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.

Design Using CDCE62005

Other Parts Discussed in Thread: CDCE62005, LMK00105

Hello Sir,

I am Vijetha H.N, Hardware Design Engineer at iWave Systems, Bangalore, India.

We are designing a PCB using CDCE62005 as the clock synthesizer. I have some queries related to the design using CDCE62005.

Design Decsription :

The design includes FPGA (Stratix V, ALTERA) and CDCE62005 is used to provide REFCLK (Thransceiver Referance clock) to FPGA. The SPI interface to Synthesizer is derived from CPLD (Max V, ALTERA).

The Clock Design Flow:

50 MHz Oscillator -> 1:5 LVCMOS Clock Repeater (LMK00105SQE/NOPB) ->Sec_Ref pin of Clock Synthesizer(CDCE62005) -> 1:4 LVDS Clock Repeater (CDCLVD1204RGTT) -> REFCLK pins of FPGA (Stratix V, ALTERA).

Input Spec for CDCE62005 = 50 MHz LVCMOS (to Sec_Ref).

CDCE62005 Output Requirement = 250 MHz LVDS on 5 outputs.

REFCLK Phase noise requirement for  FPGA = -120 dBc/Hz

REFCLK Phase Jitter requirement for FPGA = 1.2 ps.

I am new to the design of clock synthesizer, so I request you to guide me by answering the below queries.

1. What should be the Bandwidth of the loop filter for the above requirements? 

2. Please mention the required factors to be considered and standard procedure to be followed to select/calculate the Bandwidth of the PLL, so that 250 MHz clock is generated without noise/jitter.

Thank You,

Regards,

Vijetha H.N 

  • Since the input is an oscillator and has very good phase noise, it is better to set the PLL loop bandwidth very high (over 300kHz).

    The CDCE62005 software GUI has a loop filter tool built-in and can help you to select the component values to hit any loop filter bandwidth.

    Based on work we did in the past, please set your PLL loop filter internal and select C1 = 1.5pF, R2 = 24kohms, C2 = 473.5pF, R3 = 5kohms, C3 = 5.5pF, Icp = 1mA.

  • http://e2e.ti.com/cfs-file.ashx/__key/communityserver-discussions-components-files/48/0601.CDCE62005_5F00_2014_5F00_01_5F00_31_5F00_Vijetha.INI

    Hi Vijetha,

    Since the input clock to the CDCE62005 will be a clean source, I suggest using a wide loop bandwidth.

    The configuration used in Table 2 of the datasheet should be a good starting point for the settings used to generate 250 MHz output from 50 MHz input clock (PFD = 25 MHz = 50 MHz ref in / 2).  This configuration used a Loop BW of 400 kHz, and achieved 0.4 ps rms jitter (10kHz to 20MHz) with a clean reference input. 

    I attached the INI file for CDCE62005 software GUI which has the configuration described above which should be applicable in your application.

    Regards,
    Alan

  • Thank you Madhu and Alan for your suggestions.

    Here I am not cleared with the term "clean source". When the source is called a clean source? 

    The 50 MHz generated from the oscillator passes through a Clock repeater and then goes to the Synthesizer input. So here what are the parameters I need to consider to see whether the source is clean or not. If you mention the parameters to be considered, then I can get the respective values for the 2 ICs (Oscillator and Clock repeater).

    Regards,

    Vijetha 

  • Hi Vijetha,


    A clean source to the CDCE62005 is one with less than 2ps, rms jitter (12k-20M offset). Given that you have chosen an oscillator input, it should meet the required jitter numbers and should work fine with the loop filter suggested.

  • Hi Sir,

    I have looked into the JITTER information provided by the oscillator manufacturer and it is as shown below.

    They have mentioned multiple parameters related to jitter. 

    1. RMS Period = 0.186 ps

    2. Peak to peak period = 4.577 ps.

    3. Phase Jitter (12KHz - 20 MHz) = 0.083 ps.

    4. TJ RMS = 0.570 ps.

    Out of these which value should be considered to know what is the actual Jitter (RMS) at the synthesizer ref. i/p.

    (There is 1 more reason for asking this question; I want to know what is the total jitter observed at the REFCLK input of FPGA in the design. To know this I need to add jitters contributed by individual parts.)

    Regards,

    Vijetha

  • Hi Vijetha,

    You can use the oscillator RMS phase jitter spec (83 fs rms) and LMK00105 additive phase jitter spec (100 fs rms) to compute the total phase jitter at the Synth input = sqrt (83^2 + 100^2) = 130 fs rms.  This is a very low jitter input to the CDCE62005 synthesizer.

    Regards,
    Alan

     

  • Thank You Alan.

    Here is another couple of queries:

    1. We need to set the type of charge pump (Narrow band/wide band). Which 1 is preffered in this application?

    2. There are 2 registers SEL_DEL1 and SEL_DEL2 which should be 0 to set short delay for fast operation and 1 for long delay (recommended for Ref i/p< 150 MHz). Here what is the meaning of fast operation and where this delay is effective?

    3. There are 2 VCOs in the PLL. Which VCO I can use for my design?

    4. In the configuration document you provided (.INI) what is the meaning of PORTS and EXTERNAL COMPONENTS values mentioned? We only understood the register values and the software team is checking the values against the requirement. We also understood that the values mentioned for INPUTS are the frequency values (in MHz). Can you help me in understanding the meaning of the values given for other 2 parametsers (PORTS and EXTERNAL COMPONENTS).

    There is also a set of values mentioned by Mr. Madhu to set the loop bandwidth. Whether the register settings and the discrete values by Madhu are same? To confirm this can you tell me how to decode these register values (with 1 example) ?

    5. It is mentioned that the table 2 can be a starting point for our descign using 50MHz ref input. But it is given for 25 deg C. My application will run up to +60 deg C. Is there any deviation in these values at this temperature?

    Regards,

    Vijetha

  • hi Vijetha,

    (1) Wide band is better.

    (2) Please set "1" for SEL_DEL1 and SEL_DEL2 since your ref freq is < 150MHz. This is for the smart MUX operation and this is especially useful when trying to do the automatic switching between primary and secondary. The delay within the MUX is minimized for faster clocks to enable reliable auto switch.

    (3) Please use VCO1 since that matches the frequency plan you want.

  • (4) Ports indicate output port in Alan's file and External components indicate external R and C values for external loop filter. But since you have clean source, please use internal loop filter with the values I suggested.

    (5) We should be within 1ps across entire temp range of -40C to 85C.

  • Dear Madhu,

    Thanks lot for the guidance.

    To verify the R/C values of Loop filter and to calculate the Bandwidth of PLL, I have used the CDCE62005 EVM Software 1.4.8 in which I entered the suggested values. 

    When I entered the values, I found following observation:

    1. When the Charge Pump current value was 1mA the BW calculated from the tool was around 1.7 MHz, with a warning on 3rd order pole BW.

    2. When the charge current reduced to 300 uA, the warning disappeared with BW=473 KHz.

    I would like to know how you've derived capacitor and resistor values w.r.t bandwidth value (is it based on formulae substitution or trial and error method) ? If it is based on formulae substitution (R,C values vs BW value) then I kindly request you to inform the formulae you've used.

    Regards,

    Vijetha

  • Hi Vijetha,

    I made a typing error. I did go through the tool but I used 200uA charge pump current. Another option is 300uA with C3 = 5.5pF for a higher 3rd order pole.

  • Dear Madhu,

    Yes you are correct, with 200 uA of charge pump current value I am  able to achieve Loop bandwidth around 300 KHz - 400 KHz. I just wanted  to know (as I asked in previous post) how these values (R,C, charge pump current value, Bandwidth) are achieved? Is there any procedure/formulae to design these values?

    It would be greatly helpful, If you share the method/formulae you've followed to design these values of PLL filter.

    Vijetha

  • Hi Vijetha,

    The procedure is based on closed loop analysis including frequency response of a PLL. You can calculate the -3dB point and the phase margin with these formulas. The detailed theory behind this is given in the excellent book by my colleague, Dean Banerjee. It is here http://www.ti.com/tool/pll_book

    The theory is implemented in the loop filter design tool in the CDCE62005 EVM GUI. You can use that software to design your loop filter.

  • Thank You Madhu and Alan for the support. And the book is excellent, thanks for that too. 

    Regards,

    Vijetha