RM48L952: Crystal circuit recommendations?

Hi,

Please refer to 6.6.1.1 of device datasheet:

The correct CLOAD is important for proper operating frequency. Crystals are available with a variety of CLOAD values.

CLOAD = (C1 * C2 ) / (C1 + C2 ) + CS

CS is the stray capacitance in the oscillator circuit. Stray capacitance is a function of trace lengths, PCB construction, and microcontroller pin design. For a typical design, CS should be approximately 2pF to 4pF. Because C1 and C2 are normally of equal value, the calculation for a typical circuit simplifies slightly to: C1 and C2 = (CLOAD – 3pF) * 2

NOTE: TI strongly encourages each customer to submit samples of the device to the resonator/crystal vendors for validation. The vendors are equipped to determine which load capacitors will best tune their resonator/crystal to the microcontroller device for optimum start-up and operation over temperature and voltage extremes

Hi,

Working with Anders on this topic, he is now on vacation.

As you suggest, we are following the capacitor recommendations of chapter 6.6.1.1.

We are unsure about crystal series resistor.

Without the series resistor, the oscillating voltage reaches levels outside specification for the device. We are also having a higher than typical power dissipation in the crystal.

To get specified typical power dissipation in the crystal, we had to increase the series resistance to 7.5kOhm.

What are the recommendations for a crystal series resistor?

Hi,

Thank you.

We want to operate at max operating frequency and reduce the dissipated power in the crystal.

Perhaps there are no recommendations for crystal series resistor except for "a large external series resistor could cause the oscillator to fail to start"?
We need to know where is the limit, what is too large for the external series resistor, 1k? 10k? 100k?

Hi,

Did you find any specific limits for this series resistance?
We need to get a specific limit or a way to calculate it.
Meanwhile, we will start testing different values in the lab, but without knowing what different failure mechanisms there are, it will be hard to know if the test method is realistic or if we can trust our results.

Is the failure with a big resistor due to too low voltage on the MCU comparator at the OSCIN pin? Anything else? Will this vary over time? Temperature? Other factors?

Best regards,
August Hultman

Hello,

That's unfortunate that we can't find any information about the external crystal interface. The series resistor is not needed for the TI device, but we want to limit the power level of the crystal, therefore it is needed.

I was testing the crystal circuit with 0, 7.5k, 18k and 33k series resistor in room temperature.

18pF CL, 16MHz crystal rated for 100µW typical drive level.

Using the ECLK output, I found that the duty cycled varied with different resistors (expected):
0 Ohm: approx 45% duty cycle
7.5k: approx 38% duty cycle
18k: approx 9% duty cycle
33k: not starting (0% duty cycle)

I will use this as an indication of minimum amplitude margin for the input. It seems like the 7.5k is not affecting the crystal circuit by any big amount.
If possible I will also test this with different temperatures.

We might send the board for validation of load capacitance at our crystal manufacturer, but they won't be able to tell us what the TI device limits are for amplitude etc.

If anyone finds any more information in this topic, please reply and inform us.

For me, this topic is not resolved yet, but I cannot deny the suggested answers in this thread as I am not the thread starter.

Hi August,

You said the power dissipation of the crystal is 7x higher than the typical value. The typical crystal power consumption is 100um. I understand that the power dissipated in the crystal has to be limited, otherwise the crystal can fail. The maximum power consumption of the crystal ( ABLS-16.000MHZ-K4T) is 1mw. 

How do you calculate the power dissipated by the crystal? 

I don't have a current probe to measure the current. I measured the voltage level at OSCIN: Vpeak-2-peak = 3V (RM46Lx Launchpad), and the voltage across the crystal is around 46mV (AC, RMS). 

The estimated dissipated power: 46*46/ESR = 53uW, the impedance of the capacitors is not included. 

Hi,

Thank you, I think you are leading us in the right way.

We have been using this formula for power calculation:

I was digging in our documentation and found that we probably used the wrong voltage in the formula.

I will redo the measurements from the beginning next week and hopefully this topic will be resolved.

Sure, let me know your new measurement and power dissipation calculation.