• Resolved

LMX2595: Lock time at 12 GHz

Part Number: LMX2595

Good day

I am busy using the LMX2595EVM to evaluate the performance of the device.

As part of my investigations, I am looking at the lock time when performing frequency changes. This is done over a wide range of frequencies to see the impact of other peripherals (such as VCO dividers/multiplier, etc).

For reference, I am measuring the time between the last bit of R0 being sent (to trigger a VCO calibration) and LOCK DETECT going high.

For example, I will have a start frequency (FStart) of 4 GHz and a stop frequency (FStop) of 15 GHz). Or FStart of 8 GHz and FStop of 19 GHz. I'm sure you get the picture.

I have however noticed something strange. When FStop is 12 GHz, and Fstart falls within one of the other VCO Cores (ie not VCO4), then the lock time is drastically longer. If FStart is in the same VCO Core, then the problem is not there. 

At first, I thought it might be because 12 GHz falls on a VCO core edge (VCO4 fCoreMax), but frequencies that form the fCoreMin or fCoreMax of other VCO cores do not exhibit the same behaviour. 

There is a note at the bottom of page 21 of the LMX2595 datasheet that states: "In the range of 11900 to 12100 MHz, VCO assistance cannot be used, and the settings must be VCO_SEL = 4, VCO_DACISET_STRT = 300, and VCO_CAPCTRL_STRT = 1. Outside this range, in the partial assist for the VCO calibration, the VCO calibration is run. This means that if the settings are incorrect, the VCO still locks with the correct settings; the only consequence is that the calibration time might be a little longer. The closer the calibration settings are to the true final settings, the faster the VCO calibration will be."

That note is the only indication I could find in the datasheet that something strange is possibly happening at 12 GHz since I can't use the VCO assist feature at that frequency.

Is there some inherent instability at that frequency, or is it some other reason?



  • Carl,

    The longer lock time does have something to do with the note in the datasheet and the fact that 12 GHz is near the boundary of VCO4 and VCO5.  To ensure that we have the ability to hold lock over temperature we need to put these special settings.  I know it likely sounds confusing, but these settings actually ensure that the VCO picks VCO5 instead of VCO4 (even though VCO_SEL=VCO4) near the boundary.  By ensuring that we pick VCO5 at the boundary, we ensure lock over temperature.  However, in the process, these settings cause the calibration to bounce between VCOs and take longer.

    That being said, if you use full assist mode and use this method to get your calibration settings, then the full assist will not be increased.  This only applies to partial assist.


  • In reply to Dean Banerjee:

    Hi Dean

    Thank you for your reply.

    Initial testing has shown dramatic improvement when using Full Assist. I still have many tests to run, but for now I will mark this issue as resolved.