measuring frequency higher than osciloscope bandwidth

Thread Starter

Kardo22

Joined Mar 12, 2014
49

crutschow

Joined Mar 14, 2008
34,281
Depends upon how much higher the frequency is than the oscilloscopes bandwidth.
If you can see the signal, than you can measure the frequency, but the oscilloscopes frequency accuracy is likely much less than the crystal accuracy.
 

Lo_volt

Joined Apr 3, 2014
316
Digital scopes are going to be limited by the sample frequency. There's really not much you can do if your oscillator frequency is higher than your sample rate.

If you have an analog scope available and if it is close enough the bandwidth of the scope, you may be able to display a heavily distorted view of the oscillator output. This should allow you to at least measure the frequency even without the high fidelity needed to view the actual signal. Of course, as crutschow says, measuring frequency is not particularly accurate on a scope.
 

tautech

Joined Oct 8, 2019
383
Hi,
I can't find any good info on it. If I only need to measure crystal oscillator (mounted on a board) frequency, can I use oscilloscope with lower bandwidth and still get accurate results? If yes, then how would I determine how much higher I can go?

One example of what I need to measure is below crystal, 180MHz version (mounted on PCBA):
https://eu.mouser.com/datasheet/2/698/REN_XL_Crystal_Oscillators_Datasheet_DST_20220302-1954959.pdf
This is one of those cases when any measurement actually changes the measurement value however with a DSO frequency counter you'll get a reasonable indication of the frequency it's operating at although if it's well beyond the rated BW of the scope the amplitude will most certainly be less. If substantially higher frequency you could be in aliasing territory but comparing the trace timebase/div against the frequency counter can indicate if it's an aliased measurement.

So in circuit and operating, an ordinary scope probe may load the crystal too much and an active probe with much lower capacitive loading the preferred measurement technique however that really depends on the circuit design as it may be quite robust enough to be probed and not impact on its operation.
Just knowing this stuff ^^ is often enough to make a judgement on whether your measurement is anything like representative.
 

ag-123

Joined Apr 28, 2017
276
I think for digital oscilloscopes, an idea is to consider aliasing. i.e. it is actually measuring a beats signal that is

frequency - sampling frequency

the max amplitude of this beats signal is likely just equal to the actual signal. But the wavy signal pattern or some such pattern is actually the beats signal rather than the actual signal. to find the actual frequency, and considering that it is beats, actual frequency = sampling frequency + beats frequency. it is likely pretty accurate.

Note that for the higher frequencies, the ADC or input stage gain could be reduced as well, so even the max amplitude could be lower than actual.

This tactic I think is used for the various SDR (software defined radios), it recovers the A - B (IF) signal by simply considering aliasing. This omits a low pass filter that is normally necessary.
 
Last edited:

drjohsmith

Joined Dec 13, 2021
852
Hi,
I can't find any good info on it. If I only need to measure crystal oscillator (mounted on a board) frequency, can I use oscilloscope with lower bandwidth and still get accurate results? If yes, then how would I determine how much higher I can go?

One example of what I need to measure is below crystal, 180MHz version (mounted on PCBA):
https://eu.mouser.com/datasheet/2/698/REN_XL_Crystal_Oscillators_Datasheet_DST_20220302-1954959.pdf
would it not be great if I could use a cheap scope to measure stuff outside its spec,
Who needs a high speed scope then.

Unfortunately, you get what you pay for
a scope for instance has an analog front end,
This is band limited / filtered ,
For a digital scope, filtering is part of the scope front end to prevent aliasing
for an analog scope, this is to prevent intermodulation

This filtering is not a birck wall
so you will probably get something through at higher frequencies than the spec,
but the accuracy will be diminishing, as the filter attenuates more,

You ask for accuracy, so as you see, it wont be accurate.

You could make something to look at the crystal output
you could band limit the crystal output,
mix that with a known frequency
and filter the output
to generate a signal in the range of your scope
its like an old analog radio down mixing,
but accuracy , you will have to calibrate the system ,
probably using a higher quality scope !
 

DickCappels

Joined Aug 21, 2008
10,153
One of the things I like about my old Tektronix scopes is that the rolloff was nice and smooth. Using the FFT mode I was able to set up a 330 MHz transmitter even though the scope was spec's to 60 MHz.
 

ScottWang

Joined Aug 23, 2012
7,397
You could use the GHz range counter to pre-divides(exp:/10) the input signal and then the scope probe to test the signal from the output of the counter, but you need to multiply(exp:*10) the frequency shown on the scope.
 

rsjsouza

Joined Apr 21, 2014
383
Hi,
I can't find any good info on it. If I only need to measure crystal oscillator (mounted on a board) frequency, can I use oscilloscope with lower bandwidth and still get accurate results? If yes, then how would I determine how much higher I can go?
One specific class of oscilloscopes that can display very high frequencies is called "Sampling" oscilloscopes - some manufacturers also call it ETS or "Equivalent Time Sampling". They can show waveforms by randomly sampling their inputs and rebuilding the waveform in the screen little by little. For that to be successful, the waveform must be repetitive, constant (just like what you would see in a fully working oscillator) and the equipment must be built for it.

As others have said, with a "Real Time" oscilloscope, accurate measurements will not be possible - you will have to account for some attenuation and/or distortion.
 

drjohsmith

Joined Dec 13, 2021
852
One specific class of oscilloscopes that can display very high frequencies is called "Sampling" oscilloscopes - some manufacturers also call it ETS or "Equivalent Time Sampling". They can show waveforms by randomly sampling their inputs and rebuilding the waveform in the screen little by little. For that to be successful, the waveform must be repetitive, constant (just like what you would see in a fully working oscillator) and the equipment must be built for it.

As others have said, with a "Real Time" oscilloscope, accurate measurements will not be possible - you will have to account for some attenuation and/or distortion.
You are dead right .
But sampling scopes tend not to be cheap , which might not be what the op wanted
 

MrChips

Joined Oct 2, 2009
30,708
You can use a binary 2^n divider circuit to bring the frequency down to the range of the oscilloscope. The consequence is a corresponding loss in precision.
 

upand_at_them

Joined May 15, 2010
940
The specified oscilloscope bandwidth is the -3dB frequency. I've used my 50MHz Rigol DS1052E (okay, hacked to 100MHz) to check 144MHz transceivers.
 
Top