• State Resolved
  • Locked Locked
  • Replies 10 replies
  • Answers 2 answers
  • Subscribers 21 subscribers
  • Views 1121 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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: 2 LMX2594 frequency locked output

Abhijit Khadilkar
Prodigy 250 points
Part Number: LMX2594
Other Parts Discussed in Thread: LMK04828, LMX2820

Hi, I am using 2 LMX sourced from same signal of 50MHz. One LMX2594 output sources the sampling clock of ADC of 2.5GHz. Another LMX2594 sources the LMK04828 clock of 1.25GHz. LMK04828 generates the

  1. SYSREF for FPGA and ADC.
  2. refclk for various JESD devices like FPGA and FPGA serdes refclk.

How to make sure that the outputs of two different LMX2594 are frequency locked? Please help. 

I use the TICS Pro tool for generating the hex file for LMX2594.

Attached is the schematic of LMX2594 board.

thanks, AbhijitHSC1389_REV1P0-SCH.pdf

  • 0
    TI__Genius 11900 points

    Hello Abhijit,

    I am a bit confused here. Based on your description one LMX2594 outputs 2.5 GHz while the other outputs 1.25 GHz. You are asking me how to frequency lock two different frequencies?

    If both the LMX2594 devices were outputting the same frequency you can have deterministic delay between them. Please refer to section "7.3.10 Phase Synchronization" of the datasheet.

    Thanks,

    Vibhu

  • 0 in reply to Vibhu Vanjari
    TI__Prodigy 250 points

    Hi, 

    The issue is: When i use the outputs from from LMX 1 for sysref, FPGA clocks and LMX2 for ADC reference clock generation, the JESD between ADC and FPGA does not achieve sync. 

    When i give the clocks from external signal source like HSX, the JESD achieve sync.

    I am suspecting that the outputs of two LMX are not tracking each other's frequency changes.

    I want both the LMX2594 outputs to track the output frequency changes if any.

    thanks, Abhijit

  • 0 in reply to Abhijit Khadilkar
    TI__Genius 11900 points

    Hello Abhijit,

    I have your schematic but maybe a block diagram with frequencies will help. Please share this.

    In your initial post you say that the SYSREF comes from the LMK device, is that not the case?

    The outputs of the two LMX devices will not track each other's frequency changes. I am not sure why you expect that. If both the LMX devices outputted the same frequency there would be a way to sync them to have deterministic phase.

    "I want both the LMX2594 outputs to track the output frequency changes if any."

    I do not understand this statement, can you please elaborate?

    Thanks,

    Vibhu

  • 0 in reply to Vibhu Vanjari
    TI__Prodigy 250 points

    Hi,

    "I want both the LMX2594 outputs to track the output frequency changes if any."

    I mean if one LMX2594 output frequency increases the other should increase in same ratio and vice versa.

    Please refer my email for block diagram.

    thanks, Abhijit

  • 0 in reply to Abhijit Khadilkar
    TI__Genius 11900 points

    Hello Abhijit,

    These are two individual LMX2594 devices that need to be programmed individually. For example let's say that both your LMX2594s are programmed to a 10 GHz VCO. On one LMX2594 you use channel divider of 4 to get 2.5 GHz and on the other you use channel divider 8 to get 1.25 GHz.

    Now let's say the output frequency oft he first LMX2594 increases from 2.5 GHz to 2.6 GHz, this means you need re-write the PLL registers to get a 10.4 GHz VCO frequency. The second LMX2594 will still have a 10 GHz VCO and 1.25 GHz output until this second device is also reconfigured for a 10.4 GHz VCO.

    Again one option here is switching to the LMX2820. The device has two outputs that come from the same VCO, however unlike the LMX2594 this device has individual channel dividers. So you can use a divide by 4 on one output and a divide by 8 on another output. To change the frequencies and have the 2.5 GHz and 1.25 GHz frequency change at the same scale now all you need to do is program one device to a different VCO frequency.

    I do not understand why you think the outputs on the two LMX2594s would track without any external intervention / re-configuring of registers. Please give me your thoughts on this. Also thanks for sending over the block diagram!

    Thanks,

    Vibhu

  • 0 in reply to Vibhu Vanjari
    TI__Prodigy 250 points

    Hi Vibhu, Can you please send me the hex register file for the two LMX components, suitable for frequencies.in the block diagram.  

    I am generating the files using TICS Pro but i want a golden file for reference. I don't have an EVM in the lab.

    Thanks, Abhijit

  • 0 in reply to Abhijit Khadilkar
    TI__Genius 11900 points

    Hello Abhijit,

    You can select the "Default configuration" in the tool bar to start at a known good configuration.

    I am not sure what you decided whether you want to use two different output frequencies from the devices or you'd like to SYNC them with the same output frequency.

    If you do not plan to use SYNC here is what you need to do:

    1. Start from default configuration
    2. Enter your input frequency 50 MHz in Fosc
    3. Enter your output frequency in RFoutA
    4. Set PFD_DLY_SEL = 3

    Tip: When you mouse over yellow/red highlighted boxes, the tool will tell you what is wrong or how to correct it.

    Thanks,

    Vibhu

  • 0 in reply to Vibhu Vanjari
    TI__Prodigy 250 points

    Hi Vibhu,

    When i configure the output of both LMX at 1.25GHz it has a duty cycle of 65-70%. For the same set of outputs configured to 2.5GHz the duty cycle seems to be 42%.

    Also the 1.25GHz waveform has does not look clean as 2.5GHz waveform. How can i improve it?

    Please refer the attached files. thanks, Abhijit

  • 0 in reply to Abhijit Khadilkar
    TI__Genius 11900 points

    Hello Abhijit,

    Can you confirm that your device is locked? The easiest way to do this is to close MUXout_SW (refer to EVM schematic) pins 2 and 3 so the LED is connected.

    The LED will light up when the device is locked.

    Always toggle FCAL_EN to lock VCO after changing frequency. This will likely not be needed because you will be using the same VCO frequency (10 GHz) with different channel dividers.

    With higher channel divider values the waveform is expected to look more and more like a clipped sine wave. So this behavior is somewhat expected however your waveforms don't look the cleanest.

    If you are measuring these outputs single-ended, then make sure you terminate unused output pins with a 50-ohm termination.

    Please let me know if any of these suggestions clean up the waveforms.

    Thanks,

    Vibhu

  • 0 in reply to Vibhu Vanjari
    TI__Prodigy 250 points

    Hi Vibhu,

    The easiest way to do this is to close MUXout_SW (refer to EVM schematic) pins 2 and 3 so the LED is connected.--- I checked rb_LD_VTUNE in R110[10:9]. I am seeing 10 in these two bits after the LMX is configured.

    The LED will light up when the device is locked.---I am using a board which does not have LED.

    Always toggle FCAL_EN to lock VCO after changing frequency. This will likely not be needed because you will be using the same VCO frequency (10 GHz) with different channel dividers.---Making this bit 0 and then 1 after every configuration write of LMX SPI registers.

    With higher channel divider values the waveform is expected to look more and more like a clipped sine wave. So this behavior is somewhat expected however your waveforms don't look the cleanest.---After changing the PLL divider, i am able to get 50% duty cycle. Now i am using 625MHx instead of 1.25GHz. See attached waveform.

    If you are measuring these outputs single-ended, then make sure you terminate unused output pins with a 50-ohm termination.---Yes.

    Please let me know if any of these suggestions clean up the waveforms.---Now the waveform looks clean. Thanks for your support.

    Thanks, Abhijit