The goodman focus value ranges from − 2000 to + 2000 the best practice (from experience) is to sample every 200. If you want to refine you should not go below a 100 step. If you already know the approximate focus value for a given setup, you don't need to do the full range, just take enough samples so it is possible to make a fit or see the trends. At least five points.
Note
Always use the narrowest slit available
For imaging, an image of the slit is obtained, illuminated by an uniform light such as a quartz lamp. A special ROI is used for faster reading. This does not affect the focus of course.
import matplotlib.pyplot as plt from ccdproc import CCDData
ccd = CCDData.read('_static/im_g_SDSS.fits', unit='adu')
fig, ax = plt.subplots(figsize=(8, 2))
ax.set_title("Imaging - g-SDSSnDome Lamp 40%")
ax.imshow(ccd.data, cmap='gray', clim=(484, 513))
In spectroscopy a comparison lamp is observed and after every line is measured and fitted, an average is obtained with sigma clip rejection.
import matplotlib.pyplot as plt from ccdproc import CCDData
ccd = CCDData.read('_static/sp_400m1.fits', unit='adu')
fig, ax = plt.subplots(figsize=(8, 2))
ax.set_title("Spectroscopy - 400m1nHgArNe")
ax.imshow(ccd.data, cmap='gray', clim=(490, 836))
import matplotlib.pyplot as plt from ccdproc import CCDData
ccd = CCDData.read('_static/sp_400m2.fits', unit='adu')
fig, ax = plt.subplots(figsize=(8, 2))
ax.set_title("Spectroscopy - 400m2nHgArNe")
ax.imshow(ccd.data, cmap='gray', clim=(487, 1157))