Measuring RF power, tricky thing

Hi there, I have found measuring RF power a tricky thing to do if not special equipment is available and I would like your comments on this.

I have an HF power oscillator that is supposed to output 2W. When I measure this in the watt meter it outputs 2W. When I measure it in the scope, it outputs about 28Vpp ie 2W, and the waveform seems somehow distorted but not too much (to the eye).
However, if I connect it to the FFT (spectrum analyzer) I see that the main carrier is only about 160mW, and it is followed by a vast amount of harmonics up to 120MHz or so.
Because both the watt meter and the scope, are broadband, they respond to the main carrier but also measure the harmonics, which add up their power to display the total power of 2W.

It is only in the FFT that the actual main RF carrier can be measured correctly.

Are my thoughts correct?

If it appears to be somewhat distorted, it is somewhat distorted. The harmonics are taking away ~70% of the total energy? Something is not right.

Output of the power oscillator is expected to be monochromatic. Perhaps something is broken. There is nothing tricky about that!

I agree, something isn't right. You didn't mention the original frequency but even a good comb generator would be hard pushed to do that.

I wonder if pushing 28V pp into a spectrum analyzer isn't overloading it and actually creating all those harmonics. A better method would be to use an attenuator to sufficiently drop the signal until the spectrum analyzer showed a fall in the fundamental (so you know it isn't clipping) then note the relative amplitudes of harmonics. You can then add the attenuation factor to the measured level to find the initial amount.

Provided the load is resistive and the scope bandwidth is adequate, and very importantly, you have the probe attenuator properly compensated, the voltage measurement method should be quite accurate. Unless there really is a serious distortion problem, the power added in harmonics should be negligible.

Brian.

Hm... an FFT overload, I haven't thought that.
Currently I use the 1M port of the scope (that has the FFT) and I terminate the transmitter with a resistive 50R load. Then measuring the voltage accross this load with the 1M of the scope.

I do not have the images of the waveforms with me now, but it is a sinewave as seen on the scope, with slight distortions.
It is something between this ../imgqa/eboard/Antenna/rf-ukxaah20q1y.png and this http://cdn.stereophile.com/images/ar...06DCSFIG07.jpg but not so distorted as the second picture.

So based on your comments, I think that it is not possible that the transmitter outputs that many harmonics?

I can't use my generator (HP8648C) or spectrum analyzer (MS610A) at the moment - there isn't space on my workbench for them!

I did try feeding my IC756pro2 to a dummy load and used and oscilloscope with FFT (Tek MSO 2032) to look at the waveform and with 1.999MHz/~2W on the internal power meter, I couldn't even see the second harmonic it was so weak.

Without getting into complex maths, I would think a signal with ~70% of it's energy in harmonics would have to be made of short, sharp pulses rather than a sine wave.

Brian.

Saying you see only slightly distorted fundamental wave, but measure only 160 mW of 2W total output power is simply contradictory. Either one or the other measurement must be completely wrong.

If you have a square wave, it has 90% of the power in its fundamental. For the power to be all in the harmonics the pulse has to be very narrow.
Frank

Thank you all.
I believe that the FFT on my scope cannot handle such large signals (how else to explain it) whereas the scope function can.
It might be a reason why the sine is not severely distorted on the scope but the FFT is full of harmonics.