[SOLVED] Crystal with incorrect load capacitors.

We recently discovered that a 16 Mhz crystal, which is specified for 8 pF load capacitors, by mistake the assembly house fitted 12 pF instead. It drives an Atmel AT90 processor at 5v.

The equipment passed all of our electrical tests, and so far has been working for a few months with the customer.
I've been asked to make a risk assessment of the situation.

My feeling is that the only things that could happen is a slight frequency offset and perhaps a longer startup time.....but from a reliability sandpoint, what would happen?

My feeling is that the only things that could happen is a slight frequency offset and perhaps a longer startup time

Click to expand...

I think that you are correct, Providing that the start up is reliable over the full temperature range then the capacitors will be fine. A increased value of capacitors normally improves startup reliability over the temprature range at the expense of start up time. The increased value of the capacitors will not have a detrimental effect on the crystal.

Yes, you will be okay. The load caps only "complete" the xtal circuit for the specified frequency. Unless you perform critical timing based on the xtal, you will be okay. The frequency will be slightly lower than 16MHz.

Crystal load capacitance matters in two regards:
- the required load capacitance to achieve the nominal frequency, as specified in the crystal datasheet. The specification is sometimes misunderstood, it's the total capacitance seen by the crystal, not the value of the capacitors connected to both crystal terminals.
- the capacitance range that allows reliable operation of the crystal oscillator. Although it depends weakly on the crystal series resistance, it's mostly given by the oscillator parameters and is specified in the microcontroller datasheet.

In terms of reliable operation over production process, voltage and temperature variation, you'll look primarily for the second specification.

Exact frequency matters for applications like RTC or RF transmitters.

schmitt trigger said:

We recently discovered that a 16 Mhz crystal, which is specified for 8 pF load capacitors, by mistake the assembly house fitted 12 pF instead.

Click to expand...

The external load caps will be in parallel with the internal capacitance of the crystal (which will be 1-2 pF most likely) and also in series with the capacitance of the voltage driver (that is part of the micro).

The original high-Q of the crystal will be somewhat broadened (no change in the frequency though) when you consider the external added load capacitors. The crystal may even appear as an inductor to the load capacitors because of the "too small" internal capacitance.

I presume that there may be a longer startup delay but why there will be a frequency off-set? Unless the Q is reduced so much that...

Thanks everyone

I presume that there may be a longer startup delay but why there will be a frequency off-set?

Click to expand...

The crystal resonance frequency can be pulled about 50 to 100 ppm by varying the external load capacitance. This is often utilized for fine tuning of crystal oscillators. Therefore crystal datasheet specify a nominal load capacitance along with the resonance frequency.