# How to design a pierce oscillator circuit?

I need to design a crystal resonator that can be used as a clock source for a uC. The first design is for an ENC28J60 IC, it's an ethernet - spi module and needs a 25MHz frequency.

I tried to follow a lot of tutorials and sites but I still can't wrap my head around it, I'm not sure if the design would work.

I picked a crystal that looked fine to me, the QCS25.0000F18B35. It's 25MHz, Cl is 18pF, Frequency tolerance is +- 30ppm and the stability is +-50ppm.

I used this application note from STM as a reference for calculations. Since the second project is to design a STM32F207 uC that needs another crystal too.

In the document it said $$Cl\:=\:\frac{C1\cdot C2}{C1+C2}+Cs$$ to calculate C1 and C2, since they can be the same value the formula would be $$Cl=2\cdot \left(C-Cs\right)$$ If you use 18pF for Cl and 3pF for Cs it gives you a C1 and C2 value of 30pF.

So far, is this a correct way of doing this?

Next is the drive level, the maximum power the crystal can dissipate is 1mW. If I use $$I_{Qmax}=\sqrt{\frac{DL_{max}}{ESR}}=\sqrt{\frac{1mW}{40}}=5mA$$ When I then calculate the current that will flow in the oscillator using $$I_{Q\:RMS}=2\cdot \pi \cdot f\cdot V_{RMS}\cdot C_{tot}=2\pi \cdot 25\cdot 10^6\cdot \frac{3.3}{2\sqrt{2}}\cdot \left(C+\frac{C_s}{2}\right)=5.79mA$$ Obviously this is more than the maximum current, so with this setup there should be an extra Rs inserted. But how to calculate that one... I have no idea. I've read that you could calculate them using the 'insides' of the XTAL, or to take a value that could work using a potentiometer and work from there to get a workable value. But I can't test the crystal in real life. It has to be possible to calculate it right?

When I tried other values for Cl, for example 12pF, using 3pF for the Cs gives you C1 = C2 = 18pF. This should give a current of 3.58mA, and that should work. Is this a workable value too? and is it possible to use my earlier values using Rs?

• Why does a pierce oscillator....oscillate? Why are high current risky? How to ensure the circuit always can reach the N*360 degree phase-shift point, exactly 360 degrees? – analogsystemsrf May 17 '17 at 14:40
• @analogsystemsrf From what I understand you have to power the quartz first to get it to 'oscillate mode', those signals go to the inverted amplifier in the uC and that creates the squared wave. High currents are risky because you could damage the quartz, so you should limit that under the maximum tolerances. My guess is the combination of the C1 and C2 values that control a 50/50 split and/or phase-shift? – pascal0312 May 17 '17 at 15:07
• @pascal0312 The choice comes from your crystal manufacturer. I've seen situations where Texas Instruments, for example, would list out a dozen different specific crystal part numbers and in the same table show different optimal load capacitances. There are drive limitations that Microchip will apply. But that's just a cross-check to make sure you are buying a crystal that your device can drive. The crystal manufacturer is the place to go. Do you have a datasheet for it? – jonk May 17 '17 at 16:53
• Your calculations address two-of-three major design issues (frequency of oscillation, and max power). You are missing gain calculations required to oscillate. Too much added Rs can decrease gain to the point where oscillations won't start. Many designers err on the side of having excess gain, because an oscillator that doesn't oscillate is like a human with no heartbeat. – glen_geek May 18 '17 at 12:52
• @jonk Certainly. I started off with this crystal. – pascal0312 May 19 '17 at 8:51