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.

LMX2592EVM: phase noise for different range of frequency

Part Number: LMX2592EVM
Other Parts Discussed in Thread: LMX2592

Hi Team,

Could you please you help my customer to check what is the correct setting if they just wanted to look at this range of frequency.

I understand that the Loop filter on the Evaluation board is fixed. So with this current Loop Filter. how to optimise the phase noise to the Best.

 

My VCO output is

(a) 3357Mhz, 

(b)  3817Mhz,

(c) 4117Mhz,

(d) 4257Mhz

(e) 4507 Mhz,

Best Regards,

Ernest

  • Hi Ernest, 

    You're correct, the loop filter on board the EVM is fixed. 
    Please refer customer to PLLatniumSim to optimize loop filter. 

    Regards, 

    Vicente 

  • Hi Vincente,

    Thanks. 

    can you also help me check that if customer need to change the Loop Filter circuit. What is the best combination?

    Example they want to change the Crystal to 200Mhz (Max) and the Loop filter order( hardware components change) to let it be better phase noise.

    Because when they roughly try out the existing EVM, they can see improvement on the phase noise with their 11ax but It still has some Gap to achieve it.

     

    So total 2 items.

    (a) existing EVM board, how to achieve best noise configuration.

    (b) changing of EVM hardware to achieve better noise ratio.

    Best Regards,

    Ernest

  • Hi Ernest, 
    Please ask customer to use PLLatinumSim. 
    It really depends on their requirements. Do they have certain metrics they need to meet over a certain span? Do they care about in band noise, out band noise (PLL/VCO noise, etc? 
    I need more information about customer requirements before I can provide a recommmendation. 

    That's why we always push for customer to calculate optimal loop filter values using PLLatinumSim given we usually do not know know what they require. 

    Once customer has their optimal R/C values for their loop filter they change on EVM if they wish to verify.

    Regards, 

    Vicente

  • Hi Vicente,

    Thank you .

    It is quite straight forward, 1024QAM needs a s/n ratio for 35dB and above, customer requirement for the frequency they provide to you. Their Bandwidth is 20Mhz and 40Mhz.

    so as long as it can have a good margin for about 40dB S/N , I think it is good already. Their Power is about 24dBm.

    Is there any other information which you require?

  • Hi Ernest, 

    For a 20 MHz to 40 MHz span, customer is interested in only the noise floor of the device essentially. This is no longer loop filter dependent. 

    You can change the charge pump for example to change the LP BW & you will see the SNR not change at all. 
    SNR requirement is met nevertheless with LMX2592. 

    Regards, 

    Vicente 

  • Hi Vicente,

    Sorry, let me rephrase the question.

    Actually, I am looking into the S/N is for the EVM and not the Carrier-to-noise ratio which I miss out to you. 

    So can you help to check if customer is doing the design 1024QAM for the BW of 20Mhz and 40Mhz. What kind of phase noise should they achieve? example for 100Hz,1Khz, 10Khz, 1Mhz offset.

  • Hi Ernest, 
    If that's the case the phase noise performance they achieve at certain offsets will be dependent on their reference and their loop filter. If their reference clk is extremely clean, they will have 


    Please have customer use PLLatinumSim to estimate phase noise for their configuration. 

    For example, 
    If I simulate a 200 MHz input with a 3357 RFout utilizing a 3rd order loop filter & a 30 kHz LP BW & ~72 phase margin you can have the following phase noise at the offsets listed: 

    To ensure we are both on the same page, can you reiterate what customer is requesting? I understand they're asking about the phase noise of the device at certain offsets - but nevertheless I rather confirm to be certain. 

    Regards, 

    Vicente 

  • Hi Vicente, 

    (a) Can I check with you that, When I do simulation there is no shown any error, but I change the value on the EVM, it become lock to other frequency. 

    Is it due to Charge pump or Loop filter ?

    (b) For the actual software is only charge pump max is 4.843mA.  So from you simulation is max at 24.219mA. I had select the correct device LMX2592. So I am not sure what go wrong. 

    (c) Can you assist me on this. how to have a best phase noise to achieve a suitable use for 1024 QAM. the VCO output is from 3357 to 4527Mhz Range. Either 2 order or 3 Order Loop filter and what is the Charge Pump current.    

    Thank you very much

    Leroy

  • Hello Leroy, 
    A. What are the values you edit that result in this frequency shift? What frequency do you expect and what did you actually see? 
    B. There is a charge pump gain multiplier to the right, highest multiplier value available is 2.5 thus -> 4.843mA * 2.5 = 24.219mA
    C. Of course. In this case we can set Kpd to maximum value and set calculation to "optimize jitter" and have PLLatinumSim calculate optimal values, I have it set to 2nd order as well, you have an RMS jitter of of ~58fs for a integration BW from 12kHz to 20MHz. The values for C1/C2/R2 are shown below. 

    Does customer have any certain phase noise requirements at certain offsets? An RMS jitter requirement for a certain band of interest? 
    I am not familiar with QAM unfortunately. 

    Regards, 

    Vicente 

  • Hi Vicente, 

    I think my LMX have some issue (maybe damage), I will try to exchange another new part once i receive it and try it again. 

    Thank you for the support.