Improving moisture and flux immunity of 32768 xtal circuit?

[BasePointer]

Hi,

We are using low power external crystal 32768Hz with Rf: 10Mohm, Rs: 4.7kohm, load capacitances 18pF and 12pF. Actually it works without any problems under normal conditions. We lived some problems after mass productions with our crystal circuit. Most of them didn't work (doesn't start-up). To modify them, we unpluged the MCU, cleaned the PCB and soldered the same MCU again. They start working.

So, I wonder if there is a way to improve moisture and flux immunity of our low power 32768 xtal circuit.

MCU is LC60 and we have to use that High Gain Oscilator(HGO) is disabled for our application.

Thank you.

[bigmac]

Hello BP,

If removing and re-soldering the MCU then results in normal operation of the oscillator, this is suggestive that you either have a short circuit (solder bridge) or an open circuit (bad joint).  Rather than removing the MCU, I think you really need to identify the actual cause in each production unit, to see if there is a common pattern.

Using a high Z oscilloscope probe at the oscillator pins of the MCU might help - these should be biased to about half Vdd.  If the voltage at either pin corresponds to a supply rail, this is indicative of a problem.

I would doubt that moisture would represent an issue during factory testing.  If you think that there is flux or moisture present you should be able to confirm whether this is the problem by cleaning the existing boards, and then properly drying them.  Of course, the act of cleaning the board may also have an effect on an intermittent solder joint.

If you are using RoHS manufacture, identification of suspect solder joints is probably more  difficult.

Regards,

Mac

[Alban]

Hello,

- How did you determine the Rs value ?
A Negative Resistance Margin test is usually to be done.

- It could also be bad soldering in production with voids/bad contacts/increased resistance.

32kHz oscillators are not very sensitive from my experience, therefore I am surprised you have oscillator start-up issues.

Alban.

[BasePointer]

Hi Alban,

I'm not sure that negative resistance test can be done for low power oscilator mode(HGO=0) of LC60. Because the datasheet of LC60 (page 339) indicates for Rs typical 0ohm and max:10kohm.

Actually, even if I use 220kohm for Rs, it works. I reduced it to 4.7kohm due to some loss of lock problems that we lived at EMC tests. Any lower value than 2k2ohm for Rs causes crystal to stop.

If I set HGO=1 before desoldering MCU that we lived start-up problem in production, the crystal starts oscillating. So I think this is not a void/bad contact/increased resistance problem.

Regards,

BP.