) (08/31/88)
THE RD54 IS NOW WORKING on the MAC!!!
I got the RD54 successfully installed as an RLL device. It is
one manufactured 10/87. It seems that older RD54 drives CANNOT
be RLL encoded.
It now shows up on the Mac as a disk with 227,875k on it!!! I have
it plugged into the INTERNAL SCSI port on my MAC-II directly to
the 50-pin connect on the ACB-4070 board.
I loaded it with 117meg of stuff (64meg was 2 32meg files (Partitions))
and let Disk Express Examine Volume/Crunch Desktop/Unfragment files
to give the drive a real workout. It performed like a champ. It
did constant I/O for 5 hours without a hitch.
I had one question though for the group...If I do a 'SHOW DEVICE/FULL'
of an RD54 on my UVax II, it shows as 1225 cylinders, but on 2 other
uVax IIs, it shows as 1221 cylinders...How come? I used 1221 on my
installation just to be safe, but why the discrepency???
I did NOT enter any bad sectors into the drive table. The Adaptec
manual states that it does ON THE FLY sector removal and updates
its own defect list by moving sector formats down on the disk.
So the users format program should never see any bad sectors from
a format. Anyway, my RD54 has 103 factory bad defects and the format
defect list that the controller stores in 18 'zone' defect maps (See
the manual for details) listed 104, including all the ones on the
drive, so the ACB-4000 series does it for you!!!! I played with
sending direct SCSI commands to the drive to do this....
Also, you know when the drive does a recal or seek or whatever
and it makes the loud hum noise? On a Vax it only does that on powerup
when the heads load. It does it for most accesses on the Macintosh
especially the 'Updating Desktop' message on a copy. What is that
sound? Is it a seek or a recal or what? If it is a recal, I have
a problem, if it is a seek, why does it do it for 1-2 seconds 3
times for every desktop update?
I set the drive to 1:1 interleave, since I have a Mac II, but I think I am
going to reformat as 2:1. The drive seems kinda sluggish, the interleave
might be the problem. The driver I now use along with the 50% speed
improvement given by RLL encoding might be too fast for the driver.Though,
I am not sure of this though, since the ACB-4070 controller has
a 1k buffer and reads in multiple sectors on every read. Maybe
the 2:1 causes this algorythm to slow just enough for it to hit
every other block better than every block. I'm not sure....
Phil Hunt