# How to measure RF power with an oscilloscope?

I'd like to measure the RF power coming from an RF amplifier using an oscilloscope.

The signals are sinusoids with a frequency of 50-120 MHz. The maximal power coming from the amplifier is 20 W, which is attenuated by a -3 dB RF attenuator. This power is delivered directly to the load (50 ohms), which is an ultrasonic transducer. The cables used are SMA coaxial cables.

I'm planning to put a T-splitter between the load and the attenuator, and use that to send the signal to a high frequeny digital oscilloscope set to 1 megaohms.

However, the oscilloscope can only handle 9 V RMS on its inputs at such a high frequency. So I thought that I could use a compensated 1:10 passive probe to attenuate the signal. But now I see that these probes aren't designed for frequencies like 100 MHz, and to cut the coaxial cable arrangement (for the probe) may not be the best idea, either. What is your opinion?

Alternatively, we've got some 50 ohms -3 dB RF attenuators, so I could wire them in series and put them between the T-splitter and the scope set to 50 ohms. Is this going to work? Won't they present a different load to the amplifier? Do I need to set the scope to 50 ohms in this case, or can I leave it on 1M (so it can handle a greater voltage)?

• Directional coupler would be easiest. You can pick them up from Minicircuits. – Peter Oct 31 '15 at 21:52

## 2 Answers

As your signals may go up to 120MHz, it's best to leave the scope on 50$\Omega$ input impedance.

Ideally you'll split the power with either a directional coupler, a 6dB power splitter, or a Wilkinson divider. To preserve accuracy you want something that is matched on all ports.

Then stack your 3dB attenuators in series until you are down to a power your scope input can handle.

• This seems like a good plan. The OP is planning to use a 3dB attenuator anyway. Maybe he can use a splitter instead of an attenuator at that location. The splitter insertion loss will probably be slightly higher than 3dB. If that is acceptable, that should work. He would then stack up as many attenuators as needed, as you suggest, on the path to the oscilloscope. – mkeith Oct 31 '15 at 20:53

Here is a diagram of my idea. It is a bit idealized. I think the splitter will attenuate the signal by a bit more than 3 dB. If the oscilloscope is not connected to out1, it would be a good idea to terminate it with a 50 Ohm load. Otherwise the insertion loss may be reduced.