TCAN4550-Q1: About crystal selecting

Hi teritama,

Yes, please include the series dampening resistor, even if it is set to a 0-ohm as an initial value on the first prototype build.  This will allow for the circuit to be analyzed and optimized for performance and stability which may require a different value resistor after this process is complete. 

The issue is that there are two perspectives when looking at this circuit and doing the optimization which must be balanced.  The first perspective is from the crystal's needs which specifies a particular capacitive load and drive level to oscillate at a particular frequency.  The second perspective if from the TCAN4550 and the overall load created by the crystal, capacitors, resistors, etc. located external to the device and connected between the OSC1 and OSC2 pins.

The crystal specifies a total load capacitance of 8pF, which includes the pcb parasitic and TCAN4550 pin capacitance as part of that capacitance.  Once these parasitic capacitances are included, the actual value of the load capacitors needed to create an overall capacitance of 8pF will be less and possibly in the 2pF to 4pF range (each).  This is application specific and will vary by the pcb layout.

The TCAN4550 is expecting an external load of 8pF to 24pF worth of capacitance and an ESR of up to 60 ohms.  The crystals specify a max ESR and the actual measured ESR on any given crystal is usually less than half of this spec from my experience.  Also, the TCAN4550 is not including it's own pin capacitance in the expected needs for the overall capacitive load expectations which is a significant portion of the total capacitive load.  Therefore, when the capacitance is optimized for the crystal using load capacitors that are in the 2pF to 4pF range, this is below the expected value and it can cause the automatic gain control circuit to rail out at the minimum edge of the range and prevent it from reducing the output current to a lower level needed to reduce the crystal oscillations.  This will result in an oscillation voltage that is too large and this can lead to a stability issue with the single-ended detection circuit thinking the OSC2 pin has been "grounded".

The other issue is this parasitic capacitance is not stable with temperature because it is not made up of the same type of dielectric material used in the actual capacitors.  Therefore, when the temperature shifts, the capacitance will shift, and without optimization, the circuit could in theory drift into an unstable condition.  Most optimization is done at room temperature, and then when the board is heated, the board may not be stable.

Adding the series dampening resistor will reduce the current and drive level to a low enough level that small changes in load capacitance due to component tolerance and temperature won't cause a stability issue.

As I've outlined in the Optimization App Note, the recommendation is to start with a 0-ohm resistor and then make some tests and optimize the circuit by changing the capacitor values and then increasing the resistor value as necessary.  Typical values are less than 100 ohms, with some as low as 10 ohms when this is needed.  Some applications are stable without a series resistor, but if the resistor is omitted from the design, increasing the load capacitors is the other other solution to creating stability and this method will shift the oscillation frequency slightly.

Regards,

Jonathan