• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 1 answer
  • Subscribers 21 subscribers
  • Views 986 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.

LMX2531: Reference Oscillator Input level and required signal type

Eric Hooker
Expert 1445 points
Part Number: LMX2531

Hello,

We are using the LMX2531LQ257. We have encountered an issue where some synthesizers don’t lock on our PCB. We found the lock issue is due to the OSCin level. Our oscillator frequency is around 20 MHz.

Page 6 of the data sheet specifies “Oscillator Sensitivity” as 0.5 Vpp min and 2.0 Vpp max.

  1. Does this specification assume the oscillator signal is a clipped sine wave or a square wave?
  2. If the answer to question 1 is no, is it okay if the oscillator signal is a sine wave?
  3. Does this specification vary with oscillator frequency? In other words, is 0.5 Vpp sufficient at 20 MHz? We have found that 0.5 Vpp is too low at 20 MHz.
  4. If the specification does vary with frequency, what is the minimum voltage level required at 20 MHz?

Thanks,

Eric

  • 0
    TI__Expert 7855 points
    Hi Eric,

    1. These are specifications for peak to peak voltage values, either sine wave or square wave is included
    2. yes, in fact sine wave is what we typically use
    3. there will be some dependency across input frequency, but the lower and upper bound specifications are chosen so that it works over PVT for the worse case frequency.
    4. Are you using the EVM? Are there components in the input path? What voltage level were you able to go down to before not locking (for ex. 0.7V, 0.8V)?

    Regards,

    Brian Wang
  • 0 in reply to Brian Wang
    TI__Expert 5370 points
    Dear Eric,

    Sorry to hear you have issue with lock.

    You should not have issue at 0.5 Vpp single ended driving OSCIN (cap to ground on OSCIN*) with 20 MHz on your board.

    I am wondering if it is possible that your calibration setting are off to operate with 20 MHz. check register 7 and make sure you have XTLDIV=2 or more. If you are on the edge and the calibration engine fail to operate at the speed you will fail to lock.

    Regards, Simon.
  • 0 in reply to Brian Wang
    Expert 1445 points
    Hi Brian,

    We are not using the EVM. We are using our own PCB design. The input path is like the eval board. We can't go below about 0.9 V.

    Thanks,
    Eric
  • 0 in reply to Simon Damphousse
    Expert 1445 points
    Simon,

    We will check register 7.

    Thanks,
    Eric
  • 0 in reply to Eric Hooker
    Expert 1445 points
    Simon,

    Register 7 is okay. I have another question. Is it possible that harmonics of the 20 MHz clock could be an issue? I have seen harmonics of VCO's cause problems for synthesizers. I usually follow a VCO with a LPF to attenuate harmonics before feeding the signal back into the synthesizer. (I am talking about other synthesizers without an internal VCO.) I looked at the 20 MHz clock signal and the harmonics were only about 10 dB down. Also, is the OSCin pin more sensitive at frequencies > 20 MHz? If so, perhaps the reference divider is "confused" by one of the 20 MHz harmonics?

    Thanks,
    Eric
  • 0 in reply to Eric Hooker
    TI__Genius 9920 points
    Hi Eric,
    For a square wave, it is normal for only 10 dB down on the 3rd harmonic. It would be helpful to debug with a scope to capture a waveform for this 20 MHz on OSCin input. I'm thinking the input schematic impedance match or reflection.

    Regards,
    Shawn
  • 0 in reply to Shawn Han
    Expert 1445 points

    Hi Shawn,

    I did look at it with a scope but I don't have the scope clock waveforms handy.  I did simulate the clock waveforms with Keysight ADS and they matched scope measurements very well.

    Below is the waveform when the OSCin was terminated with a shunt 49.9 ohms in series with 0.01 uF.  The peak to peak level (scope and ADS sim) was 0.96 Vpp.  A LMX2531 did not lock with this OSCin waveform.  Notice the waveform is also distorted somewhat.  The clock freq is actually 19.2 MHz. 

    Below is the waveform when the OSCin was terminated with a shunt 200 ohms in series with 0.01 uF.  The peak to peak level (scope and ADS sim) was 1.57 Vpp.  The same LMX2531 did lock with this OSCin waveform.  Notice the waveform is also more linear.  So, this is the solution we are using now -- changed the OSCin shunt resistor to 200 ohms.  By the way, we have seen this issue on only two LMX2531 parts.

    Thanks,

    Eric