--- On April 24, 2013, after 3:47:06_UTC there was a processor interruption on HMI
which required a manual reboot of the processor.
No solar images were taken for several hours following that event.
From that time on until April 30, 18:33_UTC, the wavelength tuning motor 
positions were incorrect (their actual positions during an observables 
sequence were not the intended ones). 
After these motor positions were reset the tuning of HMI went back to normal.
However, the observables (Dopplergrams, magnetograms, continuum intensity, etc.)
with a T_REC in the time interval 2013.4.24_3:47:06_UTC to 2013.4.30_18:33_UTC
are of lesser quality due to the subpar tuning, and are marked as such: 
their QUALITY keyword has the bit 0x20 raised to warn the user.

--- 12 and 24h period oscillations in HMI observables:
The large variations in SDO orbital velocity, combined with errors in the HMI filter transmission profiles and
in the model of the Fe I line profile used to determine the observables, result in oscillations in numerous
quantities computed by the observables code. These oscillations have periods of 12 and 24 hours.
The Sun-SDO radial velocity (OBS_VR) varies daily by up to 6-7 km/s.
This issue is more significant in the line of sight observables (Dopplergram, continuum intensity, magnetogram) 
that in the vector camera ones, and in the presence of strong magnetic fields than in the quiet Sun.
Moreover, in very strong active regions, a phenomenon of saturation may be reached as the velocity resulting
from the actual photospheric motions and the Zeeman split of the Fe I line may exceed the boundaries 
of the look-up tables used to infer the Doppler velocity.

--- artifacts in observables resulting from transient events
The line-of-sight and vector-camera observables require a temporal interpolation of level 1
filtergrams during their computation. Some weights used for the production of definitive observables 
have negative values. In the presence of transient events, like flares, this interpolation scheme may result in
artifacts in the HMI observables.

--- from April 3, 2013, to June 19, 2013, the same flat fields were accidentally used when producing the level 1 data 
(the flat fields were not updated every week, as they should have been). As a result, there is a jump in intensity between
eastern and western solar hemispheres on the continuum intensity data of the hmi.Ic_45s series, by about 
0.75% of the peak intensity for T_REC on June 19, 2013.
A similar issue affects the side-camera intensity data (e.g. hmi.S_720s). Observables with T_REC after June 20, 2013 (0:0_TAI)
were computed from level 1 filtergrams with updated flat fields.