Pulse duration from interferometric autocorrelation

[Voxynn]

Hi,

I have some interferometric autocorrelation traces of ~20fs pulses. Does anyone know how to convert to or calculate the actual pulse duration from the fringes of the trace?

Thanks,

Richard

 

[DrFurious]

Are your pulses Gaussian, Sech, or Lorentzian?

Use the pulse type to figure out the time bandwidth product $$\delta t \delta \omega$$. The spectrum tells you $$\delta \omega$$. For example, a Gaussian has $$\delta t \delta \omega = \frac{2 \log 2}{\pi}$$.
Use this equation to solve for $$\delta t$$

 

[DrFurious]

Woah sorry. $$\delta t \delta \omega$$ is $$4 \log 2$$. What I quoted was $$\delta t \delta \nu$$. In case you wonder, $$\nu$$ is ordinary frequency in hertz and $$\omega$$ is angular frequency.

 

[Voxynn]

My pulse is gaussian (or near enough). I know how to calculate time/bandwidth product for actual spectra (wavelength vs intensity etc) but how can I extract similar info from the autocorrelation trace? I've attached a picture of the trace I have.

 

[DrFurious]

Units?

 

[DrFurious]

Check out page 34 here:
https://urresearch.rochester.edu/fi...ionalItem.action?itemId=7332&itemFileId=13953

Looks like the number is 1.5331, but I don't know. If it's the right answer, do repost.

 

[Voxynn]

The data is taken from an oscilloscope, so the units are time and voltage (intensity).