2

I disassembled a usb wifi adapter. I was surprised to see a 40MHz oscillator, when wifi works at 2.4GHz. How the output of the oscillator is scaled to the required frequency?

706Astor
  • 1,054
  • 1
  • 8
  • 19
akm
  • 255
  • 3
  • 8

1 Answers1

2

Generally when you need to scale up a clock you use a PLL (Phase lock loop).

A very crude explanation is that you have a voltage controlled oscillator at roughly the required frequency. You divide the output of this down to your reference frequency (a factor of 60 in this case) and then compare it to your reference clock input. The difference between the two is then used to adjust the frequency of the voltage controlled oscillator.

The system will take a short time period to lock on and stabilize but after that you end up with a clock which is an exact multiple of the input.

This is glossing over a lot of the details but google for "PLL theory of operation" or similar and you'll find more details than you ever wanted to know.

Andrew
  • 766
  • 3
  • 9
  • But why a 2.4 GHz oscillator not used ? @Andrew – akm Dec 19 '16 at 10:33
  • Partly because running a PCB trace at 40MHz is a lot easier than running one at 2.4GHz, partly because a trace at that speed would radiate EMI and potentially jam your own receiver but mainly because you can't get 2.4GHz pieces of quartz. Crystals are mechanical devices, they are a small lump of quartz that has been cut to resonate at the required frequency. Physical constraints on how you can cut them mean that they are easiest to make and most reliable between around 1 to 50 MHz, most systems will use a clock input in that range and step it up or down from there as needed.. – Andrew Dec 19 '16 at 10:51
  • Wouldn't the trace will radiate and jam after the step up? @Andrew – akm Dec 19 '16 at 11:46
  • That will be inside the chip and so a very short distance at low power. – Andrew Dec 19 '16 at 11:52
  • Then all the devices even those running at few MHz, if there are, should also use step up procedure, to avoid interference. is this the case? @Andrew – akm Dec 19 '16 at 13:21
  • 1
    It depends. If the trace is short relative to the wavelength then you have less of an issue. 2.4 GHz in FR4 (the standard PCB material) has a wavelength of around 3 cm. 100 MHz has a wavelength of 75 cm. Keeping traces under a small fraction of 3 cm is tricky. Keeping them under a fraction of 75 cm is a lot easier. But you are missing the main point: You can't easily get oscillators outside a certain frequency range. – Andrew Dec 19 '16 at 14:15
  • But this page https://en.wikipedia.org/wiki/Electronic_oscillator#Negative_resistance_oscillator lists various oscillators which work in GHz range. Why these are not used? – akm Dec 19 '16 at 14:27
  • Because they cost a lot more. It's not about what's possible. It's what you can get a reliable supply of cheaply. Electronics design isn't about what's possible, it's about what's most cost effective. I can get a thousand crystal s in the 40MHz region delivered tomorrow for a few cents each. – Andrew Dec 19 '16 at 16:54