walking spur more susceptible if both LO's are using fractional values?

2. does the order of the modulator make a difference when using the DEN values?

3. should they use DEN = 10000 on all fractional values?

4. could using a larger DEN at rational numbers improve spur performance?

5. 1st and 3rd order modulators have given better spur performance while 2nd order has not been nice. is this typical?

Thanks,

Kevin

  • Hi Kevin,

    At 5750MHz output, with DEN = 10000, the spurs frequency should be 50MHz, I don't expect the spurs level is high at this frequency. What is the measured spurs level and frequency?

    I need more specific usage information before I can answer other questions. For example, what are the frequencies of interest, phase noise and spurs requirement, etc.

  • tune frequency = 5750MHz

    LO frequency = 14750MHz. but the goal is to use the LO in 1 MHz step sizes within a specific frequency range.

    Spur minimum spec = -60dB. does not seem like LO is capable of this spec. any idea on how to improve is welcome.

    Phase noise spec = -106dBc/Hz at 10kHz offset.

  • Hi Kevin,

    Above 11GHz, we are not able to hit the target phase noise.

    With 1MHz channel spacing, between 7000MHz and 15000MHz, there are 8000 channels, for sure fractional spurs and IBS are unavoidable.

    Here is my loop filter design, you can load it into PLL Sim.

    https://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/48/e2e.sim

    In practice, a super clean reference clock is required in order to hit the target phase noise at 10kHz offset. Both PLL and the reference clock will contribute noise at this offset frequency.

  • Hi Noel,

     

    customer mentions that the phase noise is fairly close to what they need. please provide an answer for each question below.

     

    1.       Do you recommend the denom value to be different for fractional verse integer configurations?

    2.       Does it make sense to run the fractional at 10,000 and the integer at the max value?

    3.       Would running the fractional at a value say of 100,000 since the number is rational help with spurs at all?

     

    Thanks,

    Kevin

  • Hi Kevin,

    In general, we don't need to change DEN from time to time. A rational DEN works for both integer and fractional channels. An irrational DEN may help reduce spurs in fractional channels but may hurt in integer channels. 

    If your application requires only a few frequency channels, maybe it makes sense to have slightly different configurations for each channel in order to optimize performance. Since in your case, you have 8000 channels, so it would be better to keep the same DEN for all channels. We should also use a rational number for DEN.