LMX2820EVM: Phase Noise

Part Number: LMX2820EVM
Other Parts Discussed in Thread: LMX2820, USB2ANY, LMX2592EVM

A phase noise of LMX2820EVM is bad.

Result of a measurement is far apart from simulation.

Please give me advice to become same as a simulation result.

  • Hello,

    Give us some time to look into these settings you have made and we will get back to you shortly.

  • Hi There,

    Can you provide a phase noise plot of your OCXO?

  • Hello,

    Thank you for your reply.

    I submit measurement data of OCXO.

    Regards,

    Arai

  • Hi Arai-san,

    Your OCXO phase noise is very good, hope the spurs are not real and is just a measurement issue.

    Given this good phase noise, I don't know why the phase noise of LMX2820 is bad. 

    Were you using the "EVM default mode" setting but just change the output frequency to 1.7GHz?

    Could you also try other reference clock? For example, signal generator? You could try with 800MHz reference clock and then set Pre-R = 4 to get the same 200MHz fpd. 

  • Hi Noel-san,

    I set it and measure in the "EVM default mode".

    When fout=6GHz(RFOUTAN)

    When REF IN = + 10 dBm

    I seemed to improve some results, but the result still has a problem.

    Other REF input is difficult, want to wait.

    A phase noise has difficult good signal source.

  • The second photograph is a mistake.

  • I show addition data.

    REF=1GHz OCXO, fpd=200MHz(Pre-R=5),


    Please give me advice.

  • Hi Arai-san,

    When did you order this EVM? What is the part number as shown in the top marking of the device? Is it LMX or PLMX?

    The noise bump between 10kHz and 100kHz could be reduced by increasing the bypassing capacitor at pin 3. Change this capacitor to 4.7µF or larger.

    Higher input frequency or input power improves phase noise suggests that the slew rate of the reference clock is also important. If possible, use a square wave reference clock. Also try use a sma connector (aka I-join) instead of sma cable to connect the OCXO and EVM together.

  • Hi Noel-san,

    >When did you order this EVM? What is the part number as shown in the top marking of the device? Is it LMX or PLMX?

    1. It is "LMX"

        order number : T01091457

    > The noise bump between 10kHz and 100kHz could be reduced by increasing the bypassing capacitor at pin 3. Change this capacitor to 4.7µF or larger.

    2. I will try this after.

    > Higher input frequency or input power improves phase noise suggests that the slew rate of the reference clock is also important. If possible, use a square wave reference clock.

    > Also try use a sma connector (aka I-join) instead of sma cable to connect the OCXO and EVM together.

    3. Square wave reference clock is difficult.

       Is "I-join" connected by SMA-M to SMA-M.

       the OCXO and EVM because SMA-Female.

  • Hi Arai-san,

    You've got the latest silicon, your EVM is fine.

    Yes, use male sma i-join like the picture in below.

  • Please give me some time.

  • Hi Noel-san,

    I'm sorry late reply.

    I report result of a measurement.

    > The noise bump between 10kHz and 100kHz could be reduced by increasing the bypassing capacitor at pin 3. Change this capacitor to 4.7µF or larger.

    Changed to 2.2uF.

    Phase noise has improved by about 0.5 dB.

    Do I need to change Register2's INSTCAL_DLY as a result of the change?

    Also, I'm not using INSTCAL_DLY right now. (default)

    > Also try use a sma connector (aka I-join) instead of sma cable to connect the OCXO and EVM together.

    Phase noise is about 1dB worse.

    Please give me advice.

  • Is REFERENCE 100MHz square wave in figure 8-2 on page 30 of the data sheet?

    Is it a sine wave?

  • Hi Arai-san,

    Yes, and the output power is 10dBm.

  • I understand that the reference signal is a +10 dBm sine wave.

    Even if the input level of the reference signal is changed to +10 dBm, the simulation and the measured value are still different.

    Do you have any other advice? Or is it possible to get a demo?

  • Hi Arai-san,

    I am afraid I don't have any cue for phase noise difference. Not sure if there is other factor, other than slew rate and phase noise, that may affect the PLL noise of this device. This requires further investigation from our side.

  • Just FYI, I've spent some time with the LMX2820 eval module recently and found that it is relatively sensitive to power supply noise.  The LMX2820 data sheet says, "The internal power supplies are provided by integrated LDOs, eliminating the need for high-performance external LDOs," but this is a bit optimistic on TI's part. Slight smile

    Try powering the board from a dedicated low noise 3.3V LDO instead of running it directly from your bench supply.  I've seen good results with the LT1963 followed by some additional LC filtering (pi filter with 27 uH and 10 uF ceramic on both sides).

  • Hi Noel-san,

    We can't confirm it anymore.

    Does that mean you can't demonstrate?
    Can TI remeasure the LMX2820EVM at 6GHz and show the data?
    Please disclose the power supply, reference signal, and reference signal level used for the measurement conditions.
    Also, please give me a measuring instrument.

  • Hi John-san,

    Thank you for your advice.Slight smile

    I will try.

  • I supplied a power supply with 3.3V through LDO.(LDO=ADM7151ACPZ-04)

    But, the phase noise did not improve.

    I do not have what to do.

    Please help me.Dizzy face

  • Hi Arai-san,

    I will ask my team to take some data to you, please stay tuned.

  • Hi Noel-san,

    How is the investigation after than going?

  • Hi Arai-san,

    Sorry to keep you waiting but I am sorry to tell you that the team has been busy with new product development, all the test equipment is occupied at the moment. Let me push again and get you something in next week. Thank you for your patience. 

  • Hi Arai-san,

    Here is the plot, with EVM default mode and change the output divider to Divided by 4. Reference clock is 100MHz Wenzel.

  • Hi Noel-san,

    Thank you for measuring the data.

    Is this data measured using mass-produced EVM?

    Actually, our USB2ANY has a problem.

    For this reason, we are using the supplied USB2ANY of LMX2592EVM.

    Is there any change in measurement due to the difference?

    USB2ANY we received had an initial failure and have never been able to use it.

    To request a replacement, we’ve called TI support team, opened a support request, asked on e2e , and we’ve also requested product return.

    However, none of them were in charge for the replacement, and our request was declined.

    We just want USB2ANY to be replaced. Can you support our request?

  • Hi Arai-san,

    Every USB2ANY is identical, you can use the USB2ANY that came with LMX2592EVM.

    I cannot replace your USB2ANY, I don't have this capability.

    If you use a new PC that have never been connected to an USB2ANY (or, you can remove the USB2ANY driver from the PC that you have been using with USB2ANY), when you plug in the non-working USB2ANY, will you get prompted to update the firmware?