ADRV9009 multichip sync lost after frequency hopping

Are you hoping the frequency directly from 600MHz to 800MHz and vice versa? hope you are running init calls every time you are hoping(since the change is more than 100MHz). Are you running the full init sequence along with both RFPLL and BBPLL  MCS tracking enabled when you are hoping frequencies?

• Are you hoping the frequency directly from 600MHz to 800MHz and vice versa? hope you are running init calls every time you are hoping(since the change is more than 100MHz).

Yes，I hoping the frequency directly from 600MHz to 800MHz and then from 800MHz to 600MHz . But I did not run init calls do other operation , like exit fhm or re-run

• Are you running the full init sequence along with both RFPLL and BBPLL  MCS tracking enabled when you are hoping frequencies?

NO, after hopping, I did't do other operations, not re-run init sequence an not exit fhm. I thougth the member enableMcsSync in taliseFhmMode_t struct  set to 1 will be worked for hopping . And the most important reason is that I need more than 1000 hops per second, if I re-run the full init sequence, It must be out of my requirement.

I also found if I just exit the async chip's FHM mode (full exit), its tow channels will be synced again. I viewed the topic:FHM Calibration Resolution, I think exiting FHM will cause the calibration and tracking functions to re-run, that cause their phase re-sync again. But from exit  to enter fhm and vice versa will consume a lot of time, especially the commands of the three chips need to be transmitted in sequence by SPI.

• Are you hoping the frequency directly from 600MHz to 800MHz and vice versa? hope you are running init calls every time you are hoping(since the change is more than 100MHz).

Yes，I hoping the frequency directly from 600MHz to 800MHz and then from 800MHz to 600MHz . But I did not run init calls do other operation , like exit fhm or re-run

• Are you running the full init sequence along with both RFPLL and BBPLL  MCS tracking enabled when you are hoping frequencies?

NO, after hopping, I did't do other operations, not re-run init sequence an not exit fhm. I thougth the member enableMcsSync in taliseFhmMode_t struct  set to 1 will be worked for hopping . And the most important reason is that I need more than 1000 hops per second, if I re-run the full init sequence, It must be out of my requirement.

I also found if I just exit the async chip's FHM mode (full exit), its tow channels will be synced again. I viewed the topic:FHM Calibration Resolution, I think exiting FHM will cause the calibration and tracking functions to re-run, that cause their phase re-sync again. But from exit  to enter fhm and vice versa will consume a lot of time, especially the commands of the three chips need to be transmitted in sequence by SPI.

Can you change the frequency from 600Mhz to 700MHz and then check the phase sync? Can you check if the phase sync is there if you stay at a particular frequency for a long time? 

I have tested the frequency from 600MHz to 610MHz and from 600MHz to 750MHz, It was also somtime async. If all channels are synced, they are always synchronized during my long observation period. If some channels are asynced  with other, the asynced channels are always asynchronized during my long observation period.

Hope you following the RFPLL frequency change procedure as mentioned in UG.

Are you seeing the phase change only in the third transceiver?

• Hope you following the RFPLL frequency change procedure as mentioned in UG.

RFPLL frequency change procedure only describes the normal way to change RFPLL freq, but not the freq hopping mode. As my understanding of the UG, the freq hopping mode includes all operations that chaper of RFPLL frequency change procedureI describe. I have no need to do the RFPLL frequency change procedure manully. Is that right?

• Are you seeing the phase change only in the third transceiver?

No, all transceivers may change phase, it 's random.

Can you try changing the frequency manually instead of FHM (radio off --PLL freq set--radio on)and then check the phase sync? 

I tested LO changing manually, all channels phase was synced , async  phenomenon did not appear.

Can you compare the QEC performance when you are changing freq using FHM vs normal frequency change where phase change was not there?

I encountered the same issue. Have you resolved it?