• Resolved

LMX2594: some issues about vco calibraion of lmx2594

Prodigy 140 points

Replies: 7

Views: 183

Part Number: LMX2594


I have some issues about VCO calibration of  lmx2594:

 1.My RFout of lmx2594 is 7~8.3GHz,and i do no assist for vco calibration.Does this mean the vco will calibrate automatically when i set Fcal_en=1?And what's the reliablility of it?Does this calibration will change with temperature(-40C~25C~50C) change?If it is does change whether it means the vco unlock?And it maybe need calibration again?

2. For the vco calibration,I see some recommends from the lateset datasheet in table 5 of page 23.And i compared my settings,I found i didn't set the VCO_SEL&vco_daciset_strt&vco_capctrl_strt as the recommends. I chose a frequency point(eg.7671.6Mhz) in the band I used above and then exported the register configuration from TICS as the initialization registers configuration.When i change to other frequencies,I just program N-divider value and PLL numerator and denominator and then program FCAL_EN (R0[3]) = 1.Does this matter?Or I'd better configure the three parts according to your recommendation?

3.If my RFout change to 9~12.3G,did i have to use two initial parameter for vco calibration?One is for 11.9~12.1G,the other is for other frequeny bands.


  • Hi there,

    1. VCO calibration will be initiated when R0 is programmed with FCAL_EN=1. VCO calibration speed depends on the reference clock frequency. The max. speed is 200MHz. Use CAL_CLK_DIV to change this frequency. Lower speed will ensure the calibration is more robust. So we can always use a lower speed well below 200MHz. Furthermore, LMX2594 has internal LDOs, they are power up when Vcc is applied. It takes time for the LDOs to power up and settle down. It is recommended not to program the device until it is power up for, say a few milliseconds. Or follow datasheet's recommended initial power up programming sequence.

    2,3. your programming sequence is correct. The suggested setting in table 5 returns shortest VCO calibration time. You may not necessary follow. However, I would suggest use VCO_SEL=4, VCO_DACISET_STRT=300 and VCO_CAPCTRL_STRT=1 for all VCO frequencies.

  • In reply to Noel Fung:

    OK,thank you.I will take your advice in my design.~~~

  • In reply to Noel Fung:

    hi Noel,

     For the vco calibration, I have another question as follows.I set the same frequency on my board,but  when i reboot my board several times,i have found some differences in reg112 when i readback eachtime. eg.the frequency is 7625.24M,reg 112 may be 0x0111,0xd0,0xcd,0x00cb,0x00c8 and so on,while the reg110(0x0428)and reg111(0x0093 or 0x0094) are basically the same when readback.did it work normally?


  • In reply to user5959213:

    Hi there,

    Please try use a lower state machine clock and redo the test again. Or you can try set ACAL_CMP_DLY to a higher value, for example 20.

  • In reply to Noel Fung:

    Hi Noel,

      Yes,I have already lower my state machine clock with change R1 from 0x0809 to 0x080A(my OSC=368.64M).

      I have another question,If i do VCO auto calibration at 25C,but my board should work from -40C to 55C,which means 25C configuration should work over the whole temperature.If vco cannot run robust,then there maybe unlock in lower or higher temperature.Do you have some suggestions for me to avoid this happen?


  • In reply to user5959213:

    Hi there,

    We support 125C continue lock without re-program. That is, if you lock it a -40C and then bring the temperature to 85C, it will remain lock.

  • In reply to Noel Fung:

    Hi Noel,

      Yes,I know what you mean,and i believe most chips of lmx2594 on my board work as you say.But there is always one chip or more(until now we found one) doesn't work as it was stated.As a designer, i just want all of them work robust.So I see what else I can do to make them work robust.