Crystal Emission: Understanding Fundamental and Harmonic Frequencies"

[likephysics]

Say I have a 25MHz Crystal that's connected to a Microprocessor/Microcontroller or spread spectrum clock generator. Does the crystal emit Electromagnetic radiation at its fundamental frequency or it's harmonic frequencies?

I keep hearing - " I see a peak at ~50MHz, it must be the 2nd harmonic of the crystal". I just find that hard to believe. Crystal has a sine wave output, so no harmonics unlike a square wave.
Any EMC experts care to add your 2 cents.

 

[berkeman]

Say I have a 25MHz Crystal that's connected to a Microprocessor/Microcontroller or spread spectrum clock generator. Does the crystal emit Electromagnetic radiation at its fundamental frequency or it's harmonic frequencies?

I keep hearing - " I see a peak at ~50MHz, it must be the 2nd harmonic of the crystal". I just find that hard to believe. Crystal has a sine wave output, so no harmonics unlike a square wave.
Any EMC experts care to add your 2 cents.

2nd harmonic components would generally be from duty cycle issues. How far down from the fundamental is the 2nd harmonic component? How are you making these measurements?

It's common for fundamental mode oscillators to overtone in the 3rd harmonic on startup, but after startup, there generally is not much harmonic content (at the input side of the oscillator gate -- the waveform gets squared up a bit at the output of the gate, and hence you will get some odd harmonic content there).

 

[mdjensen22]

Also, could it be emissions from the device interfaced to the crystal? I believe the processor or clock generator would be converting the sine wave to a square wave at some point. It is not uncommon for processors to give off emissions at some harmonic of their clock frequency.

 

[Studiot]

Crystals do not emit electromagnetic waves.

Their oscillations are a mechanical phenomenon.

The mechanical oscillation between the faces alters the capacitance of the crystal. This change in capacitance can be used within an electrical circuit to turn the mechanical oscillation into an electrical one.

The electrical oscillation can then be broadcast as an electromagnetic wave via a suitable antenna.

 

[waht]

Say I have a 25MHz Crystal that's connected to a Microprocessor/Microcontroller or spread spectrum clock generator. Does the crystal emit Electromagnetic radiation at its fundamental frequency or it's harmonic frequencies?

An xtal sitting alone all by itself will not oscillate. In order to use an xtal as an oscillator you need an active circuit with a proper positive feedback path. Both of these elements will determine how the xtal will oscillate.

You can have a case of pure output sine wave oscillator up to a certain point. Beyond that point there will always be non-ideal stuff present due to non-ideal components and non-linearly that creates those harmonics.

You can have a case of a square wave oscillator, which ideally should be a sum of odd order sine harmonics, but because of the non-linearities present in the circuit, even order harmonics could be present as well. But these will not make a square wave into its second harmonic unless you design the oscillator really bad. That eventually leads to a comb generator, where the xtal is used in such as way as to generate nth order harmonics up to like 500 or more.

 

[likephysics]

Also, could it be emissions from the device interfaced to the crystal? I believe the processor or clock generator would be converting the sine wave to a square wave at some point. It is not uncommon for processors to give off emissions at some harmonic of their clock frequency.

It could be, but not at 50MHz. The processor or clock gen output is square wave output. So Only odd harmonics, whereas 50MHz is even harmonic of base freq.

 

[likephysics]

2nd harmonic components would generally be from duty cycle issues. How far down from the fundamental is the 2nd harmonic component? How are you making these measurements?

Measured in an Anachoic chamber.

 

[likephysics]

One more EMC question - what should the min length of the trace be to act as an Antenna?
Lambda/4?
Does it have to be lambda/4 at least or can it be lesser?