the clock function is not working in esb_ptx example for nRF52832

[Masoumeh]

Hi, 

as I said before I am working with nRF52832. I can connect with nrf and program it. for example I can make an LED to blink. but when I run the esb_ptx example the program doesn't pass through the clock_initialize() line. I am using the CX2016DB32000D0WZRC1 chip as the crystal. can you help me to find the problem.   

[Einar Thorsrud]

Hi,

but when I run the esb_ptx example the program doesn't pass through the clock_initialize() line.

There is no function called clock_initialize in the SDK 17 (which I see you are using in another question). There is a clocks_start() in the esb_ptx example, though. Is that the one?

That is implemented like this:

void clocks_start( void )
{
    NRF_CLOCK->EVENTS_HFCLKSTARTED = 0;
    NRF_CLOCK->TASKS_HFCLKSTART = 1;

    while (NRF_CLOCK->EVENTS_HFCLKSTARTED == 0);
}

So the only thing that could make this not return is if the HFCLKSTARTED event never happens. Can you double check with a debugger if this really is the problem, and the while loop never exits? If so, there is likely a problem with the HF crystal on your HW.

[Masoumeh]

you are right, it is clock_start(). when I comment clock_start() the program runs properly and the LED start blinking, but when the clock_start() line is active, it stops there. I checked the crystal an I even changed it but noting happen. the crystal that i am using has all the features that is said in datasheet. 

[Einar Thorsrud]

Hi,

Masoumeh said:

can it be a problem with the crystal?  or the nrf52832 IC?

Yes. I cannot pinpoint exactly where the problem is, but the example should work and there is really no way the clocks_start() function could fail if the HW is functional. So something in your HW is preventing the HFXO from starting properly.

Have you had your custom board reviewed by Nordic? If not, I suggest you upload your schematics and layout and get it reviewed. If that is OK, then there is likely something fishy with the components, PCB, soldering or other, and so it make sense to test another board and see if that works or not (chances are you will not have the same physical defect on the next one).

[Masoumeh]

This is my PCB layout. in the right part of it you can see the nRF52832 and its components. the schematic of the nordic is exactly like datasheet. I just remove the LFCLK crystal. 

 PCB.PcbDoc

[Einar Thorsrud]

Hi,

There are several potential issues here:

1. Make sure the value of the load caps matches the CL value of the crystal: Ccap = CL *2 - 4 pF. 12 pF load caps, C19 and C22, is correct for a crystal with CL = 8 pF. If a different crystal is used, the load caps must be changes accordingly.
2. Make sure the crystal is soldered the correct way. Turning the crystal 90 degrees to make it easier to solder will not work.
3. The crystal you are using, CX2016DB32000D0WZRC1, is out of spec. The stability and tolerance is too high. The total needs be be +/-40 ppm or better.

[Masoumeh]

I ordered the crystal due to the spec in below link. 

https://www.mouser.com/ProductDetail/Kyocera-Electronic-Components/CX2016DB32000D0WZRC1?qs=mHG8nxlouK5LiZGb%2FcBxwg%3D%3D

you mean that the crystal cannot work for this IC?

[Einar Thorsrud]

Hi,

Masoumeh said:

you mean that the crystal cannot work for this IC?

The problem is the stability and tolerance. Sp it may or may not work, and it could vary from device to device. This means that if you are lucky it could be good enough to play with, but it is for sure not suitable for mass production as you can never know if a device will work or not without testing it (and happen to be lucky and get one that performs better than expected).

[Masoumeh]

Do you have any suggestion for crystal?

[Einar Thorsrud]

You can use Q22FA12800025, which is used on the DK, or any other that matches the requirements in the electrical specifications for the HFXO. Please refer to this post for more.

[Masoumeh]

is it possible to make a clock by signal generator and test the nRF52832 by this clock? does it make any problem for nRF? 

[Einar Thorsrud]

Hi,

In most cases, no. The radio is very sensible to phase noise, and a PLL based clock source will not be good enough, and that is what most signal generators use. You need a crystal based oscillator of some sort.

[Masoumeh]

I checked my Crystal by an inverter and it's working as you see in the picture. is it possible that the inverter that i used is capable of working with the crystal but the nRF52832 isn't? my inverter is 74AC04DR.