I've been measuring the power from my ATSC transmitter by looking at its output on my spectrum analyzer. It turns out, however, that that probably is not an accurate way to judge the transmit power.
I found a PDF that has an explanation.
A spectrum analyzer is a radio receiver, whose output is hooked up to the Y axis of an oscilloscope. The X axis is a sawtooth wave, like a normal oscilloscope, but in addition to sweeping the X axis of the display, it also adjusts the frequency of the receiver.
There's a bit more to it than that, however.
One of the characteristics of the radio receiver that's part of a spectrum analyzer is its "Resolution Bandwidth." This is the bandwidth of the receiver. That is, in order to plot an amplitude value on the display, how wide of a swath of the RF spectrum do you sample in order to determine what that amplitude is?
The resolution bandwidth of a SA is tied to the span of the display. The larger the span, the higher the resolution bandwidth, which means the wider a signal has to be to actually show up. At the same time, the smaller the resolution bandwidth, the slower the sweep has to be in order to display the finer grained samples.
If the signal you're watching fits inside of the resolution bandwidth, then the entire signal contributes to the amplitude that shows up on the display. But if the signal is wider than the resolution bandwidth, then at any one given moment, you're only plotting a fraction of the signal's power on the display.
It turns out that if you know the 3 dB bandwidth of the signal (for ATSC it's 5.83 MHz), and the RBW of the SA, then you can calculate a correction factor: 10 * log (RBW / Signal BW) dB is how far down your SA will show the signal. That is, you subtract that number (which is probably negative) from the displayed power level to obtain the real one.
I took a picture the other day that showed my transmitter's signal at the 2nd division down from the top, and the reference level was -4 dBm, with 10 dB per division. That's a signal strength of -14 dBm. But the RBW was 100 kHz, so the correction factor is -17.65 dB, which actually put my output power at +3.65 dBm, or about 2.3 mW.