HFXO does not work on nRF52832

[silvio3105]

Hi,

I cannot start HFXO on nRF52832. I'm using 32MHz crystal TSX-3225 32.0000MF20X-W0(from Mouser) wtih 12pF caps UMK105CG120JV-F(also from Mouser).

Checked with beeper on multimeter, no shorts between pins and between pin and GND.

Looks ok on MCU pin side(QFN package).

This is code

int main(void)

{

    NRF_LOG_INIT(NULL);

    NRF_LOG_DEFAULT_BACKENDS_INIT();

    NRF_CLOCK->TASKS_HFCLKSTART = 1;

    while (NRF_CLOCK->EVENTS_HFCLKSTARTED == 0);    

And PCB

I'm out of ideas.

[Kazi Afroza Sultana]

Hello,

We have an [68] CLOCK: EVENTS_HFCLKSTARTED can be generated before HFCLK is stable errata that could be relevant to your issue. If EVENTS_HFCLKSTARTED starts before HFCLK is started.

Can you check the Crystal start-up time of your 32MHz?

[rmptxf]

We also run into a similar issue, and the start time was more than a 1ms on best cases (not even fixed).

What should we do in that case of high start time?

[rmptxf]

This is how we get the startup time on NCS.

NRF_TIMER1->BITMODE = TIMER_BITMODE_BITMODE_32Bit << TIMER_BITMODE_BITMODE_Pos;
NRF_TIMER1->TASKS_CLEAR = 1;
NRF_CLOCK->EVENTS_HFCLKSTARTED = 0;
NRF_TIMER1->TASKS_START = 1;
NRF_CLOCK->TASKS_HFCLKSTART = 1;
while(NRF_CLOCK->EVENTS_HFCLKSTARTED == 0);
NRF_TIMER1->TASKS_CAPTURE[0] = 1;

LOG_INF("HF Clock has started. Startup time: %d uS", NRF_TIMER1->CC[0]);

[hmolesworth]

Most useful, we will adopt this. For other users it may be helpful to do the measurement in uSecs:

   NRF_TIMER1->PRESCALER = 4; // Time in uSecs when using 1MHz timer clock. fTIMER = 16 MHz / (2^PRESCALER)

[silvio3105]

Mine app stucks at while loop.

[hmolesworth]

The data sheet for that crystal implies 7pF CL although the Mouser listing states 12pF, so 12pF load caps are not correct in either case.

CL = (CaCb)/(Ca+Cb) + Cstray where Ca=Cb=12pF+Cpin=15pF and Cstray (not pins) say 1pF
     = 8.5pF

CL = (CaCb)/(Ca+Cb) + Cstray where Ca=Cb=9pF+Cpin=12pF and Cstray (not pins) say 1pF
     = 7pF

If the Mouser description of 12pF is correct then try 19pF

CL = (CaCb)/(Ca+Cb) + Cstray where Ca=Cb=19pF+Cpin=22pF and Cstray (not pins) say 1pF
     = 12pF

[silvio3105]

Mouser says 7pF https://hr.mouser.com/ProductDetail/Epson-Timing/TSX-3225-32.0000MF20X-W0?qs=f9yNj16SXrLFb%2FNL%252B2Lczg%3D%3D

Datasheet says 7pF to infinity

DigiKey says 12pF https://www.digikey.com/en/products/detail/epson/TSX-3225-32-0000MF20X-W0/7731972

I made mistake with not including 4pF for Cpin.
For 7/5pF caps have to be 4pF while for 12/5pF caps have to be 14pF, according to this calculator - https://ecsxtal.com/crystal-load-capacitance-calculator/

Seems like CL is 7pF since I guess 12pF will be enough for 12pF CL.

[silvio3105]

Tried 4pF, 8pF(stack od 2x4pF in parallel) and 20pF(2x4pF + 12pF in parallel) - still cannot start it.

I belive CL is 7pF as stated in datasheet. Calculated caps with "((8+4+1)*(8+4+1))/(2*(8+4+1))" - from nRF52832 clock section.

C = C1 = C2

(C + Cpin + Cs)^2 / (2 * (C + Cpin + Cs))

Cs = 1pF
Cpin = 4pF

Result is 6.5pF which is close to 7pF from datasheet.
Connection is tested, solders on MCU pins look good.

 Kazi Afroza Sultana 

[hmolesworth]

Perhaps try a different crystal; the part number TSX-3225 32.0000MF20X-W0 does not match the part number format on the data sheet. As an aside, nRF52832 pin capacitance is 3pF (as used in my formula), not 4pF, but it doesn't make much difference.

CPAD Pad capacitance 3 pF