confirming the pulse-bandwidth-product for fs laser pulses

mamoamamoa

Hey everyone,
hope this is the right place to ask this question so I'll just put it out here.

I've been measuring the pulse duration of a mode locked Ti:Sa laser with an autocorrelator. For a central wavelength of λ_0 = 800 nm and the spectral bandwidth of the Ti:Sa Laser (FWHM) Δλ = 8 nm, the Gauss fitted autocorrelator funtion gave me a pulse duration of (FWHM) Δt = 132 fs. I'm trying to confirm that Δt * Δf ≥ 0.44.

Δt needs to be deconvoluted by a factor of 1/√2 for a Gaussian pulse shape and Δf should be given by Δf = Δλ * c / (λ_0)^2, which would give me 0.35 and that can't be true. Any help is really appreciated. Thanks,

Mamoa

Edit: Δλ and λ_0 are pretty stable so I'm not worried about those. That means either my pulses are not Gauss shaped (setup problem) or I'm measuring the wrong pulse durations (autocorrelator problem), would you agree?