Talk:Specific detectivity

"D* = sqrt(A * f) / NEP" and "normalized per ... and frequency bandwidth" must be wrong if unit is "cm * sqrt(Hz) / W" (which is correct) because [NEP] = W / sqrt(Hz). Where should the additional sqrt(f) come from? Could anybody who has a clue correct this and explain, what D* is good for? Thanks and regards. — Preceding unsigned comment added by 212.114.240.166 (talk) 13:04, 15 February 2016 (UTC)

I have seen this formulation repeated, for example in this document from Boston Scientific, but I agree that the units are incorrect. I think the appropriate formulation is $D^{*}={\frac {\sqrt {A}}{NEP}}$. In the second reference article, D* is defined as ${\sqrt {Af}}D$, where $D$ is detectivity in $W^{-1}$, so no ${\sqrt {Hz}}$ dependency. — Preceding unsigned comment added by 140.247.179.70 (talk) 19:30, 16 May 2016 (UTC)

To follow up, I think the main weakness here is a failure to define terms or specify units. The stated definition may make sense if NEP is defined in Watts. Note that this definition differs from the Noise-equivalent power definition on Wikipedia and also the usage immediately following defining NEP in terms of Responsivity and Noise Spectral Density.

140.247.179.70 (talk) 14:48, 17 May 2016 (UTC)

I agree the core mistake in this entry (my original, i suppose) is the definition of NEP as units of W. I see that the NEP article here on Wikipedia references microbolometer sources, however for infrared photo-detectors like HgCdTe, InGaAs, etc. NEP is often only in Watts (or specifically pico-watts). To follow the convention that Wikipedia has put NEP in W/sqrt(Hz) I don't mind changing this to be consistent. Perhaps an addendum to the NEP page in that sometimes NEP is defined without bandwidth? I could cite a few sources there.

Cbuurma (talk) 16:41, 3 June 2016 (UTC)