frequency bin , amplitude and time resolution

Can somewone help me with understanding what frequency bin is?
Many strophysical papers use this notation, but i do not understand why i need a frequency bin, i would just Fourier transfom the series and look at the powers and decide what frequency belongs to my signal...

Now, when talking about signal, especially sinusoids, what does Amplitude A of about 8% mean, why "%"? 8% of what?

and the last quostion: one paper says: the signal was binned to a time resolution of ... milliseconds. What ist time resolution in this context?

Can someone help?

practical Fourier transforms are done in the digital domain. This implies that the result is a discrete set of values, and not a continuous function of frequency.
Each discrete frequency which is output by the DFT is called a "bin", and covers a range of frequencies. Much like a barchart....

So even though you may have a particular bin with the largest power/ amplitude, it does not necessarily mean that its exact center-point frequency is the max frequency. You need to look at the powers in the adjacent bins (or bars) and do an interpolation to get a more accurate frequency value as the max value. Several fairly simplistic methods have been used for this, each with its pros & cons.

Regarding the 8%, it obviously means a percentage of the total power of the signal in the captured sample. Not amplitude.

Lastly about milliseconds, i'm not sure what is the reference. However a longer sample of your signal which has been sampled at a very high rate would likely give you a more precise (i.e. frequency resolution) than one sampled at a lower rate. I would have to look at a larger context of the paper to say exactly what is meant.

thank you for your respond!
In the paper of Israel and Stella, 1996, they write:


"V0332+53 is a 4.4 s X–ray pulsar accreting from a Be star in a transient X–ray binary. Its
light curves are characterised by a pronounced red–noise type variability. The 4.4 s pulsations have a small amplitude (A≃8%; Stella et al. 1985 and references therein). We re–analysed a 4
hr long 1–9 keV light curve obtained on 1984 January 24 with the EXOSAT ME, during which the source was relatively faint (average rate of∼36 counts s^-1). The light curve was binned to a
time resolution of 938 ms and a single sample spectrum calculated over a 16384 point interval ..."


would this provide you some more information?
what do they mean by "average rate of∼36 counts s^-1"?

- - - Updated - - -

by the way, the paper says something about "the binned data", do you think they are still talking about the same thing (in the frequency domain)?

Only a little more information.... i'm not an astro expert.

My earlier description of 8% is probably incorrect. From this text it would seem that the 8% refers to the relative amplitude of the pulsars intensity variation with respect to the mean pulsar brightness (or magnitude)

the 36 counts s^-1 means 36 counts / second. S^-1 implies s to power minus 1, or 1/s .

My limited knowledge says that a pulsars absolute magnitude is proportional to its pulsing rate. So i'm not sure why a rate of 36/sec is faint... it must be very very distant !!

the 938 ms and 16384 points issue is still not clear to me ...sorry.

thank you