LMX2572: Muxout pin & Lock Detect in mode VCO Cal + VTUNE

Hi Noel,

Regarding the lock detect.

The understanding of how the LD work is quite OK for us. However we observe "bounces" of the LD muxout pin depending of the configuration. We observe the MUXOUT in lock detect mode for several frequency jumps without removing the reference clock after a jump.

Here is an example of what we observe in partial assist mode with:

• Oscin = Fpfd = 100 MHz
• Jumping from 3300 MHz to 3310 MHz
• LD_TYPE = 0 (VCO)

Here, the same configuration but with LD_TYPE = 1 (VCO+VTUNE):

This behaviour is ok for us and coherent with the datasheet.

 Here, the same configuration but jumping respectively from 3300 MHz to 3500 MHz:

Now, this is what we observe in full assist mode with:

• Oscin = Fpfd = 100 MHz
• Jumping 3300 MHz to 3310 MHz
• LD_TYPE = 1

Here, the same configuration but jumping respectively from 3400 MHz to 6400 MHz and from 6400 MHz to 3400 MHz :

...

Our question is about the LD HIGH and LOW alternance between what is look like the lock time.

Regarding the Block programming.

We have tested the block programming in case of ADD_HOLD equals to 0. We observed, it does the same operation as described in the datasheet section "7.5.4 Block programming" p22, which is set the beginning address an give just the data bytes in decreasing order.

For ADD_HOLD = 1, it is not obvious for us how it is working. As we read it in the datasheet, it looks like it allows us to freeze address decrement in block programming and write the same register sending only data bytes. We would like to know how to start and stop block programming with ADD_HOLD=1. It is not describe in the LMX2572 datasheet as expected  in the block programming section. We suppose it works as described in the below example:

• Change ADD_HOLD = 1 in R0
• Pull CSB pin LOW
• Write 0x7Caaaa for R124
• Write 0xbbbb for R124, followed by 0xcccc for R124, and so on.
• After the last register write is completed, pull CSB pin HIGH to finish Block Programming.
• Change ADD_HOLD = 0 in R0

Are we right on the way it is working with ADD_HOLD equals to 1 ?

Hi Malo,

With LD_TYPE = 1, lock detect will validate Vtune, if Vtune is bouncing between the thresholds, then MUXOUT will also be toggling. Check fout vs time to confirm this.

For the block programming, I am pretty sure we need ADD_HOLD = 1, I will verify this again tomorrow.

Hi Malo,

I checked block programming, right, it works only when ADD_HOLD = 0, I need to confirm this with the designer, I will get back to you later.