• State TI Thinks Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 2 answers
  • Subscribers 22 subscribers
  • Views 384 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: LMX2594 Ignores forced VCO_CAPCTRL and show differen phase relationship between programming cycles.

Manuel Milla
Prodigy 45 points

Part Number: LMX2594

Hi, 

I am trying to synchronize the output of two LMX2594 devices on the same board. In order to have the same phase relationship after each powercycle, I am forcing the VCO_SEL, VCO_CAPCTRL and VCO_DACISET values. 

This is the PLL Configuartion for the both LMX2594 Devices. 

image.png

I set different VCO_CAPCTRL and VCO_DACISET while keeping VCO_SEL for each device. Each LMX is programmed reading the register values from a TICS generated file. As you can see in my log, only register 19 (VCO_CAPC) and register 16 (VCO_DACISET) are different. 

Device LMX2594_1 is programmed with VCO_CAPCTRL=127 (0x7F) and VCO_DACISET=357 (0x165)

Device LMX2594_2 is programed with VCO_CAPCTRL=130 (0x82) and VCO_DACOSET=399 (0x18F)

Capture.PNG

As you can see in the log, I launch the programming script having two different results:

In one of them, the LMX devices seems to be well aligned.

A capture )

lmx2594_sync.png

In the second one, they show a phase shift

 B Capture)

lmx2594_no_sync.png

My questions are:

Why the VCO_CAPCTRL are not the same as I forced ? 

Also, even if they are different, the phase shift between the LMX is "constant" but different each programming cycle...this is, at each programming cycle, they show eiter A behavior or B behavior. Shouldn't they always show the same phase shift ?

Thank you very much for taking the time in reading my question.

 

 

  • 0
    TI__Guru** 122220 points

    Hi There,

    The phase between the devices may not necessary be identical as the propagation delay is not the same, but the phase difference should be deterministic. 

    For example, I have two boards, after VCO calibration, their VCO data are as follows:

    device 1: capcode = 121; Daciset = 281

    device 2: capcode = 132; Daciset = 531

    If I force them to use these data (or still in auto-calibration mode), their phase difference is always constant. I used the same TICS Pro configuration as yours.

  • 0 in reply to Noel Fung
    Prodigy 45 points

    Hi, 

    Thank you for your reply. I agree with you that's what I should be observing as well...For a bit more of context, this is how my user controls pannels look:

    As you can see QUICK_RECAL_EN = 0
    VCO_CAPCTRL_FORCE=1

    VCO_SEL_FORCE=1

    VCO_DACISET_FORCE=1

    However for some reason it looks like the VCO_CAPCTRL is being ignored and I readback a different value than I programmed.





  • 0 in reply to Manuel Milla
    TI__Guru** 122220 points

    Hi There,

    This is a silly bug in the chip, in full assist mode, the read back CAPCTRL does not equal to the forced value. 

    You can verify this by measuring the Vtune voltage. Write VCO_CAPCTRL to 135, the Vtune will change. That means the device is actually using 130, not the read back value of 135.

  • 0 in reply to Noel Fung
    Prodigy 45 points

    Hi Noel, 

    Thank you! Is there a way or effectively reading back what VCO_CAPCTRL is being used or do I just have to assume that what I write in the register is the "actual" value ?


  • 0 in reply to Manuel Milla
    TI__Guru** 122220 points

    Hi Manuel,

    If you do a no-assist calibration and read back, the returned CAPCTRL value is correct.

    The bug we have is, when you do a calibration in full-assist mode, the read back CAPCTRL value is not correct.

  • 0 in reply to Noel Fung
    Prodigy 45 points

    Thank you!

    I will give it a try...maybe there's also something in the way I program the devices. In my current configuration, each LMX is getting the reference oscillator from a LMK (different LMK output to LMX ref input). I program first the LMK and then I program the LMXs, even though I am in full assist the phase relationship is different between programming cycles ... these phase diferences are "discrete" as I show in my oscilloscope captures: they differ but between two or three posible cases. Do you see any big flaw in my concept ?  



  • 0 in reply to Manuel Milla
    TI__Guru** 122220 points

    Hi Manuel,

    Did you perform a sync to the LMK device so that all output clocks are in phase?

  • 0 in reply to Noel Fung
    Prodigy 45 points

    Hi Noel,

    No, I was giving it for granted the outputs of the LMK were in phase I will certainly look into this,

    Thank you a lot !!