• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 22 subscribers
  • Views 720 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.

Original question:

LMX2594EVM: PLLatinum simulation tool - loop filter disign

LMX2594EVM: Register Values

Keith clark
Prodigy 130 points

Part Number: LMX2594EVM
Other Parts Discussed in Thread: LMX2594, LMX2595

Hi,

I have questions regarding register values for register 0 and 25 in the LMX2594 and wonder if you can help?

In register 0, D3 is responsible for FCAL_EN, should this be toggled (0 - 1 - 0) if so how long should it be high for, or can it be left as 1?

Regarding register 25, revision B of the data sheet shows that D11 = 0, D9 = 1, D3 = 0, D2 = 1, D1 = 0 and D0 = 1 - these appear to be fixed values and not responsible for any user defined functions. However, if I download the hex file from the evaluation board using TICS Pro all of these values are inverted. Which of these settings are correct?

I ask because we have problems with very high phase noise and wonder if one of these register settings could be a contributing factor?

Many thanks for any help you can offer,

Best regards

Keith

  • 0
    TI__Guru** 116070 points
    Hi Keith,

    R25 is not used, it value is not important at all. The silicon default value of this register is 0x624, so the datasheet is correct.
    We recommend keep FCAL_EN=1 all the time. Whenever you need a VCO calibration (e.g. change VCO frequency), after finish programming all other registers, program R0 to initial VCO calibration.

    Usually when the device is locked, the phase noise should be good. However, if PFD_DLY_SEL is not set right, then the phase noise could be very bad.
    You can send us your register setting to review. (preferably using TICS Pro to generate the register setting or .tcs file).
  • 0 in reply to Noel Fung
    Prodigy 130 points

    LMX2594 TI File.xlsxHi Noel,

    Thanks for your prompt reply.

    I have attached an Excel file that shows the register values for the LMX2594 from a stand alone LMX2594 development breadboard (low phase noise) and an integrated EQM module (high phase noise). The  breadboard ones were from a hex file from the TICS Pro file that programmed the development board and gives good phase noise response. The EQM values are sent to the chip via an on-board FPGA so a TICS Pro file in not available for this. The values for the registers are taken from the programming files instead.

    The high phase noise is not necessarily caused by these register values but could be caused by using a single ended HCMOS reference clock rather than an LVDS one. We plan on changing this shortly to see if we get improved performance.

    Another possibility will be to run this device from a separate LDO regulator however having run the chip from an external power supply this did not give improved phase noise performance even when the 2 other chips on the same power rail were disabled by removing the supply to them and disabling all other digital signals.

    Best regards

    Keith

  • 0 in reply to Keith clark
    TI__Mastermind 18070 points
    Hi Keith,

    I had a look at the excel file. The R0 is okay, the only difference between the two is FCAL_EN being checked and unchecked and MUXOUT_LD_SEL being in "readback" and "lock detect" mode. So R0 shouldn't be the reason. But please follow Noel's advice and set FCAL_EN to 1 if you use VCO calibration.

    R25 is not being used in LMX2594, this register is only used for LMX2595 to control the bias current of output doubler, and it has no effect on frequency below 15GHz. The Ticspro default changed this register from 0x0624 to 0x0C2B only to match with LMX2595 register settings. This register will be disclosed in the next revision of LMX2595 datasheet, but again, it has no impact at all on LMX2594.

    R75 sets the channel divider value but since you use direct VCO output this register also doesn't matter.

    Therefore, based on the given information, difference in R0, R25 and R75 shouldn't be the reason that causes phase noise degradation.

    The quality of reference input can greatly impact output phase noise. I suggest you use the same reference input source and input termination for an apple-to-apple comparison.

    Regards,
    Hao