0
\$\begingroup\$

The question is, does a RF frequency multiplier work with ultrashort pulses?

In another words, does the frequency multiplier change the pulse width of input signal?

For example, suppose we have a analog signal oscillating at 100MHz and it was amplitude modulated to be a 50ns-width Gaussian-shape single pulse (note, there was only 5 periods of signal).

  1. Will the output signal of frequency doubler/tripler be a 200MHz/300MHz gaussian pulse? What is the pulse width?
  2. What if the signal was amplitude modulated to be 50ns-width square shape pulse?

【update】 To be more specific, my signal is a frequency modulated analog f0=100MHz pulse with pulse width of ~10μs, where the frequency modulation \$\delta f=\delta f(t)\$ is a function of time during the pulse width.

For example, \$\delta f\$ is about 1MHz at t<5μs, and \$\delta f\$ jumps right away to be ~10MHz at t=5μs, then at t=8μs, \$\delta f\$ jumps to ~15MHz.

I want to extract the frequency jumping curve \$\delta f(t)\$with high time resolution (say, at least with 100ns resolution), but as you can see, the oscillation duration is very short (a few cycles). A short time Fourier transform analysis can not resolve the frequency jump accurately.

So I wonder, can a frequency multiplication process enhance the frequency difference to \$N\delta f(t)\$, but also the frequency jumping curve shape is not changed? Then the core question is, will a frequency multiplier work with short signals with few cycles?

I have read papers that uses frequency multipliers to generate wideband linear frequency modulation (LFM,\$\delta f=\alpha t\$) pulse signal from a narrow band linear frequency modulated signal . I'm not familiar with electronics, but I guess a frequency multiplier will also work on a nonlinear frequency modulated signal? Am I right?

\$\endgroup\$
  • \$\begingroup\$ I suspect a multiplier only works well over some limited bandwidth and a short pulse is composed of a wide spectrum so it probably wouldn't work all that well. \$\endgroup\$ – jramsay42 Feb 1 '18 at 0:52
  • 1
    \$\begingroup\$ "A RF frequency multiplier" is too broad. There are many different technologies that can be used to multiply a frequency, ranging from simple passive harmonic generators to active circuits like PLLs. They have very different characteristics, so you'll have to be a lot more specific about what you're talking about. \$\endgroup\$ – Dave Tweed Feb 1 '18 at 4:41
  • \$\begingroup\$ As other comments suggest, what kind of frequency multiplier? The classic tripler should be fine, as the input is rich in third harmonic; its output will be approximately sinusoidal. The XOR/delay doubler will give unsatisfactory mark/space ratio, for two examples. \$\endgroup\$ – Brian Drummond Feb 1 '18 at 10:47
1
\$\begingroup\$

There are dozens of different multiplier technologies that vary in linearity, frequency output, input , harmonic, harmonic rejection.

Do the math and then define what you need for harmonic levels below the desired harmonic then see what multipliers will perform what you need.

The technmology varies from active diodes, Schottkey didoes, pin diodes, balanced mixers, MEMs and other diode bridge structures with filters with a vary wide variation in price and quality.

But multiplying a 100MHz sine with a 50 ns square pulse is basically a synchronous 50 Ohm switch for 5 cycles. A square pulse can be considered x1

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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