• State
  • Locked Locked
  • Replies 8 replies
  • Subscribers 21 subscribers
  • Views 534 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

LMX2572: Phase shift

Ram Al Sabti
Prodigy 50 points
Part Number: LMX2572
Other Parts Discussed in Thread: LMX2594

Dear,

I would need some further help regarding the phase shift option on the LMX2572, since I am not sure if this behaviour is normal.

I have two LMX2572 boards which are sweeping between 500 MHz and 2490 MHz, with a step of 10 MHz (200 number of steps). At each frequency the boards are phase synchronised and then a phase shift for 400 steps is performed (between 0.72-0.99 degree/step, depending on the frequency, since the resolution is changed with CHA_DIV). To validate the performance both signals are summed with a RF combiner and measured is the average amplitude. The signal output is shown in the figure below:

Since the cable length of the two outputs is different, the magnitude minimum should slowly shift up with the frequency. This almost matches the expectations, but there are also two unexpected behaviours:
1. At frequency f17 and f18 the minimum doesn't follow the order which means that the phase sync was unsuccessful.
2. At some frequencies (specially between f111- f129) it looks like the frequency shift is not linear since the shape is not as expected (eg. f1)

I have repeated the measurements and the results are the same, so could this be a configuration problem or its a hardware limitation? Later I would like to set a specific phase shift for each frequency, where I then also expect that it will match the setup.

below is my a screnschot of my basic setup, from where I then just recalculate the needed values (N, CH_DIV, PFD_DEL, ...)


thank you and best regards

  • 0
    TI__Guru* 93670 points

    Hi Ram,

    We will look into this and get back later, stay tuned.

  • 0 in reply to Noel Fung
    Prodigy 50 points

    Dear Noel,

    I have completely solved the 2. problem. The RC values of the loop filter have been not correct since I have used the same values as the one on one of my LMX2594 boards. It looks that this has then under some conditions caused a strong jitter.

    however, I could not solve the sync problem. Attached is the latest measurement with a frequency sweep, which was repeated for three times. Its visible that in some cases the phase shifts for 180 degree (f82, f83), and in some scenarios it has just a phase difference of few degree (eg. f64,f183) after repeating the measurement. But also after performing multiple measurements the frequency at which this happens changes. I have also already tried to increase the MASH_RST_COUNT, but without success.

    best regards.

  • 0 in reply to Ram Al Sabti
    TI__Guru* 93670 points

    Hi Ram,

    For those failure channels, before they are phase shifted, are they phase aligned? I assume you use MASH_SEED to shift the phase, right?

  • 0 in reply to Noel Fung
    Prodigy 50 points

    Dear Noel,

    yes, first they are phase aligned for every frequency step. And yes, I use the MASH_SEED to shift them.

    If maybe the figure is unclear. The minimum at each frequency shows when the two signals have a phase difference of 180 degree (basically they substract each other) and the maximum when they are phase aligned. So if we track the maximum/minimum position, this will shift with frequency because the cables have different lengths.

    So for me its important that the signals have always the same phase difference if I continuously switch through frequencies. But as its visible from the figure, after multiple loops at some frequencies the phase doesn't remain the same. Because fpd=10MHz and i increase the frequency always in 10MHz steps I am always in sync category 1 or 2, which shouldn't be critical. Or its possible that I would have to increase some more values (like I have already the MASH_RST_COUNT) or add delays?

    Thank you

  • 0 in reply to Ram Al Sabti
    TI__Guru* 93670 points

    Hi Ram,

    In Cat.3 sync, MASH_RST_COUNT needs to be increased when OSCin is just 10MHz. In Cat.2 or Cat.1 sync, I think this is not necessary. 

    Since you said their phase are aligned before you are trying to shift their phase, so the problem is on the MASH_SEED but not the sync.

    Can you toggle the MASH_RST_N bit (make it LOW and then HIGH) before writing to MASH_SEED?

  • 0 in reply to Noel Fung
    Prodigy 50 points

    The problem is definitely in the sync. Now I had the opportunity to connect the two outputs directly to an oscilloscope and the phase shift works as it should, only the phase sync doesn't.

    At f=670MHz I can never get a phase sync, but the phase between the signals has always a constant phase.
    at f=1320MHz the phase is switching between phase sync and 180 degree, on every time I send a phase sync pulse. The same thing is then also happening at some other higher frequencies.


    My procedure to change the frequencies is as following:
    at default both board have MASH_SEED = 0 and MASH_RST_N = 1, VCO_PHASE_SYNC_EN = 1
    loop:
    1. board1 MASH_SEED = 0, toggle MASH_RST_N
    2. set frequency on board1 and board2
    3. toggle RESET (This step probably wouldn't be even needed)
    4. send all registers
    5. toggle FCAL_EN
    6. Send a sync pulse
    7. phase shift board1

  • 0 in reply to Ram Al Sabti
    TI__Guru* 93670 points

    Hi Ram,

    Please try below configuration.

    No QUICK_RECAL_EN and MASH_SEED_EN. Use default VCO calibration parameters.

    It should not be matter, but worth try set LDO_DLY = 1. I have double the MASH_RST_COUNT value, you may need to change this value according to your lock time.

    Test procedure:

    1. Vcc power up the devices

    2. Program all the registers so that they all lock to 670MHz

    3. Program PLL_N = 268 (decimal) and then VCO_PHASE_SYNC_EN = 1

    4. Program PLL_N = 536 (decimal) and then VCO_PHASE_SYNC_EN = 0

    Do they sync?

  • 0 in reply to Noel Fung
    Prodigy 50 points

    Dear Noel,

    i have followed your suggestion, but have even increased MASH_RST_COUNT=300000. Now when I use the software sync with your suggestion it syncs, but not always. After multiple retries in many cases it doesn't. Bellow is a video
    successful sync

    unsuccessful sync

    Its also interesting that in some moment the phase is already in sync after lock, but then after applying the sync procedure not anymore.
    And from your procedure first, step 1-2. have been applied to both boards and then after lock 3-4. for both boards