4
\$\begingroup\$

I have followed some reference designs that uses crystals like 40 MHz for the MCU, and I have not considered the trace width to do impedance matching. The result is that the boards works fine (I kept the track as short as possible). I have not seen any note anywhere that one should consider trace width to have an impedance matching (or at least get closer to it) as above 3 MHz is considered high speed signal. I wonder why one does not consider the wave reflections on the crystal's traces? I don't see any note about it on crystals' datasheet neither.

Even if the reflected waves does not change the frequency of the oscillation, it should cause more noise in the board am I wrong? Even if one would not need to use extra L-C matching circuit for it, but at least wouldn't it be better to use a little bit thicker or thinner line width to improve the reflection depending on the crystal and MCU?

\$\endgroup\$
  • \$\begingroup\$ For simple clock ususlly there is not a problem since pulse have a trapezoid shape and consiquently slow rise time (i.e harmonics it is not an issue). But you have to taking into account when stipe lines in PCB used to distribute a master clock around. In this case uniformity of signal speed and propagation delay on trace should be considered. \$\endgroup\$ – GR Tech Jan 14 '15 at 15:15
9
\$\begingroup\$

A 100MHz oscillation has a wavelength of 3 metres and, a general rule of thumb is that you don't need to consider track length impedance matching if the trace length is less that one-tenth of the wavelength.

So, if you are using a 100MHz oscillator and it is sited at least 0.3 metres from the chip it feeds, then you ought to consider doing something; either shortening the track or impedance matching.

At 100MHz the output will probably be more sinusoidal than square so there shouldn't be a problem AND if you try to consider the sq wave harmonics of a much lower frequency oscillator you'll still conclude that things only start to become a problem when you have ridiculously long tracks and frequencies above 100MHz (half theory half rule of thumb).

\$\endgroup\$
  • \$\begingroup\$ The wavelength of a 100 MHz signal on PCB is not 3 m, it is only about 2 m. \$\endgroup\$ – Uwe Jun 13 '16 at 15:02
  • \$\begingroup\$ @Uwe thanks for your clarification (and the other one!). \$\endgroup\$ – Andy aka Jun 13 '16 at 16:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.