When using the PXA to measure the power of a signal in the complex domain I observe a 3dB difference between an IQ sinewave and a IQ modulated signal (e.g. WCDMA, DVB-H etc.), even though they have the same (real) power. Any comments?
When you modulate a signal, its bandwidth grows. You need to make sure to integrate over the signal bandwidth to measure the entire signal power. So, in the Spectrum Analyzer app you would increase your RBW or use the Channel Power measurement and make sure the integration / channel bandwidth is large enough. In the IQ Analyzer mode, you can also increase your integration bandwidth.
Of course we integrate over the entire signal bandwidth. For example when using a 10MHz LTE (RF) signal, we integrate [-5MHz, 5MHz] in the complex IQ domain.
Just to make clear on the measurement: I'm using a direct conversion receiver to down-convert an RF signal and measure its power in the PXA. I compare the results when using a CW (tome) at RF and when using a modulated signal (for example LTE, WCDMA, DVB-H) both with the SAME power level. When measuring in spectrum analyzer mode the integrated power of a modulated signal over Signal Bandwidth/2 and compare it either with peak or with the integrated power of the CW tone in baseband in either I or Q channel the results match. However when I change to PXA I+jQ mode then for a CW I see the same power as in spectrum analyzer mode but for modulated signal signal I see 3dB less (integrated over Signal Bandwidth this time to account for negative freq as well). This does not make sense, so I wonder if there is something in the FFT used for I+jQ PXA measurements that we are missing....
It appears you are using the front panel I/Q inputs. Is this correct?
When you performed this measurement: “However when I change to PXA I+jQ mode then for a CW I see the same power as in spectrum analyzer mode but for modulated signal signal I see 3dB less (integrated over Signal Bandwidth this time to account for negative freq as well).”
Did you connect BOTH the I and Q signals from the DUT to the PXA front panel I and Q inputs when you performed the I=jQ measurement? If you did not, this is the likely cause of the 3 db difference.
I did an experiment with the I/Q inputs. When I generate a W-CDMA signal, the integrated power when I remove either the I or Q input when using I+jQ input path drops by 3 dB. Conversely, when I generate a multi-tone signal with two tones at the same phase, then when I remove the I input, the signal disappears from the display. When I remove the Q input, no difference in power is seen. When I offset the phase of the second tone by 90 degrees, then when I remove either I or Q, the power drops by 3 dB. When I offset the 2nd tone by 180 degrees, then when I remove the Q input, the signal disappears from the display (all of the power is in the Q channel). Now, when I compare the integrated power between the I+jQ input and just I only input in the case of the two tones that are at the same phase angle, the power is 6 dB higher for the I only path. When I compare the integrated power between I+jQ input and I only input when the phase angles are offset by 90 degrees, then I see a 3 dB higher result for the I only path. Therefore, when the power is contained in just one of the I or Q inputs, you will see the signal measure 6 dB lower when you set the input to I+jQ as opposed to using a single channel (I only or Q only) input. This is because of the Fourier Transform of Asin(wt) = 1/2A u(W-w) + 1/2A u(W+w). When the power is equally in the I and Q channels, you will see the signal measure 3 dB lower when you set the input to I+jQ as opposed to using a single channel (I only or Q only) input.
When you modulate a signal, its bandwidth grows. You need to make sure to integrate over the signal bandwidth to measure the entire signal power. So, in the Spectrum Analyzer app you would increase your RBW or use the Channel Power measurement and make sure the integration / channel bandwidth is large enough. In the IQ Analyzer mode, you can also increase your integration bandwidth.
Regards -