USRP BasicTX and LFTX - what is the maximum transmit power?

Let's say this first: if you really want to know what the transmit power in your setup is, you will need to measure it yourself. Don't rely on whatever you might find in the net.

Having said that, Juha and I were searching the net with little success. We wanted to know the maximum TX power output levels that the USRP-boxes we had on our desk could produce. The various pieces of information we found were either confusing, not relevant for our hardware or simply conflicting.

Practically all RF power amplifiers (PA) expect a nice 50-ohm load such as a properly impedance-matched antenna for the frequency range in question. If there is a problem with the antenna (matching), the reflected power can destroy the amplifier's final stage unless you take precautions. When prototyping, you have the responsibility and there are at least three ways to engineer this:

1. Use a sufficiently rugged system that survives any load
2. Isolate the reflected power so that it does not go back to amplifier but gets dumped elsewhere
3. Monitor the reflected power and adjust the gain of the PA to avoid bad things happening.

Our choice was #1 mostly because it is a simple approach that would still provide a sufficient power level at the output. The advice from the Mini-Circuits technical support revealed that if we can keep the input level to ZX60-100VH+ less than -5dBm, the amplifier should survive anything Svalbard can offer. Our most likely failure case would be open load i.e. the antenna is gone with the wind or a broken feedline. We already had to repair one antenna cable when testing the HF setup at Svalbard, so things like this do happen.

So, what is the output from the USRP? Do we need to add an attenuator a.k.a. "pad" to limit the input to the PA or not?

The measurement setup was simple: a 50-ohm dummy load was connected to the USRP and an oscilloscope was used to measure the output peak-to-peak voltage. A simple flowchart in GNU Radio Companion produced a sinusoidal wave with adjustable frequency and amplitude. We had two different daughterboards, BasicTX and LFTX. We set the amplitude to maximum and measured the power levels at several frequencies.

f (MHz)	BasicTX (dBm)	LFTX (dBm)
1	-1.7	4.7
2	-1.6	4.7
3	-1.6	4.7
4	-1.6	4.6
5	-1.6	4.5
6	-1.6	4.5
7	-1.6	4.4
8	-1.6	4.2
9	-1.6	4.2
10	-1.6	4.0
11	-1.6	3.9
12	-1.6	3.8
13	-1.6	3.5
14	-1.7	3.4
15	-1.7	3.2
16	-1.7	3.0

So, the BasicTX outputs a maximum of about -1.6dBm while the LFTX can reach 4.7dBm. We also did a quick measurement to see what we get when adjusting the amplitude. We did this test at 3MHz.

Level @ 3MHz	Measured (dBm)	Nominal (dBm)	Delta (dB)
1	-1.7	-	-
0.9	-2.5	-2.6	0.1
0.8	-3.5	-3.6	0.1
0.7	-4.6	-4.8	0.1
0.6	-5.9	-6.1	0.2
0.5	-7.4	-7.7	0.3
0.4	-9.3	-9.6	0.4
0.3	-11.5	-12.1	0.6
0.2	-14.7	-15.6	1.0
0.1	-19.5	-21.7	2.2

The amplitude changes pretty much in a way you'd expect. The comparison to theoretical values shows that at larger amplitudes things are ok. We did not spend too much time wondering about the large differences at small amplitudes: the measurement setup simply could not provide the accuracy to say much more. Measuring small things is difficult and requires great(er) care.

Our measurements were done in a quick-n-dirty style in order to select correct system components before continuing with testing the HF radar prototype in field. One always expects to destroy a few components when developing stuff, but a few quick checks can save a lot of money/time. To be on the safe side, we need a 4dB pad for BasicTX while the LFTX would need a 10dB pad. Of course, we did not have any 4dB pads and used a 6dB model. And then measured the PA output and adjusted the BasicTX output level to minimise harmonics, but that's another story.