14 Sep
2015
14 Sep
'15
6:40 p.m.
More testing this evening.
Drive 1 ran for 8 ten-minute AJRLIA read-write passes before I stopped it,
with a single error flagged during pass 6:
WD1 0235
WD2 0000
ER 4001
CB 1117
indicating a Data CRC error and a Read header without CRC command. Not
completely unexpected, a lot of data is being read and written and there is
a soft error rate of 1 in 10^9 bits and hard error rate of 1 in 10^12, IIRC.
Meanwhile I keyed in the Oscillating Seek program from the RL01/02 Users
Guide manual (page C-6) and set it for 511 cylinders. It seeks without any
problems swinging the head from Track 0 to Track 511 and back, on both
drives. No Fault, no nothing except seeking. Perhaps I was overthinking the
"wrong command" theory.
I did reseat the 2708 EPROM (state machine) on the controller card.
One VERY interesting thing I learned from a close reading of the manual and
this test program though:
The controller reports Ready status *immediately* when a seek command is
given, even though the seek has not completed!! (pg. 5-15, para 5.3.4)
And that seek can take up to 100 ms for 511 tracks on an RL02. A long time
in computer terms, even at 1.6 us per instruction cycle...
Now I'm wondering if AJRLHA diagnostic doesn't wait long enough for the seek
to complete before trying something else. The Oscillating Seek does (by
calling for a dummy Read Header command).
Still don't understand the Write error that shouldn't have been there from
that diagnostic... Not happening tonight though.
I can run RL2FMT on either drive, boot OS/8 with no problems (other than the
brief flash of both Fault lights), PIP files from one drive to the other.
Could OS/8 do its initial Seek without waiting long enough for it to
complete, too?
Would love to hear from any experts on the internal workings of OS/8
especially relating to RL02's.
-Charles