AWR1642: If AWR1642 isolate Digital ground and Analog ground, any risk?

HI Vivek

We found that noise is from PWM swith GND, not power supply, because our board design don't have enough board space to let RF antenna keep away from  power componenets.

PS: RF antenna is on top side of PCB, but another side(bot side) is for Power components .ex: PMIC, and "12V=>5V" PWM power solution.

It just wants to isolate digital and analog ground, but I knew AWR1642 document don't have the suggestion.

So do you have any concern or risk to let us check? Or how to do the design? Thanks,

HI Vivek

Our PCB is 8 layers design. L1/L2 are rogers RO3003 and L5~L8 layers are FR4. L2 is for whole antenna's reference GND of rogers RO3003 and L5/L7 are for whole GND. So total have solid GND planes on our boards, and I think GND plane is enough. But it is not good design for the placement overlap between top side-PCB antenna and  bot side- PMIC power design, so it seems antenna is coupled from PWM Switch GND noise.  

Below analysis is why we think RF spur is from PWM switch GND , not from power supply.

1.) enable PMIC(LP87524B/J-Q1) Spread-Spectrum Mode to reduce 4MHz Switching Frequency noise.

      it can reduce 10dBc.( orignal noise: noise floor: 45dBc/ this RF spur: 63dBc, then if enable SSM, it becomes 53dBc from 63dBc.)

2.)When rework another one board provide power supply/GND -" PMIC*4 power-3.3&1.8&1.2&2.3V, GND" to  2nd DUT" , 2nd DUT don't have RF spur.

    PS: PMIC power of 2nd DUT don't work, so don't have 4Mhz switch frequency noise   

3.) When rework another one board provide power supply/GND -" PMIC*2 power-1.2&1.8V" to 2nd DUT, 2nd DUT still don't have RF spur.

     But when only add 2.3V power loading(ps: add one parallel 5 ohm on 2.3V power supply), it becomes to have  this RF spur.

    PS: PMIC*4 power consumption checking is as below.

          When 2nd DUT remove 1.2V/1.8 power supply from another one board power supply, 2nd DUTdon't have RF spur. Then when 2nd DUT add 2.3V power loading(PS: parallel 5 ohm), it still has RF spur .        

          (1) 3.3V- lowest power current.

          (2) 1.2V-~ 800mA( peak)

          (3) 1.8V-~1.2~1.3A(peak)

          (4) 2.3V-~400mA(peak)

Hi Vivek

Our design is co-layout LP87524B and LP87524J, and the analysis is the same result about PWM switch GND noise if follow up our check items.

We add 5 ohm parallel on  2.3V's output. It means 2.3V add  2.3V/10 ohm power current.

DUT1/DU2 two are for the same design boards. DUT1 is only for power supply and GND and DUT2 is for RF spur measurement board.

As list to you about item2 last feedback, 

     (2) We rework to turn off DUT2's PMIC*4 power , and then use wire to connect DUT1's power supply , so DUT2 board don't have PMIC 4MHz PWM  switch frequency , and DUT2 use the same design power

          supply from DUT1. It seems RF spur disappear.

As list to you about item3 last feedback, 

     (3) We also rework to trun off DUT2's PMIC's 1.2V&1.8V(ps: 3.3V&2.3V still work on DUT2), and then use wire to connect DUT1's PMIC's 1.2V&1.8V.

          So DUT2 board also have 4MHz PWM switch frequrency, but this noise should be lower. It seems RF spur disappear.

         next step, 

                        We only rework to add DUT2's 2.3V's power loading(PS:  add 5 ohm parallel on  2.3V's output. It means 2.3V add  2.3V/10 ohm power current.), so PWM 4MHz PWM  switch frequency becomes     

                        strong.

                        Then this measurement could get RF spur for 4Mhz noise.            

    So why do you still think it is power supply coupling, not for PMIC PWM swith GND?