I am trying to generate an 11 MHz pulsed chirp signal with center frequency of 4 GHz at a power level of +10 dBm. I have an arbitrary waveform generator that can generate the correct chirp signal, but can only supply a power level of -40 dBm. I am determining this using a spectrum analyzer, and measuring the average amplitude of the 11 MHz band. The waveform on the spectrum analyzer is generally flat, with maybe +/- 0.5 dBm variation.

So I added two amplifiers in series (part number ZVA-213-S+). As far as I can tell this should give me a minimum of 40 dB gain, or a typical of 50 dB gain. Unfortunately there is some kind of saturation happening, and I am maxing out at around -16 dBm. I can lower the power level of the arbitrary waveform generator until I see the power level dropping, but once I start increasing the power level, I can't get past -16 dBm even though the waveform generator claims its output power is still got room to increase. The spec sheet of the amp claims that it can provide an output of +33dBm, typically.

I repeated this test using a single tone pulse at the same frequency. With this I could see the expected gain of 50 dB.

I verified everything was fine in terms of providing power to the amplifier. It's the recommended 12 VDC, and the power supply can provide up to 5 amps.

Is there some fundamental difference when applying a chirp signal to an amplifier versus a single tone that would cause this?

  • \$\begingroup\$ Measure the current the amp is drawing when you amplify the chirp. If it is about the same as when using the sine, then the amplifier probably isn't saturated. \$\endgroup\$
    – JRE
    Commented Sep 28, 2018 at 18:04

2 Answers 2


Are you sure you are using that SA correctly? Chirps are tricky to measure, especially as 11MHz likely exceeds the widest res bandwidth that instrument is capable of.

What RBW have you selected?

Basically the SA is looking at a bandwidth much narrower then 11MHz at any given time and is sweeping that band to produce its plot, your chirp however is only spending a tiny amount of time in that bandwidth during each chirp... Throw in some averaging and get nonsense.

My bet is that your generator is putting out something a good 30dB higher then you think it is, and that you are massively saturating the SA input stage, it is for this kind of thing that owning a RF power meter as well as a SA is so very useful (They however have their own issues).

  • \$\begingroup\$ I agree. I would set S.A. to 4GHz with widest BW and Video Filter then reduce until you minimize noise floor without affecting peak amplitude. Then go on zero sweep with > 11MHz IF BW centred on spectrum and use X axis like a scope to see carrier envelope power level. \$\endgroup\$ Commented Oct 4, 2018 at 19:10
  • \$\begingroup\$ you don't need to do this, most of the SA can measure wide-band power. the option is most commonly called "Channel Power Over Bandwidth" \$\endgroup\$
    – Dan
    Commented Oct 14, 2018 at 7:33
  • \$\begingroup\$ @Dan Depends on the generation of SA in play, modern units with digital IFs of reasonable bandwidth can of course just do channel power, while older mix/filter/log amp/video filter style gear clearly cannot. My modern Aeroflex can, my CRT based Anritsu, not so much. \$\endgroup\$
    – Dan Mills
    Commented Oct 14, 2018 at 18:25

The spec sheet of the amp claims that it can provide an output of +33dBm, typically.

No it doesn't; it says that the third order intercept point is +33 dBm and that is not what the maximum power output is. The quoted figure is +24 dBm and this is probably when powered from a supply of +12 volts. Even at this level there will be 1 dB of amplitude compression so you might be realistically looking at something like 20 dBm and, as you haven't drawn a schematic, I am assuming your you have made a mistake in your schematic.

The data sheet says this: -

withstands open/short load at 1dB compression point output power

And that implies to me that if you maxed out the power and were not careful about a correct load you could have damaged one or both parts.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.