Troubleshooting HF clock that is not starting

[mark.blackburn.orpyx]

I am troubleshooting a board with no working RF. On looking deeper it appears that the HFCLK is not starting (NRF_CLOCK->EVENTS_HFCLKSTARTED is never set).

This is on a PCB that was working previously. I want to troubleshoot it deeper to see if it's a

* Soldering issue

* Component issue (capacitors or crystal)

* ESD issue

If I assume the crystal is not working, what is the expected behaviour of the XC1 and XC2 pins on a nRF52840 when the HFCLK is started by setting NRF_CLOCK->TASKS_HFCLKSTART

Does one of these pins get driven high? If so which one?

Is there a way to tell if the nRF52840 XC1/XC2 pins have become unsoldered. I was thinking that I could do a diode check between XC1/XC2 and VCC and GND on the assumption there are ESD protection diodes on these pins. Would that work? Which power rail should I use for this check? VDD, VDDH, or one of the DEC pins?

[mark.blackburn.orpyx]

> Seems like XC2 line isn't driven high for some reason.

Should I suspect that something is wrong with the crystal, capacitor or their soldering? Or could something be wrong with the XC2 pin on the IC?

Yes, the load capacitance of the crystal is 10 pF..

[Jared]

Hi,

mark.blackburn.orpyx said:

Should I suspect that something is wrong with the crystal, capacitor or their soldering? Or could something be wrong with the XC2 pin on the IC?

I would suspect either the crystal or the soldering. Have you tried changing the crystal on the PCB?

mark.blackburn.orpyx said:

Yes, the load capacitance of the crystal is 10 pF

mark.blackburn.orpyx said:

There are 2 12 pF load capacitors connected to the XC1 and XC2 pins and to ground.

Then the crystal isn't properly loaded. The parallel load capacitors should be 16 pF if the internal load capacitance is 10 pF. However, I don't think this is the culprit in your case. But it would be good to change them to the correct size later. 

regards

Jared 

[mark.blackburn.orpyx]

I tried changing the crystal. It still won't start.

[Jared]

Hi,

What percentage of samples produced is affected by this? Could you try changing the IC from a working sample to a non-working sample and see if it fixes the problem? It could help us understand what's causing this. 

regards

Jared 

[mark.blackburn.orpyx]

> What percentage of samples produced is affected by this? 

We haven't characterized are failures to this detail yet. There may be other failures like this but I haven't confirmed. We have other units where the radio (BLE) is not working, but haven't confirmed if the crystal oscillator is the cause. Do you have an idea how often this might occur?

> Could you try changing the IC from a working sample to a non-working sample and see if it fixes the problem?

I don't have that capability without taking the board to our contract manufacturer. I'll perhaps reassess if we think this is occurring often enough to do that work.