1
\$\begingroup\$

I am designing a board but space is not a luxury that I have at the moment, can I get away with that design or I must use a SMD resonator?

enter image description here

\$\endgroup\$
  • 1
    \$\begingroup\$ The frequency might be interesting for some people \$\endgroup\$ – PlasmaHH Aug 31 '18 at 5:59
  • 3
    \$\begingroup\$ I think this will work. But do remember that Xtals with this footprint are often quite high. If you first solder the chip and caps to the PCB and the Xtal later there should be no issue. You will have an issue if you need to probe the pins of the chip. For mass-production I would go with an SMD type Xtal. Any reason why you need this large type? \$\endgroup\$ – Bimpelrekkie Aug 31 '18 at 6:02
  • 6
    \$\begingroup\$ Hey, @LucasAlexandre, by convention, MHz is capitalized just like that. It is really not correct to write "mhz." Just FYI. I think the layout will work. Especially if you don't need to pass radiated emissions. \$\endgroup\$ – mkeith Aug 31 '18 at 6:35
  • 1
    \$\begingroup\$ also, "space is a luxury" means you need to save on PCB space. But "I want to save on SMD oscillators"… well, honestly, one would be a cheap, simple solution, the other a cludge to use up your stock of through-hole components on a board that's otherwise mainly SMD... \$\endgroup\$ – Marcus Müller Aug 31 '18 at 7:19
  • 1
    \$\begingroup\$ Oh, and if you can, don't break the ground plane by running the reset signal there \$\endgroup\$ – Marcus Müller Aug 31 '18 at 7:23
2
\$\begingroup\$

You could "Get away" with a design like this. A good clock design keeps the traces as short as possible, and the capacitors and traces symmetrical to keep the parasitics equal. A copper pour can also be a good idea on the top layer around the components to increase capacitance to ground, and shunt high frequency signals to ground.

The biggest problem with crystals is radiated emissions, and depending on the country your in this could be a problem. It's best to use good clock design practices to avoid problems.

One thing you might be able to do is mount the crystal on the other side of the board to decrease spacing.

\$\endgroup\$
2
\$\begingroup\$

Load caps should be between the MCU and resonator, otherwise they act as an antenna and might pick up extra noise, which, depending on the operational conditions, might mess with your timings.

I would suggest leaving resonator "horizontally" (space-wise seems to fit) and move load caps to the side of the MCU, between resonator and MCU itself.

In theory some older PTH resonators could be more thermally stable than SMD due to larger die, albeit suffer from some frequency deviations from nominal due to manufacturing tolerances.

\$\endgroup\$
  • \$\begingroup\$ You have the concern backwards, it is not picking up noise but radiating an unintended signal likely with harmonics too. \$\endgroup\$ – Chris Stratton Aug 31 '18 at 13:26
  • \$\begingroup\$ @ChrisStratton antennas work both ways. As I mentioned, it depends on operational conditions. If you have some strong radiators nearby, you might pick up. Usually oscillator drive levels are so low, that they can be considered as a negligible radiationwise. \$\endgroup\$ – stiebrs Sep 3 '18 at 8:43
1
\$\begingroup\$

It depends on what you mean by getting away with it. There are a few considerations:

1) Your Xin and Xout traces are of different lengths. It looks like the difference is small enough not to be an issue, but without knowing more about your application, I don't know.

2) If it's an HC49-U, it will really loom over your chip and make it hard to probe. If HC49-S, it won't get in the way quite so much.

3) If you decide to switch to SMD, keep in mind that they have very different -R and pullability than HC49 (they will be worse than HC49-S, which in turn is worse than HC49-S).

\$\endgroup\$

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.