How to caculate wavelength of a pulse wave?

[ntr2010]

I am designing a ultrawide band transmitter which emits short pulses.To decide how long for a trace can be counted as transmission line,I need to know the wavelength of the pulses.
These are periodic Gaussian pulses of period 100ns.Pulse duration is 1ns.Two possible thoughts are:
1 Since the Fourier Transform of a single pulse have a central frequency of 4GHz and bandwidth of 2GHz,wavelength is light speed in medal(copper)/4Ghz.
2 Since wavelength is the distance of pulse travels in one period,it will be pulse duration*light speed in medal.
Obviously these two are different.
Please help me.

Thanks a lot!

 

[Mityan]

If you have just a rectangular pulse why you talk about wavelength?
But if your pulse is a shape for sine wave you should know its frequency and so estimate a wavelength.

The distance of pulse travels in one period,it will be pulse period*light speed, NOT pulse duration*light speed. So 30 meters for 100 ns.

- - - Updated - - -

And what is medal?

 

[davenn]

Hi ntr2010

you have made a bit of an error in your method

Your error is within the first part of your question. A signal with a period of 100ns will be represented as a fourier series with a lower frequency of f=1/T=10MHz plus overtones. Not 4GHz (although 4GHZ will be part of the spectrum due to the small pulse). You cannot consider the pulse width alone for the frequency analysis. You have to take into account the full period of the signal.

The two equations
wavelength=c/f and
wavelength = c*T
where c=speed of light, f=frequency, T=period are otherwise identical since f=1/T.

cheers
Dave

 

[jossion]

I think ntr2010 is talking of UWB pulses of 2GHz bandwidth centered at 4GHz. The pulses are generated at a trigger of 10MHz. I would suggest not to take 4GHz or 2GHz as a measure. For designing a transmission line you have to make sure that the group delay of the line is constant from 3GHz to 5GHz. EM simulators are the easiest way to get through and have more insight.