HF sinewave distortion meter, how to?

Hello, what techniques are used to measure the distortion of a sinewave HF oscillator?
Are there any relatively simple circuits I can look for?

Regularly a job for the spectrum analyzer. Of course the answer depends on the THD range you are looking at.

FvM said:

Regularly a job for the spectrum analyzer. Of course the answer depends on the THD range you are looking at.

Click to expand...

So you basically measure the harmonics level on the SA to derive the purity of the sinewave?
I am looking to measure a sinewave that seems visually undistorted on the screen of the oscilloscope.

For large distortion level, the classical method for audio THD measurements could be used, a notch filter (band stop) for the fundamental or a high pass.

FvM said:

For large distortion level, the classical method for audio THD measurements could be used, a notch filter (band stop) for the fundamental or a high pass.

Click to expand...

The visual (on the scope) distortion of the HF (10MHz or so) sinewave I need to measure is very low, so I think I am talking about low distortion levels.
Is it enough to measure this distortion by measuring the harmonics levels on the SA?

Can I also use this circuit http://www.agder.net/la8ak/images/10MHz-N.gif to cancel out the fundamental, so that I do not affect the dynamic range of the SA too much?

There are three methods to measure the distortions:
1. The one was mentioned above, using a narrow band reject filter (or a notch filter, or a steep high-pass filter) with at least 80dB rejection. The measurement is done comparing the signal before the filter, and after the filter, with a reference signal.
2. Second method is using a selective voltmeter (or a selective receiver). This method can find harmonic distortions and also the intermodulation distortions, measuring the spectral amplitudes.
The setup compares a reference signal to the measured fundamental signal and also to its harmonics. Known equations can calculate the distortions of the signal, function of the amplitude of the harmonics.
3. The 3rd method is using a Spectrum Analyzer, which the same as a selective voltmeter is measuring in frequency domain, but instead gives the spectral density of the signal (which depends also by the distortions), practically measuring the average power in a specific bandwidth.

neazoi said:

The visual (on the scope) distortion of the HF (10MHz or so) sinewave I need to measure is very low

Click to expand...

If you start seeing distortions in the scope, that means rather high level of harmonics. And don't forget that the input bandwidth of the scope, which might attenuate harmonics.

Why don't you start measuring with the spectrum analyzer "as is", without a notch filter? If the harmonics are 40 or 60dB down, you will easily see them.

And if you use a notch filter, don't forget to add a 10dB attenuator, so that the oscillators isn't working into a totally mismatched load.

volker@muehlhaus said:

Why don't you start measuring with the spectrum analyzer "as is", without a notch filter? If the harmonics are 40 or 60dB down, you will easily see them.

Click to expand...

Do you think the DSP function (spectrum analysis) of a DSO can be used as well?

- - - Updated - - -

vfone said:

The 3rd method is using a Spectrum Analyzer, which the same as a selective voltmeter is measuring in frequency domain, but instead gives the spectral density of the signal (which depends also by the distortions), practically measuring the average power in a specific bandwidth.

Click to expand...

Very useful info!

neazoi said:

Do you think the DSP function (spectrum analysis) of a DSO can be used as well?

Click to expand...

Ok, this is much more limited in dynamic range. Have a look at the instrument specs.

volker@muehlhaus said:

Ok, this is much more limited in dynamic range. Have a look at the instrument specs.

Click to expand...

I know it is a little bit out of topic, but mine is the hp54520, is it useable for this measurement or not to spend time with it?

neazoi said:

I know it is a little bit out of topic, but mine is the hp54520, is it useable for this measurement or not to spend time with it?

Click to expand...

It has 8 bit resolution, so the dynamic range for FFT is limited. But why don't you give it a try? Maybe the harmonics are "bad enough" to see them.

volker@muehlhaus said:

It has 8 bit resolution, so the dynamic range for FFT is limited. But why don't you give it a try? Maybe the harmonics are "bad enough" to see them.

Click to expand...

Thanks a lot!

If you follow your signal source with a 50 ohm attenuator (so you know the source impedance) then follow it with a parallel tuned circuit and another 50 termination . So viewing the output, adjust resonance frequency for minimum output. Swing the frequency to the +- 3dB points, now you know the Q of the circuit. Because the second harmonic is double the frequency, you do not need a sharp notch.
Now you can calculate the amount of " blow through" at 10 MHZ and then see if this voltage will materially affect the level as measured which is basically not 10 MHZ.
Frank