WIZnet Developer Forum

W5500 Crytsal specification

Are there any examples of crystals that work reliably with the W5500.
No reference schematic defines with which crystals it works.
Most of them use 18pF as load capacitors.
According to the formula CL = C1 / 2 - Cs the load capacitance of the crystal would have to be between 14pF and 16pF (assuming 5pF to 7pF stray capacitance).

On the other hand the Crystal Selection Guide Application Note for the W5100S (which is supposedly also valid for the W5500) states that a crystal with 14pF load capacitance and 7pF shunt capacitance (which most crystals seem to have) can have at most 25 Ohms ESR to satisfy the gain margin criteria for stable operation.

Neither DigiKey nor Mouser sell crystals that match these criteria, so I am somewhat at a loss at the moment.

Can somebody please list parts that would work? Preferable with matching load capacitor values?
Or did I misunderstand something?

Please give examples of your calculations for the crystals you think most appropriate from the list you have.

I found some crystals which I think would work. For example ECS-250-12-33-AGN-TR.
It has 12pF load capacitance, 5pF shunt capacitance and 40 Ohm ESR.
So the gain margin = 0.00843 / (4×40×(2𝜋×25×10^6)^2×(5×10^(−12)+12×10^(−12))^2) = 7.3887 > 6.9897 would be fine. It would require 10pF to 14pF load capacitors.
The ECS-250-10-36B-CTN-TR with 10pF load capacitance, 3pF shunt capacitance and 50 Ohm ESR has a gain margin of 10.108137019 which is also enough. It would require 6pF to 10pF load capacitors.

I was wondering which crystals WIZnet is using in their designs to require 18pF load capacitors, because they would have to have a very low ESR and/or shunt capacitance and I found none matching these criteria.

It would help a lot if someone could list crystals they have used successfully with the W5500.

Gain margin has relation to the oscillation startup. If condition is not satisfied, oscillation will not start, and device will not work.

It is only possible to know that crystal they use has 18 pF minimal load. There’s stray capacitance added (traces, pads, XI/XO input capacitance, internal crystal capacitance). The nearest realistic value is 20 pF for the crystal. Values of caps, logically, will depend on the PCB design, length of the traces, their width, and even on devices located near oscillator circuit. Using math is good and must be used as a guidance. a starting point, and then, if you want better results, you tune the values.

You must use parallel resonant crystals.

I did not see WIZnet publishing list of part numbers for crystals they use / recommend.