dBV was the implementation on the TDS3000 series. I've verified the 5 series FFT and SpectrumVu readings are dBm. What can be off is when using 1 Mohm channel impedance or a probe. Our dBm calculation assumed load (50 ohms) doesn't change. External loading is assumed in these cases.
This is easy to check with a sine (pure tone). The single peak in the FFT should match the Vrms measurement. I measured 87.32 mVrms on my sine. Across 50 ohms this is 152.5 mW or -8.17 dBm. My FFT reading was within a dB of this.
When you have multiple peaks, it is harder to match to the Vrms measurement. You can get close by taking the majority peaks, convert to Watts to sum, then calculate Vrms. My measurement was 118.4 mVrms.
Are you familiar with SpectrumVu? This is a digital down conversion within our 5 series acquisition hardware that generates baseband IQ. The 5 series then analyzes baseband IQ the same way a spectrum analyzer would. Additionally, SpectrumVu and analog acquisitions can be enabled at the same time. The only cost is halving the max sample rate. Since you are well bellow the maximum sample rate, you have headroom to turn it on. I can think of three key benefits of SpectrumVu;
-independent analog and spectrum displays letting you optimize each. With traditional FFTs, there is usually a tradeoff between optimizing the time-domain display or the frequency-domain display.
-automatic markers that annotate peak frequency and power.
- use of a chirp z-transform to make SpectrumVu more accommodating to various RBW/record-lengths. (FFTs are constrained to powers of 2 vector lengths. you only have indirect control of FFT RBW via record length. FFT peaks can vary almost 1 dBm based on optimum or sub-optimum record lengths.)
