Author Topic: 8MHz trace - best practices  (Read 992 times)

0 Members and 1 Guest are viewing this topic.

Offline i_am_fubarTopic starter

  • Regular Contributor
  • *
  • Posts: 53
  • Country: gb
8MHz trace - best practices
« on: February 17, 2020, 07:52:08 am »
I have a wireless video transmitter that needs an 8MHz clock input. To save board space, my plan is to generate it from a PIC and output it via a CCP pin.

The trace is 1.6" long and running on a double sided 1.6 FR4 board. Knowing that this is going into RF territory, what considerations should be made?

Cheers
 

Online fourfathom

  • Super Contributor
  • ***
  • Posts: 2005
  • Country: us
Re: 8MHz trace - best practices
« Reply #1 on: February 17, 2020, 08:12:17 am »
1.6" is pretty short, and the PIC isn't going to have blazingly fast signal edges, so you probably don't need to do anything.  You might put a series resistor (say 100 Ohms) at the output of the PIC to reduce potential ringing on the 8M clock.  How accurate does this clock need to be?
We'll search out every place a sick, twisted, solitary misfit might run to! -- I'll start with Radio Shack.
 

Offline i_am_fubarTopic starter

  • Regular Contributor
  • *
  • Posts: 53
  • Country: gb
Re: 8MHz trace - best practices
« Reply #2 on: February 17, 2020, 08:31:40 am »
Cheers, Looked at the wavelength of 8MHz, and it's well outside of dimensions of trace :)

Good question on accuracy.... I don't know. Transmitter datasheet is a bit poor. Can always tune the PIC if needed.
 

Offline Siwastaja

  • Super Contributor
  • ***
  • Posts: 9336
  • Country: fi
Re: 8MHz trace - best practices
« Reply #3 on: February 17, 2020, 11:32:18 am »
8 MHz frequency is irrelevant; it's a square wave, so the edge rate defines the actual frequency content.

Adding a tiny series resistor (such as a 0402 package) physically as close as the driving pin as possible indeed slows down the edge (and works as series termination).

Remember to keep the ground plane contiguous below the trace and make sure there are ground vias close to the ground pins of the ICs at both ends, so that the current has a return path as close as the trace as possible. Not super important at such edge rates, but good to have some practice. The 2-layer 1.6mm board adds some extra distance to the return current you can't completely deal with, but it's ok.
 

Online Zero999

  • Super Contributor
  • ***
  • Posts: 20361
  • Country: gb
  • 0999
Re: 8MHz trace - best practices
« Reply #4 on: February 17, 2020, 01:18:07 pm »
 

Online fourfathom

  • Super Contributor
  • ***
  • Posts: 2005
  • Country: us
Re: 8MHz trace - best practices
« Reply #5 on: February 17, 2020, 04:34:58 pm »
Adding a tiny series resistor (such as a 0402 package) physically as close as the driving pin as possible indeed slows down the edge (and works as series termination).

This is indeed "best practice" but in this case the 402 package is isn't required.  You could use anything from a 402 up to a 1/4W leaded resistor and it's going to work with these signals.  But no resistor at all is also likely to work just fine.

And yes, the little xtal oscillators that Zero999 mentioned are inexpensive and accurate.  If you aren't doing surface-mount, similar oscillators are available in a package with pins.  But if you've already got the PIC and can get that clock output for free, then go for it.
We'll search out every place a sick, twisted, solitary misfit might run to! -- I'll start with Radio Shack.
 

Offline i_am_fubarTopic starter

  • Regular Contributor
  • *
  • Posts: 53
  • Country: gb
Re: 8MHz trace - best practices
« Reply #6 on: February 17, 2020, 07:25:03 pm »
The board was one I've been putting off. But I had spare real estate on a Chinese 100x100 fixed price order I'm making. So figured I would tag it into the order for free.

Does, however, mean I have literally no space for a crystal unless o re-track everyhting. It's not super high density, but hand SMT friendly for the most part.

PIC is already there, so it's another freebie. Accepting my round off, will add at least space for a 0603. Already got solid unbroken ground pour around it. A few SPI lines in parallel, but will be running those quite slowly.

Ta
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf