You have to measure the FWHMsample of each peak of your sample (take into account only Ka1 if you don't have Ka2 monochromator) and subtract the instrumental FWHMinst at that particular diffraction angle.
To obtain the instrumental FWHM function of the diffraction angle you can pass a commercial standard for XRD (LaB6 or Al2O3) to determine the FWHMinst. Then fit FWHMinst VS 2teta with a polynome.
You have to measure the FWHMsample of each peak of your sample (take into account only Ka1 if you don't have Ka2 monochromator) and subtract the instrumental FWHMinst at that particular diffraction angle.
To obtain the instrumental FWHM function of the diffraction angle you can pass a commercial standard for XRD (LaB6 or Al2O3) to determine the FWHMinst. Then fit FWHMinst VS 2teta with a polynome.
As Marco said, you need to have a standard sample. Standards were developed which show no crystallite or strain broadening, which is why you need one. Take data on a wide range for the standard, fit each peak using some equation which defines your peaks well. I have used the equation in this paper PHILOSOPHICAL MAGAZINE, 1999, VOL. 79, No. 8, 1769-1786 with success. I might disagree with Marco about ignoring the Ka2 peak unless you are at a high-2theta where they are fully separated. Really, you should fit two peaks using the assumption that the Ka1 and Ka2 peak separation is a function of angle based upon their wavelengths.
If you find a value or function online which someone else made for their XRD instrument, it does not apply to you because every instrument is slightly different.
Once you have FWHM as a function of angle using a standard, you fit a quadratic function weighted by the uncertainty in FWHM. You should get uncertainty when you fit the FWHM from the standard peaks. Using that quadratic fit (and the uncertainty in the fit) you can get the instrumental broadening at the wavelength's you are interested in.
That paper I suggested also talks about how to pull instrumental broadening out of your measured sample data.
You can measure the instrumental broadening by using a commercial sample of either LaB6, Si or Al2O3. The instrumental broadening can be done one time for your samples, and not frequent.
You can visit my ResearchGate page and go through these works:
ESTIMATING crystallite size and microstrain by warren-averbach ? - ResearchGate. Available from: https://www.researchgate.net/post/ESTIMATING_crystallite_size_and_microstrain_by_warren-averbach#58b1dc1748954c11bf326d53 [accessed Feb 25, 2017].
Data Final - Microstructure and crystal imperfections of nanosize...
Data Microstructural characterization of Al-Mg alloys by X-ray di...