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.

  • TI Thinks Resolved

LMX2571: LMX2571 & spur (subharmonic) 1/2*Fout, -52dBc

Intellectual 290 points

Replies: 11

Views: 847

Part Number: LMX2571

Dear colleagues

On Evaluation Module "LMX2571EVM" (www.ti.com/.../lmx2571evm) I see spur with frequency of Fout/2 (subharmonic) and level - 52dBc.
Also I see harmonics of frequency of operation of the phase detector with level - 75dBc.
From where do they undertake?

The output spectrum at generation frequency 1008 MHz (OSCin=20MHz, Fpd=120MHz, N=21.0, CDIV=5):

TICSPRO-SW settings:

TICS_Rg 2018-11-07 1008 120pd 4sd 2del 325uA 800 - F12.rar

Yours faithfully,
Alexander

  • Hi Alex,

    Please try:
    - Disable Dithering
    - Set MASH order to Integer
    - use smaller charge pump current
  • In reply to Noel Fung:

    Hi Noel

    It doesn't help:
    - Disable Dithering
    - Set MASH order to Integer
    - use smaller charge pump current
    - use bigger charge pump current
    - MULT_F1=3 (was 6) and N=42 (was 21)

    I dissabled OSC (I connected Y1:1 to GND) - remained only the output frequency, which became a little less, and her subharmonica 1/2:

    I see a subharmonica 1/2 as well on my PCB

    Yours faithfully,
    Alexander

  • In reply to Tolchinsky Alexander:

    Alexander,

    If changing the charge pump current or MASH order has no impact, then this spur either has nothing to do with the LMX2571 or is coming direct from the output. If it's from the output or some rogue divide by 2, I don't know what can be done.

    But try powering down the output and the spur shoudl go away. Also, you might try the PLL PRE_N divider and changing this.

    Regards,
    Dean
  • In reply to Dean Banerjee:

    Dear colleagues

    I connected the R&S SMB100 generator to input of spectrum analyzer: the subharmonica 1/2*Fout – is absent.
    I set OUTBUF_RX_EN_F1 = to dissable - out Rx is clear.


    With my settings I see a subharmonica 1/2*Fout only for CDIV1*CHDIV2 = 5*1 (-52dBc) and 7*1 (-58dBc). Change of MULT_F1=3 (Fpd=60VHz), N=21*2=42=0x2A doesn't change a situation.


    I set PLL_N_PRE_F1=4 and N=21=0x15 – the situation has changed. CDIV1*CHDIV2:
    5*1 -> Fout*3/4@-60dBc
    6*1 -> Fout/2@-66.5dBc
    7*1 -> Fout/4@-67dBc

    Subharmonicas go from LMX2571. What to do?

    Regards,
    Alexander

  • In reply to Tolchinsky Alexander:

    Hi Alex,

    I saw the same sub-harmonic issue in my EVM.
    I checked the history, this issue comes from the odd divider.
    I can reduce the sub-harmonic level by 10dB with some hidden registers.
    Except for 1008MHz, do you need other output frequencies?
  • In reply to Noel Fung:

    Hi, Noel

    Thank You for attention.

    1008MHz is only for an example: it was necessary for me at some unmodulated frequency from range 800-1000MHz to get the cleanest spectrum in a "near" band +-500kHz (to approach generator R&S SMB100A) - I tried to make it and noticed this subharmonic 1/2...

    Regards,
    Alexander
  • In reply to Tolchinsky Alexander:

    Hi Alex,

    You need 800-1000MHz, then only part of the hidden registers work, you cannot get 10dB improvement in the entire range.
    Then I think there are two options:
    (1) stay with LMX2571 but you need to add BPF to remove the sub-harmonics
    (2) consider to use LMX2572. This is also a low power synthesizer with better performance than LMX2571
  • In reply to Noel Fung:

    Hi, Noel

    For me is more suitable LMX2572LP (e2e.ti.com/.../713608 and I will try this chip before the following redesign of my project. At this chip CHDIV changed - now it is only 2^n. It is possible to hope that LMX2572LP does not have this problem of subharmonics?

    Can you give settings for LMX2571 for the frequency of 1008 MHz with the reduced level of subharmonic 1/2?

    Regards,
    Alexander

  • In reply to Tolchinsky Alexander:

    Alexander,
    I don't know the magic settings, butboth Noel and I will be out Monday next week, but back on tuesday.
    The LMX2572LP eliminates a mux to save current, but I am skeptical that it will eliminate this spur.

    REgards,
    Dean
  • In reply to Dean Banerjee:

    Hi Alex,

    Set R58[13:12] from 00 to 11.
    Set R58[15:14] from 10 to 11.
    These two registers are for the channel divider.

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.