This is regarding IM measurements in dBC using PXA spectrum analyzer mode for 6 carriers GMSK modulation. The detector selected is "Average". I am comparing the IM (in dBc) between the Average types of "Log_Power" and "Power". Trace averaging is selected with count of 100.

When the average type is "Power", RMS value of all the samples in a bucket is calculated and displayed in dBm ( My dsiplay scale is dBm). Since trace averaging with 100 is activated, the bucket values in the sweeps are averaged in W , converted to dBm and displayed as final trace (I am assuming single sweep).

When the average type is "Log_Power", average of log of all samples in the bucket is calculated and displayed in dBm.Since trace averaging with 100 is activated, the bucket values in the sweeps are averaged in log mode , converted to dBm and displayed as final trace (I am assuming single sweep). Here the numbers displayed will be around 2.5 dB less than the real numbers (The numbers displayed with "Power" mode will be the real numbers).

So, whenever I change sthe average type for "Log_Power" to "Power_RMS", both carriers and IM products should chnage by 2.5dB and the IM in dBc should not change. But in my case, the carrier amplitudes are changing by 2dB and the IM products by 2.9 dB and the IM dBc is degrading by .9 dB.

Following are the numbers I am getting==>

Carrier IM IM (DBC)

Log_Power -6.33 dBm -77.91 dBm -71.58

Power_RMS -4.32 dBm -75.04 dBm -70.72

Why the carrier and IM products are behaving differently with different average type?. The IM products are close to PXA Noise Floor. Is this contributing?

When the average type is "Power", RMS value of all the samples in a bucket is calculated and displayed in dBm ( My dsiplay scale is dBm). Since trace averaging with 100 is activated, the bucket values in the sweeps are averaged in W , converted to dBm and displayed as final trace (I am assuming single sweep).

When the average type is "Log_Power", average of log of all samples in the bucket is calculated and displayed in dBm.Since trace averaging with 100 is activated, the bucket values in the sweeps are averaged in log mode , converted to dBm and displayed as final trace (I am assuming single sweep). Here the numbers displayed will be around 2.5 dB less than the real numbers (The numbers displayed with "Power" mode will be the real numbers).

So, whenever I change sthe average type for "Log_Power" to "Power_RMS", both carriers and IM products should chnage by 2.5dB and the IM in dBc should not change. But in my case, the carrier amplitudes are changing by 2dB and the IM products by 2.9 dB and the IM dBc is degrading by .9 dB.

Following are the numbers I am getting==>

Carrier IM IM (DBC)

Log_Power -6.33 dBm -77.91 dBm -71.58

Power_RMS -4.32 dBm -75.04 dBm -70.72

Why the carrier and IM products are behaving differently with different average type?. The IM products are close to PXA Noise Floor. Is this contributing?

Of course, back inthe day, SA modes only had log detectors, so only log power could be read. Today you have a choice.

The reason for the discrepency is that some numbers in log that are noisy go much smaller than (more negative) than the same noise going positive. And the extreme example is if the noise cancels exactly the signal and generates a zero power...take the log of that and add it to your avg and you still get -infinity!