>Did anyone have this problem before?

Yes, we have.  We have gone through some great pains trying to get
stability in our 32 kHz oscillator circuits.  In our case, the
oscillator would start OK, but we found that at times, they would just
stop for no apparent reason.

Run the negative resistance test indicated in the TI application note
below to see the proper amount of capacitance to use.  We use an 'F447
and found that using the internal capacitance was much better at
keeping stability than using external caps.

Here is the link for the app note.  I wish they had this app note out
a year sooner!  The whole note is good, but especially pay attention
to section 4.2 "Negative Resistance Method".  This was the big thing
that showed ust we needed to use internal caps.

http://focus.ti.com/lit/an/slaa322/slaa322.pdf

feilip@gmail.com wrote:

> I am using MSP430F1101 with a 32.768kHz crystal. I have 100 samples
> fabricated but I find that 20% of them could not start.

One problem I had with 32 kHz xtals on MSP430 was soldering residue 
around the xtal circuit, which was a high impedence short.  After a very 
good washing they worked.

-- 
Thad

feilip@gmail.com wrote:
> Hi,
> 
> I am using MSP430F1101 with a 32.768kHz crystal. I have 100 samples
> fabricated but I find that 20% of them could not start. At the
> beginning of the firmware, I have a loop to lock the DCO frequency to
> 2MHz using the external 32.768kHz as reference. I suspect the MSP430
> cannot lock the DCO and stay in the loop forever.
> 
> Did anyone have this problem before?

Just some thoughts:

It's not the best idea to sync RC to XTAL only once. It should be done 
periodically due to temperature fluctuation etc.

The sample code shown by TI in their app note on syncing RC to stable 
frequency source is incorrect. Under some conditions the RC frequency 
will fluctuate significantly due to an error in  RC adjustment routine.

<feilip@gmail.com> wrote in message
news:1144592958.136145.289960@j33g2000cwa.googlegroups.com...
> Hi,
>
> I am using MSP430F1101 with a 32.768kHz crystal. I have 100 samples
> fabricated but I find that 20% of them could not start. At the
> beginning of the firmware, I have a loop to lock the DCO frequency to
> 2MHz using the external 32.768kHz as reference. I suspect the MSP430
> cannot lock the DCO and stay in the loop forever.
>
> Did anyone have this problem before?
>
> Thanks for your help!
> Feilip
>

I can't help you directly Feilip, but we had some problems in the past
similar to what you are experiencing and found one xtal seemed to work best,
manuf was Cosonic, I think the part number was HT38 or similar (micro was
'F149)

I heard recently of very similar problems with another local manufacturer
who I was told changed micros.

TI have an application note that might help, I think it was this one -
http://focus.ti.com/docs/mcu/catalog/resources/appnoteabstract.jhtml?familyI
d=342&abstractName=slaa225

feilip@gmail.com wrote:

> I am using MSP430F1101 with a 32.768kHz crystal. I have 100 samples
> fabricated but I find that 20% of them could not start. At the

You verified the 32KHz is actually oscillating? Is it stable? You'll
never lock if the xtal is wobbling around.

Have you checked the app notes for the xtal? What's your Vcc? Some
xtals specify a 5.1Mohm resistor to ground on one leg if Vcc < 3V for
MSP430 applications.

Hi,

I am using MSP430F1101 with a 32.768kHz crystal. I have 100 samples
fabricated but I find that 20% of them could not start. At the
beginning of the firmware, I have a loop to lock the DCO frequency to
2MHz using the external 32.768kHz as reference. I suspect the MSP430
cannot lock the DCO and stay in the loop forever.

Did anyone have this problem before?

Thanks for your help!
Feilip