Transmitting a simple sine wave in gnuradio using the USRP - and measuring it's power

Transmitting a sine wave with the USRP is a simple matter of connecting your USRP to the computer and running an equivalent gnuradio flowchart as displayed below.

The gnuradio flowchart consists of a Signal Source block and a UHD: USRP Sink block. Sliders are added to adjust the amplitude and frequency.

The signal source is set to zero frequency with an amplitude of 1 - this serves as the baseband signal. The signal is then modulated by the UHD: USRP sink block with a center frequency set by the user. The center frequency represents the carrier frequency that up-converts the baseband signal. The gain is set to "normalized" with the maximum value of 1. This ensures that the USRP is always transmitting at it's highest power. 

The reason for setting our baseband/signal source frequency to zero is that we only measure the actual output power of the USRP at it's center frequency, avoiding any modulation effects. 

Now that your script is ready, all you need is a transmission line/antenna to connect to one of the RF-channels of the USRP. Our setup looked like this:

3D system diagram. The output of the USRP is fed into a T-piece with a 50 Ohm dummy load at the one end and a high-impedance oscilloscope probe at the other. 

The version seen on the picture is a USRP2 with an LFTX daughterboard mounted inside. As the script is running we can observe and do measurements on the transmitted waveform using the picoscope. Making sure that the probe is correctly compensated, and that we observe the actual true waveform by tweaking the settings of the digital oscilloscope accordingly, we measure the peak-voltage over a range of frequencies whilst keeping the amplitude at maximum.

The power in dBm (PdBm) is related to the peak voltage (Vp) by the formula:

                                                    

                                                    PdBm = 10 + 20 log10(Vp)

Measuring over the HF-spectrum resulted in the following output from the LFTX daughterboard:

The figure on the left displays the output power as measured when using the setup shown above, reaching 4.1 dBm at 1MHz. When adding an additional adjustable attenuator between the USRP and the T-piece, and using a 300W capable, 50 Ohm dummy load instead of the small one seen in the picture above, we get power reaching up to 4.9 dBm at 1MHz.

The first run of measurements peaked at 4.1 dBm. When including the adjustable attenuator and using a heavy-duty dummy load, the LFTX peaks at 4.9 dBm. Why the discrepancy? can it be that the added components adds up to a slightly higher impedance? 

The same measurements were conducted again, this time using the adjustable attenuator with the small black dummy load. These measurements showed that the attenuator (set to 0 attenuation)
had a negligible impact on the circuit. So it comes down to which dummy load we are using. 

Based on these measurements we can guesstimate that the maximum output power of the LFTX lies somewhere in the range between 3 and 5 dBm. 

RF circuits assume a 50 Ohm impedance along the whole transmission line. The discrepancy we observed are for for the time being believed to be caused by some impedance mismatching between the dummy loads.

Note: The input impedance of the oscilloscope was 220k.