Jim Oberg wrote:

Conveniently -- and I suspect, not accidentally -- the
orbit circularization burn at first apogee failed. May 1987, I recall.



A description of what went wrong can be found he
http://www.astronautix.com/craft/polyus.htm
Polyus was a rush job, and they gave it faulty software, and so:

"Because of acceleration stresses the original flight plan called for
the Polyus platform to ride back-end first under and in the wake of the
core stage of the Energia rocket as it passed through the Phase of
Maximum Dynamic Stress after launch. Since the engines for orbital
insertion would then be in the nose it would be necessary for Polyus to
yaw 180 degrees and then roll 90 degrees before they could be fired and
Polyus placed in its working orbit. When the platform was finally
launched on 15 May, 1987, the Polyus performed a 180 degree yaw turn and
then continued the turn through to 360 degrees. Polyus then rolled and
fired its orbital insertion engines, which caused it to de-orbit into
the South Pacific.

Polyus's failure to achieve working orbit was caused by a faulty
inertial guidance sensor. In the rush of construction an already built
sensor had been stripped from an existing Cosmos spacecraft and then
been inadequately tested, as the Polyus mock-up had been shipped to
Baikonur by the time the test equipment arrived at the Krunichev
Factory. Those responsible for the failure were immediately fired or
demoted."

Pat