Increasing Reference Clock (Oscillator)??

[Tubbs]

I have an Acer V433S (socket 2) board. I would like to replace the reference clock (oscillator) presently 14.31818MHz, like all boards, with a reference clock (oscillator) of 14.7MHz or 16MHz. Will I still have a functional board? If not, why not? What other complications may arise?
I am not referring to the PLL (Phase Lock Loop) circuit which this board has along with the jumper settings (fsb) but the actual oscillator itself.

Any info would be greatly appreciated. Thank you for your time.

Tubbs.

[Paul V]

So, you want to replace the actual cystal itself?

If it is just an xtal, and not a whole xtal oscillator IC with the necessary filter components integrated into the IC, you need to change filter components too.

A crystal will vibrate in all of its harmonics. To select a particular harmonic, a tank circuit is used -- usually a parallel LC circuit -- whose frequency of oscillation (1/root(LC)) is very close to whatever frequency you need.

The LC pair basically acts as a passband filter, with a narrow band centered on the frequency of oscillation of the LC pair. This forms the frequency selection part of the tank circuit. There may be other components in the circuit as well, but the frequency filtration will almost certainly be this LC pair (it really is the best way to do this).

So, if you up the crystal, you will need to change the frequency selection in your filter (by changing the inductor or capacitor) to center the passband as close to the harmonic of your choice as you can.

[chipbgt]

ok guys my brain just exploded....


Just when you think you have computers figgerd out.....whew!

What the heck are you guys talking about? I would definitly like to know what all this is.

[Paul V]

Oh, first off, there is a mistake in what I said. Omega, the angular frequency, is 1/root(LC). f is 2*pi/root(LC).


Well, what we're talking about in essence is the basis of a clock generator in a computer.

First, a quartz crystal is used. Advantage: this gives a VERY precise oscillation. Wheras other electrical oscillation circuits exist, they are much less precise. The disadvantage is that the crystal vibrates in all of its harmonics (i.e. at its fundamental frequency, and at every positive integer multiple of this frequency). Now, a clock composed of multiple frequencies is useless, so we need to select one by some kind of filter.

One filter we use is also an oscillator in itself, albeit not as good of one, it is the LC circuit. An inductor and capacitor are put in parallel -- this means, if no current enters or leaves the system, that iC, the current through the capacitor, and iL, the current in the inductor are related:

iC = -iL

And voltages are equal:

vC = vL.

by the fundamental equations for inductor and capacitor:

vL = L*diL/dt
iC = C*dvC/dt

Thus:

L*diL/dt = vC
C*dvC/dt = -iL

L*d^2iL/dt^2 = dvC/dt by differentiating the first equation


L*C*d^2iL/dt^2 = -iL by substitution.

So we have a second order differential equation. One solution is a sine or cosine wave with its angular frequency 1/root(LC). This is what happens in an LC oscillator.

Now, it can't be used in itself, because as I said about many electrical oscillators, it is hard to control the frequency with any precision. It CAN however be used to filter one of the harmonics of the crystal.

Think of the LC system as a pendulum like a swing. Now, if you are going to push someone on the swing, you need to push it at certain times; if you push at random times and places the swing won't go anywhere. Similarly, if you apply an oscillating signal to the LC circuit, and it is not close to the frequency at which the LC circuit will oscillate, nothing will happen, you're pushing the swing at all the wrong times.

The swing analogy is pretty good too -- in a swing you are basically oscillating between high potential energy (at the tops of the peaks) and high kinetic energy (at the bottoms). Here, you oscillate between energy stored in the magnetic field of an inductor, and energy stored in the electric field of the capacitor.

So, what we use is a quartz crystal to generate a multiple frequency signal, then use the LC circuit to filter all but one of them. By Fourier's theorem, and by the superposition theorem, both of which apply to electrical systems such as this, we can consider any signal, however complex, as a sum of infinitely many sine waves added up. Then, we can see what happens at each frequency, and re-sum all the sine waves to get the output signal.

Here this is easy -- we have only sine waves of the harmonics of the crystal to consider. Most of these will be far from the frequency of oscillation of the LC circuit, so they well be attenuated (reduced) to nothing. Only one, the one that is very close to the frequency of oscillation, will be passed.

So what we get after this filter is one single sine wave whose frequency is a harmonic of the crystal's (and thus is very precise).

This can easily be made into the step waveform a clock needs.

Whew, that's all I can think of to say on this topic for now =]

[Paul V]

To actually answer the original question fully --

What kind of oscillator does it have? I.e. is it just an IC, or is it an IC + quartz crystal + other?

If it is purely an IC, you can probably just replace it (I haven't done this nor researched this so there may be side effects I haven't forseen). The oscillation circuitry could also be a crystal driven oscillator, i.e. if there is a separate crystal on the board -- they usually look like thin metallic boxes, but with rounded edges (i.e. the cross section is an oval) and will have a frequency stamped upon them, usually in MHz. They would have 2 leads going to them.

If the cyrstal is separate, then the clock is crystal driven with other components used to filter the signal; you need to alter these components and the crystal both to get a clean and stable result. Failure to do so will result in something much like what happened to me once when a parts shop sold me a mislabeled inductor -- the clock signal I tried to generate jumped all over the place, in and out of different harmonics, sometimes there was a decently clean signal and other times none...

As you don't want that, please post more about the hardware used to generate the clock signal. Without the details of the problem it's hard to help you.

[This message has been edited by Paul V (edited 05-01-2000).]

[Tubbs]

I would like to add that the Reference Clock (Oscillator) on my board (Acer V433S) is a shiny, rectangular box. It's sides are flat as is the top and underside. The corners are rounded. It has four pins. Looking at the underside of the board, one pin is wired to the PLL (Phase Locked Loop) circuit. The other three pins are not wired. Looking at the board itself from above (top side), I can't tell if the four pins are wired to anything, it's to close to the surface of the board.

The PLL circuit seems to be wired to many other components.

Printed on Oscillator:
T 9330 The "T" appears to be a company logo.
XO - 105BIC The "O" in XO could be a zero
14.31818MHz

I hope this can help with replying to my question "Increasing Reference Clock (Oscillator)??.

Thank you for the time and effort in replying to my posting.

Tubbs.

[socalgal]

Tubbs

Please do not multiple post. This generates scattered replies, makes troubleshooting difficult, and makes comparison and references difficult in this case.


http://sysopt.earthweb.com/forum/Forum3/HTML/006220.html