Go Down

Topic: Gibbs ringing on crystal resonator (Read 477 times) previous topic - next topic

Verdris

Hi all,

I'm using a 1MHz crystal in conjunction with my external ADC, and when I scoped the circuit today to make sure everything was connected properly, I noticed some Gibbs ringing on the crystal.  I tried killing with a capacitor between Vin and GND but that didn't help.  Is this ringing normal?  Will it affect the circuit?

RuggedCircuits

Never heard it called Gibbs ringing before but I assume you're observing overshoot and then some oscillation prior to stabilizing.

High-frequency oscillations are quite a common source of overshoot and ringing due to their fast edge rates. It is worse when the crystal is far from the ADC (and by "far" I mean anything >20mm or so) or the traces between crystal and part are not as short as possible.

A little bit of ringing is OK, but more than about 10% of steady-state is a potential source of concern, not in terms of damage but just in terms of a "misbehaving circuit".

--
The Flexible MIDI Shield: MIDI IN/OUT, stacking headers, your choice of I/O pins

retrolefty

  it's very difficult to accurately display the crystals signal's real wave shape with standard scope probes as their input capacitance effects the circuit being measured. Special (read expensive) FET input scope probes are one way to make sure the 16mhz signal you are looking at is indeed just the signal rather then an artifact of the signal due to scope probe impedance and capacitance.

Again if you are using standard scope probes I wouldn't put a lot of thought or concern about the wave shape you may be observing.

Lefty

MarkT

You were using a 10:1 probe and your oscilloscope goes up to at least 20MHz?  Otherwise it could easily be an artifact of measurement.

Ringing is expected wherever there are sharp edges - its not really the same as Gibbs phenomenon which is a mathematical property of Fourier series in a perfect mathematical universe (and is only 6%) - the ringing is usually due to resonances in the stray L and C or lack of appropriate termination.  If you have an ideal brick-wall low-pass filter it will exhibit Gibbs phenomenon, but that's hard to achieve in practice!
[ I won't respond to messages, use the forum please ]

Go Up