4
\$\begingroup\$

I need to generate a 1 GHz bandwidth chirped pulse, 100 ns long. In the past, I have done this using fast arbitrary waveform generators (Keysight M8195A), but this is an expensive and very general solution to a very specific problem. For this project, I would need a more effective solution (in form factor, power draw, and overall cost).

Are there common ways to generate such a waveform? Can you provide me with some pros and cons of each method?

\$\endgroup\$
5
  • 3
    \$\begingroup\$ Two words: "dispersion filter" \$\endgroup\$
    – Dave Tweed
    Jul 16 at 2:43
  • 1
    \$\begingroup\$ @DaveTweed I'm also interested - could this really last as long as 100 ns? It feels like to delay the 1.5 GHz component of an impulse or step by this much you'd need an unreasonable number of poles. I've filtered sharp square waves to generate narrow-band impulses, but only got 5 or 10 cycles out. \$\endgroup\$
    – tomnexus
    Jul 16 at 5:53
  • \$\begingroup\$ @DaveTweed Interesting. So, if I understand correctly, you are talking about generating a ~0.5 ns-long transform-limited pulse with (bandwidth of 1 GHz) and launching it through a dispersion filter to spreads it over 100 ns. Is that right? \$\endgroup\$
    – LDPC
    Jul 16 at 6:49
  • \$\begingroup\$ Yes, that's the idea. I'm hardly an expert on the concept, but I know it has been used in radar systems. A quick search last night showed that there are many ways to implement such a filter, such as SAW devices. I was thinking that it might also be done with microstrip-based filter structures on a PCB. \$\endgroup\$
    – Dave Tweed
    Jul 16 at 10:21
  • \$\begingroup\$ I see. 100 ns however is ~30 m, considering the speed of light. So, wouldn't I need a filter that is really long disperse my 1 ns pulse to become 100 ns? The way I think about it, if I have a 1m long filter, and the refractive index at 1 Ghz is 1 and at 2 GHz is something huge like 10, then one of the components would take ~3 ns to leave the filter, and the other would take ~30 ns, leading to a 27 ns pulse. Honestly, that would work for my application - even 10 ns would. But this is a 1 m long filter, with a huge refractive index delta... \$\endgroup\$
    – LDPC
    Jul 19 at 16:46

0

Reset to default

Your Answer

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

Browse other questions tagged or ask your own question.