Of course, that depends a lot on the application.

I assume that these crystal oscillators aren't "silent" during the startup phase but often perform a frequency sweep.

For example, if you have a microcontroller with an internal (calibrated) RC oscillator, you can run it on the RC clock for a while and then switch to the external clock once it's stable.

It's more complicated with microcontrollers (like many AVRs) where you have to define the clock source in fuses during programming and can't later switch from the RC oscillator to an external clock via a register.

  Mixtel90 said  

We used to temporarily stop the Z80 clock to let DMA happen. The Z80 knows nothing about it and just carries on when the clock is restored.

In fact, that was one of the great things about the Z80.   I wire wrapped up a CP/M Z80 machine.    And I was able to debug it with a logic probe.   I made the clock circuit feeding it the 4 MHz able to feed it clock pulses via a (debounced) toggle switch.

I could toggle the switch back and forth and use the logic probe to see exactly what I/O ports or memory addresses were being accessed.  

It was really nice of the Zilog guys to make the Z80 able to work and hold state at 0 Hz.

  EDNEDN said  

  Mixtel90 said  

We used to temporarily stop the Z80 clock to let DMA happen. The Z80 knows nothing about it and just carries on when the clock is restored.

In fact, that was one of the great things about the Z80.   I wire wrapped up a CP/M Z80 machine.    And I was able to debug it with a logic probe.   I made the clock circuit feeding it the 4 MHz able to feed it clock pulses via a (debounced) toggle switch.

I could toggle the switch back and forth and use the logic probe to see exactly what I/O ports or memory addresses were being accessed.  

It was really nice of the Zilog guys to make the Z80 able to work and hold state at 0 Hz.

I did the same thing with the terrific Motorola 68000 (with 1MB flat memory capability)--also wire-wrapped. You could single-step at any rate you liked.

Oh, the times (but wouldn't want to go back).

Hi,
Weighing up all your advice, and some ploughing through the literature, I have decided that I woun't be able to determine good conditions for a crystal, so will use an oscillator module instead.

Probably a ECS-3951M-200-BN-TR, 5V supply, 50ppm, 7mm x 5mm package.

For all my past projects I have used an HC49U cased crystal alone without problems, with drive requirements up to the 1.2mW level. Now I wonder if even they were overstressed!

With best regards, Paul.

Hi,
On the pcb layout for the above oscillator module, is it worth putting a guard ring on the same side as the module?

There is a ground plane on the opposite side. The literature I've seen doesn't give any such details for a module like this, just pad layouts.

With best regards, Paul.

Providing the signal track is kept sensibly short and direct then no, you almost certainly won't need a guard ring, especially with a ground plane. The only thing to think about is the capacitance along the track rounding the corners of the clock. Unless the frequency is very high this usually isn't much of a problem. The signal level is already high and has a far better SNR ratio than you have on a XTAL oscillator amplifier.

Hi Mixtel90,
Thanks. I just checked the track length - osc. o/p to micro clock pin - 5.1mm, so there should be no problems regarding distance.
With best regards, Paul.

I've run hdmi over much longer lengths than that. :)