BLE is not advertising on customized nrf52832 board

[Navdeep]

Hi,

 I have designed a customized for nrf52832 with a trace antenna. I haven't used any low pass filter and made the changes inside sdkconfig.h file to use RC and that board worked fine (ble was advertising).

Now, I have designed a new customized board for the same but using chip antenna (712-1006-1-ND)  this time and also add low pass filter with 12pf capacitors. The part numbers for LPKF and antenna are given:

Antenna- 2450AT18B100E

LPKF- FC-135 32.7680KA-AC0

 But the problem is, the BLE is not advertising at all this time. I have removed the low pass filter also just to check if that causing the problem still its not advertising.

I have attached the screenshots for schematic and pcb. This time the board thickness is 0.6mm instead of 1.6mm and according to that thickness , the transmission line width is calculated as 47mils. Please let me know if that thickness is causing any issue? because as compared to the previous board( which was working fine), I have just done 3 changes-

1. changed the board thickness from 1.6mm to 0.6mm

2. add lpkf crystal

3. and changed the trace antenna to chip antenna

4. Also I have changed 0.8pf capacitor that I'm connectimg with the antenna from 0402 pack to 0201 (25v) (which is not the issue, i guess)

Also, is it possible that with decrease in board thickness, the 32mhz crystal is not working fine? ( Crystal orientation is perfect)

[kphalak]

So you are saying, its the same project with same firmware hex file, just hardware changes as you mentioned?
you should changed the configuration settings in SDK_config.h regarding 32kHz LF crystal you added.

Did you try on board debugging? does BLE_STACK_INIT function execute successfully?

[Navdeep]

No, I'm not using the smae hex file, I have already changed the SDK_config.h file as new board is having LF crystal.

I have seen alot of posts from for "BLE advertising problem on customized boards" and most of the times, it was because of 32Mhz crystal  wrong orientation, but in my case the orientation is fine

[kphalak]

Components are fine but I have had read someone mentioning their circuit started working after changing capacitor value to 22pF.

[Navdeep]

Actually , the formula to calculate the capacitor value is

C=2CL- Cpcb-Cpin ,where Cpcb+Cpin=4pf , Cl=8pf of crystal

So , C=12pf. But according to attached screenshot, this Cpcb can vary according to the layout, so may br this time, due to decrease in PCB thickness and layout thing, this Cpcb is varying and may be I have to play with C1 and C2 values. What do you say kphalak

[haakonsh]

The frequency tolerance and accuracy is to low. You need to aim for a maximum of 50ppm = Ftol + Facc + Aging.

Also The loading capacitors might be too far away from their pins, the trace between C4 and its pin is about 2mm which adds ~2nH inductance. The GND plane is also split and the decoupling capacitors ground connection is therefore through a via who adds an additional ~1nH inductance. This might prevent the crystal oscillator from starting up. 

[kphalak]

I was also thinking so haakonsh, but I had used that exact part number for crystal and it works fine. So its at least not a crystal to blame here.
But Ya good point about loading capacitors. I am also eager to see what will end up solving the issue.

[Navdeep]

The problem was of ground plane. The 49th pin was not grounded properly through the ground plane. we changed the design and its working fine now. Thanks for your help guys.