question about oscillators

[bigendian]

Hi all,



what is the difference between such a 4 pin crystal (where just 2 are used) and a simple 2 pin crystal ?
I copied the design of the oscillator from the above picture, but using a simple 2 pin crystal. Oscillator does not work, by oscilloscope i don't see any signal. Is it possible i used an improper crystal ?

[ataradov]

They are functionally the same (assuming everything else is matched). But there is more to the oscillator design than just the form factor. You need to properly match characteristics and load capacitors.

Also, oscilloscope may put too much load and some oscillator circuits may not be able to handle that. You may need an active probe for that.

Note that the same 4 pin package is also used by signal generators, which are entirely different in operation. But they also use more than two pins.

[PCB.Wiz]

what is the difference between such a 4 pin crystal (where just 2 are used) and a simple 2 pin crystal?

If they are both genuine crystals, the only difference is usually the 4 pin package grounds the case.
In rare cases there are also 4 pin packages with both crystal and thermistor present, used for TCXO construction. 

4 pin packages can also be Active Crystal oscillators, with Power and ground connections, and sometimes OE or EN pins.
The price of these continue to fall, and they save the BOM+placement cost of 2 caps needed for crystals.

[bson]

The ground pins turn the case into an RF shield.  The two crystal pins aren't connected to ground, or to the case.

[bigendian]

Thanks all.

I found the schematic of that crystal in the picture:


So it should be a simple crystal, correct ?

[Hiemal]

Yep! Simple crystal.

An actual full oscillator would have VCC, GND, out and some have a 4th that makes the OUT pin go high impedance.

[ataradov]

So it should be a simple crystal, correct ?

It is, but keep in mind that there is more to the crystal specification than just the nominal frequency.

Those C44 and C46 capacitors have their value matched to the crystal and oscillator circuit inside the device. So, you either need to get a crystal that has  all the parameters close to the original one or recalculate the values.

[bigendian]

HI ataradov,
thanks a lot, very kind as ususal. Yes, i did a recalculation, values should be correct.

Is the resistor necessary ?

[ataradov]

Is the resistor necessary ?

That depends on the oscillator circuit in your device. Either read the datasheet for the device if available, or If the reference schematic had it, keep it as well.

[bigendian]

ok,i have only a reference schematic, electric datasheet is not available. thanks a lot

[moffy]

When doing a measurement make sure you are using a x10 probe, higher BW and lower capacitance.

[bostonman]

Also, oscilloscope may put too much load and some oscillator circuits may not be able to handle that. You may need an active probe for that.

How do you know if a probe is an 'active' one or not? It's rare I measure crystals directly, but, when I have, usually I get a fairly normal looking wave (but I always try to measure the frequency from an IC chip that it clocks). So maybe I've been using an active probe and didn't know it. Usually my probes are the typical x10.

Also, as someone mentioned, the two other pins may be EN and OE, but they could also be a voltage reference for external tuning - assuming this is more than just a 'crystal' opposed to an 'oscillator'.

[ataradov]

How do you know if a probe is an 'active' one or not?

Price. Active probes cost $100s and you don't end up with them by accident. But really, if it has just a basic BNC connector, then it is passive. Active pobes take additional power. It will come either from the scope (pads near the BNC inputs) or from a separate power supply.

Usually my probes are the typical x10.

That would be a passive probe.

Also, as someone mentioned, the two other pins may be EN and OE, but they could also be a voltage reference for external tuning - assuming this is more than just a 'crystal' opposed to an 'oscillator'.

Do you have those pins connected anywhere on the reference board? From what I see, it is just a typical crystal.

Different oscillators circuits have different sensitivity, I've seen some that handle the probing fine and others that really don't like additional load. You also need to try probing both sides. one of them is a driving output and another one is the input. Probing the output is preferred.

What you need to capture is the startup process. Power cycle the device and capture what happens on both pins in single shot mode. You should see a step from 0 V to some voltage. You may see oscillations on top of that step that die out. If this happens, then CPU tries to enable the oscillator, but it does not sustain oscillation. This may be because of the wrong circuit design or your load.

If you don't see any voltage change, then it does not even try to start and the issue is some place else.

[bostonman]

Price. Active probes cost $100s and you don't end up with them by accident. But really, if it has just a basic BNC connector, then it is passive. Active pobes take additional power. It will come either from the scope (pads near the BNC inputs) or from a separate power supply.

Ahh okay. Then I am positive my collection doesn't include an active probe or capable of using one with my scope; the only pads I have near my scope inputs are to tell the scope whether the probe is x1 or x10.

Also, I don't probe crystals enough to justify the need for an active probe.

On a side note, usually I refer to single (two legged) crystals as 'crystals' and anything that has additional circuitry, in a can that's temp and/or voltage controlled, etc... as an 'oscillator' with the 'crystal' being inside the 'oscillator'.

[ataradov]

On a side note, usually I refer to single (two legged) crystals as 'crystals' and anything that has additional circuitry, in a can that's temp and/or voltage controlled, etc... as an 'oscillator' with the 'crystal' being inside the 'oscillator'.

Oscillator is the circuit that produces the logic-level output. Crystal is a passive component that may be used by an oscillator for timing. In case of the MCUs that use crystals for timing, the oscillator circuit is internal to the device.

In case if the standalone MEMS oscillators, they include the oscillator circuit along with MEMS-based timing element.

[bigendian]

Thanks a lot.

I finally solved my issue, oscillator was not running due to poor soldering a a QFN quite hard to solder properly.
Clock is 27Mhz and i could read the signal properly by my x10 probe.
So looks i did a good design (with all grounded around) and proper capacitor values.