How do you check a crystal oscillator circuit without scope probe effects?

[e100]

Are there test instruments that can non-invasively detect the signal?

[Bud]

Make a few turns of a thin wire into a loop half inch diameter and connect to the scope probe, then place this probe close to oscillator circuit.

[Kim Christensen]

Easiest way is to look at the buffered clock signal elsewhere.
You can kind of cheat by holding the probe tip close to the circuit, but you will disturb it slightly even then. You won't know the amplitude, but at least you can tell if it's oscillating near it's intended frequency.

[edpalmer42]

Use your scope probe to hold one end of a resistor in the 10K - 100K range.  Probe the circuit with the open end of the resistor.  It isolates your probe from the circuit.  The amplitude will obviously be affected, but the frequency and wave shape will be unaffected.

[edavid]

Use your scope probe to hold one end of a resistor in the 10K - 100K range.  Probe the circuit with the open end of the resistor.  It isolates your probe from the circuit.  The amplitude will obviously be affected, but the frequency and wave shape will be unaffected.

Of course the wave shape will be affected, it's a low pass filter.

OP, you don't say what frequency you are talking about, or quantify "non-invasively", or say what you are trying to accomplish.

You can usually probe an AT crystal oscillator with a normal 10-15pF 10X scope probe.

If it's a tuning fork crystal, you need a 2-3pF active probe.  With patience, you can find something like a Tek P6202 for a reasonable price.  Or, build a simple JFET follower out of an MMBF4416.

If you designed the circuit yourself, add a buffer circuit to the board.

[bdunham7]

It can be difficult and depends on the frequency involved.  A 100X probe with 4-6pF input capacitance may work OK most of the time to see if the oscillator is alive or dead, but it will still have at least some influence on the circuit.  A strong oscillator with an AT crystal may survive a 10X 15pF probing, but a typical CMOS oscillator like the ones built into an MCU using a smaller crystal may just collapse when subject to that much load.

[radiolistener]

Are there test instruments that can non-invasively detect the signal?

yes, you can check it with a radio receiver which allows to receive main carrier, or harmonics of oscillator. It also allows to estimate if signal produced by oscillator is clean and if it has significant frequency drift.

There is no need to connect something, just put radio receiver or it's antenna near oscillator and scan the band which should contains it's carrier.

For example, if oscillator has 10 MHz frequency you can use short wave receiver. If it has 50 MHz, you can use VHF receiver tuned at 150 MHz (3'rd harmonic)

[e100]

Use your scope probe to hold one end of a resistor in the 10K - 100K range.  Probe the circuit with the open end of the resistor.  It isolates your probe from the circuit.  The amplitude will obviously be affected, but the frequency and wave shape will be unaffected.

Of course the wave shape will be affected, it's a low pass filter.

OP, you don't say what frequency you are talking about, or quantify "non-invasively", or say what you are trying to accomplish.

You can usually probe an AT crystal oscillator with a normal 10-15pF 10X scope probe.

The context is the MCP2515 Niren style CAN bus breakout board that ships with a 8MHz crystal as standard which makes no sense as you need a 16MHz crystal to work with data streams in the 500 kbp/s to 1 Mbp/s range. Some people have been replacing the crystal with a 16MHz version (because it's easy) without reducing  the capacitor values (because they didn't read the datasheet) and then complaining about unreliable operation.

So I was wondering what simple test you could use to check for a working 16MHz oscillator.

[Wallace Gasiewicz]

So this is a separate xtal and then an oscillator, oscillator in the 2515.

If so the obvious way to check is to find a source of the osc freq away from the osc. Clock OUT? I think with the MC2515 Clock Out is 1/2 of osc freq. You can check this with the probe directly.
You may be able to just probe the OSC Out also without affecting the freq much. Might be suitably "buffered" at this point.

If you just make a loop of your scope probe and it's ground lead, you should also be able to pick up the freq you are looking for. You might get both the osc freq and the clock freq.

If you have a spectrum analyzer, you can use the scope probe with the ground lead attached to the tip also.

[voltsandjolts]

Read Section 8 'Oscillator' of the MCP2515 datasheet, it contains all you need to know.
It also explains the CLKOUT feature which you can use to attach your frequency counter.

[Swainster]

I used to use a HP 8920A RF test set to adjust DECT phone baseband oscillators. Just a piece of coax with the shield stripped off a couple of cm, connected to the RF in, with the stripped end held next to the crystal case was enough to pick up a signal. With the 8920 you can dial in the target frequency and it will tell you the error. Perhaps you could do something similar with a TinySA? Might need a RF preamp though.

[e100]

So this is a separate xtal and then an oscillator, oscillator in the 2515.

If so the obvious way to check is to find a source of the osc freq away from the osc. Clock OUT? I think with the MC2515 Clock Out is 1/2 of osc freq. You can check this with the probe directly.
You may be able to just probe the OSC Out also without affecting the freq much. Might be suitably "buffered" at this point.

As far as I can tell, you have to program the registers via SPI to enable the clock out, and that task is only possible if the oscillator is running and the chip believes it is fully functional.

Therein lies the problem, a poorly performing oscillator won't get you to the point where you can talk to it via SPI.

[edavid]

The context is the MCP2515 Niren style CAN bus breakout board that ships with a 8MHz crystal as standard which makes no sense as you need a 16MHz crystal to work with data streams in the 500 kbp/s to 1 Mbp/s range. Some people have been replacing the crystal with a 16MHz version (because it's easy) without reducing  the capacitor values (because they didn't read the datasheet) and then complaining about unreliable operation.

It seems highly unlikely that is the issue.

So I was wondering what simple test you could use to check for a working 16MHz oscillator.

Just probe the oscillator out pin.