2016-03-10 12:04 PM
Hello,
I'm experiencing an issue with a STM32F101 using an 32.768kHz LSE.The LSE being used is the 7pf variant shown in the link below with 9pf caps as CL1 & CL2http://www.ecsxtal.com/store/pdf/ecx-71.pdfThe problem is that periodically(every 25 to 50 boards) the LSE does not start.Does anyone have an idea what might be happening here?2016-03-10 01:05 PM
Well the oscillator is quite fussy. If you've eliminated gross failures like opens/shorts, or wrong components then you should perhaps try different caps. I'll presume you picked 9pF based on the app note computation.
ST designs use these, with 6.8pF capshttp://search.alkon.net/cgi-bin/pdf.pl?pdfname=07353.pdf
2016-03-11 08:45 AM
Hi Clive,
Correct the 9pf CL1 & CL2 were calculated from the AN2867 formulas.I have changed the capacitor values to 4.7 pf and yes in fact it starts to work. The troubling part is that if I then change the caps back to 9pf it continues to work so I'm not able to establish that in fact it was the 9pf cap that was really the issue.The next time I have this happen(if ever) I will try a parallel feedback resistor across the crystal to help startup.Pat2016-03-12 12:52 PM
> The problem is that periodically(every 25 to 50 boards) the LSE does not start.
Do you wait long enough for the LSE to start? JW2016-03-13 03:45 AM
Hi!
How you change capacitors on PCB? Solder tip or hot air? Maybe the initial PCB has something that dissapears (violates) after heat treatement?2016-03-14 06:20 AM
Hi Jan,
Yes in fact I left one over night just to see if it would startup at all.Pat2016-03-14 06:24 AM
Hi Dmitry,
I had wondered about this as well so I tried to just re-work a few boards with both hot air and a solder iron but it had no effect.2016-03-14 07:00 AM
>> Do you wait long enough for the LSE to start?
> Yes in fact I left one over night just to see if it would startup at all. Well, that sounds to be long enough indeed... :D Dunno, then; the proper capacitors' usage as Clive said is an imperative though. JW