Choosing the value for the xtal capacitors

[Simon]

Hi, I need to select the values for the crystal capacitors (C1 & C2) for a nRF51422 pcb.

From page 9 of "Crystal Oscillator Design Considerations" I see:

C1' = C1 + C_xc1 + C_pcb

C1' = 2CL

therefore:

C1 = 2CL - (C_xc1 + C_pcb)

Where:

CL is the load capacitance from the xtal datasheet, C_xc1 is an unknown with a typical value of 4pF from Table 19 of the 51422 Product Spec v2.0 C_pcb is an unknown (the reference design uses .5pF and my design is reasonably close to it)

So it seems like my initial guess for C1 & C2 is = 2CL - (4pF + .5pF)

Here are my questions:

1. Is C1 = 2CL - (4pF + .5pF) for an initial guess reasonable? I would like to get as close as possible for my first guess and don't know if there is a better approach.

I have seen three values of C_xc1 to choose from: 4, 1, and 3.5 pF.

The Product Spec states 4pF as typical and 1pF as a test value.

I got 3.5pF from calculating its value from reference design as follows: C_pcb = .5pF (pg 61 of Prd Spec 2.0), CL = 8pF and C1 = 12pf (pg 63). C1 = 2CL - (C_xc1 + C_pcb) 12 = 2*8 - (C_xc1 + .5) C_xc1 = 3.5pF

2. Given that C_pcb and even C_xc1 are unknown I will likely have to try many load capacitors until I find the right values. I will have to add and remove these surface mounted caps by hand.

In the "Crystal Oscillator Design Considerations" this is written: "Too low capacitive load will result in a crystal oscillator frequency higher than specified, or too high a load results in a lower oscillation frequency."

Knowing this I will be able to tell if the capacitance is too high or too low. Any suggestions on how I can quickly narrow down to the correct value?

Thanks!

[Håkon Alseth]

Hi,

It seems that you have done some research and found the right formulas for calculating the caps for the XTAL. You have got everything correct, but to clarify: "C_xc1" is the pin input capacitance, which is 4 pF nom. for the nRF51x22 part.

C_pcb can be up to 2 pF, depending on your design and track-length. Usually it's in the 1 pF area. What I have done to verify that the caps are set correctly is to probe the XC1 pin to verify frequency and amplitude. What you need to keep in mind when doing such a measurement is to adjust for the load of the oscilloscope probe. The load capacitance of these probes are always shown on the probe.

Let's say that you have calculated C1=C2 = 12 pF. Your probe loads with 8 pF, then the C1 cap should be 4.0 pF when doing this measurement. The frequency should be 16.000 MHz ( + jitter of max 40 ppm if Bluetooth application, 60 ppm if proprietary)

The amplitude (Vpeak) should be within 0.2V to 0.4V, depending on the crystal.

Best regards Håkon