Questions about metal can oscillators.

[MarkS]

I'm using this for a project: https://www.digikey.com/en/products/detail/ecs-inc/ECS-2100AX-100/827269


What I want to know, and the datasheet is unclear on, is whether I need any external components. This will be the clock input to a MC68000 CPU. Is it still good practice to add caps? Do I need any sort of buffering or waveform shaping circuitry? It will be here tomorrow from Digi-Key, so I can do some tests with my oscilloscope, but all I have are solderless breadboards to test and that is fraught with issues.


Are there any best practices, or do I just simply slap on power and ground and run the output to the CPU?

[jonpaul]

logic lvl out XO. see spec sheet
MCU..see spec sheet.

both MCU, XO must use same VC 3.3 or 5V, and ground.

MCU may have option pin for bare xtal or XO

j

[MarkS]

logic lvl out XO. see spec sheet
MCU..see spec sheet.

both MCU, XO must use same VC 3.3 or 5V, and ground.

MCU may have option pin for bare xtal or XO

j

It's TTL logic and the oscillator provides the correct output. That's not what I'm asking, however.

[Dave]

It doesn't hurt to add some capacitance between the power pins.
The datasheet states you can load the output with up to 10 TTL loads. Add buffering only if you intend to clock more than 10 ICs with it.

[barshatriplee]

You can post this to digikey forum too. Digikey employees may explain for accurately.

[BillyO]

As Dave suggested, all you need is some supply bypass capacitance as close to the power pins as possible.  If you have room, use two.  A .1uF and a .01uF.  However, you do not need any capacitance on the output signal.

[radiolistener]

Is it still good practice to add caps?

if you mean caps on a clock line - no, this oscillator already have internal circuit and output buffer. But adding caps on a power supply lines is a good idea.

I'm not sure about exactly this square shape DIP oscillator, but I bought very similar square shape DIP oscillator from aliexpress and it has very bad performance - very unstable and drifting frequency. Rectangle shape DIP oscillators from China with ROJON marking is much better. They have PLL synthesizer inside, but they are pretty stable and good noise performance (not ultra low noise, but pretty good for synth-based oscillators).

[srb1954]

I'm using this for a project: https://www.digikey.com/en/products/detail/ecs-inc/ECS-2100AX-100/827269


What I want to know, and the datasheet is unclear on, is whether I need any external components. This will be the clock input to a MC68000 CPU. Is it still good practice to add caps? Do I need any sort of buffering or waveform shaping circuitry? It will be here tomorrow from Digi-Key, so I can do some tests with my oscilloscope, but all I have are solderless breadboards to test and that is fraught with issues.


Are there any best practices, or do I just simply slap on power and ground and run the output to the CPU?

Unless otherwise specified by the xtal oscillator manufacturer it is always best to assume that external power supply bypassing caps are required. It doesn't need to be a very large cap but you do need to minimise the track inductance from the oscillator power pins to the bypass cap to reduce the supply ripple seen by the oscillator circuitry. Connecting the oscillator power pins to power and ground planes in a multi-layer board can help considerably in reducing that inductance.

With each clock transition there will be a spike in the supply or ground current as the oscillator O/P charges and discharges the load capacitance. With the capacitance of the 68K clock input and the specified oscillator rise and fall times those current spikes are going to to be of the order of 10-20mA peak; the bypass capacitor needs to supply those current peaks locally assuming the bulk supply can't deliver that current instantaneously. Taking those current peaks into consideration, and allowing a reasonable amount of supply ripple, you might find that only a few nF is sufficient for a bypass capacitor. It doesn't hurt to add more as long as the circuit inductance is kept as low as possible.