🚀 go-pugleaf

From article <5381@leadsv.UUCP>, by kallaus@leadsv.UUCP (Jerry Kallaus):
> I just added a C= A2090A and a Seagate ST4096 harddisk to my 2000.

Welcome to the club!

> Does anyone have any idea why only 8 out of the 9 heads on this
> drive may be used?

The original A2090 only supports 8 heads. This was supposed to be
expanded to >8 with the A2090A according to "highly placed sources" :-)
(Dale Luck and a couple of others).

Unfortunatly, this doesn't seem to be the case. I got one of the first
A2090A's, and a friend got his A2090A this month. All the part number
stickers are the same on all the PALs and PROMS, so it looks like there
have been no updates made yet.

> Prep accepts 9 as the number of surfaces.
> However, when I run FORMAT, for each cylinder counted by the
> FORMAT display, from the drive there are 9 quick clicks for the
> format followed by what sounds like 9 real fast clicks for
> the verify, and then the FORMAT display cylinder number increments
> by one and the sound pattern repeats.

Yep, that's exactly what happened to me, lots of funny noises when you
try formatting with 9 heads. If you drop down to 8 heads, the format is
nearly silent, just one "click" per track.

> When FORMAT said it was formating something like cylinder 127,
> the system seemed to hang in a loop, retrying to format that
> cylinder.

Ditto, with 9 heads, the format dies about cylinder 126. Sometimes the
system crashes, sometimes the format just hangs.

If you format with 8 heads, it will format out to cylinder 1023 without
any problem. It took mine just under a minute a megabyte (my two 34M
partitions formatted in 32 minutes each).

>  The drive actually has 1024 cylinders.  Note that
> 128*8=1024.  It seems to me that it's behaving as if there
> was one head that spans 9 cylinders instead of 9 heads
> spanning one cylinder.

No, the number of heads doesn't relate to the number of cylinders at
all. What you've got is 17 sectors * 512 bytes per sector = 8704 bytes
per track. 8704 * 1024 tracks = 8,912,896 bytes. If you multiply that by
9 heads you get 80,216,064 bytes (note that Segate is cheating here, 1K
is 1024, not 1000, so this "80 meg" drive is really only 78.3 meg). And
if you are forced to use only 8 heads you get 71,303,168 (69.6 meg).

>  It also seems like someone only
> allowed 3 bits somewhere for the number of heads.

Bingo! That's the problem exactly.

> Any ideas why this is?

The original ST506 spec allowed 3 bits for head select, and that limits
you to 8 heads. When bigger drives were created, one of the other signal
lines was stolen for a 4th head select line. This gives you capacity of
using 16 heads, but I'd happily settle for 9.

I'd guess that when there's an attempt to access the 9th head, whatever
signal line that should be the 4th head select gets hit, and you get
some unwanted reaction from the drive, rather than selecting head 8 like
you wanted. (Anyone know what signal was stolen to make the 4th head
select line?)

Dale Luck mentioned that it was simply an EPROM change to fix the A2090A
to use more than 8 heads. And I remember someone else saying that the
A2090A hardware supports >8 heads, and the limit was in the software.
So, if anyone at (or near :-) Commodore can provide a way to use more
than 8 heads there will be at least 3 of us eternally in your debt! 

All I want for Christmas is my missing 8.7 megabytes!

-- 
Blaine Gardner @ Evans & Sutherland    580 Arapeen Drive, SLC, Utah 84108
Here: utah-cs!esunix!blgardne   {ucbvax,allegra,decvax}!decwrl!esunix!blgardne
There: uunet!iconsys!caeco!pedro!worsel!blaine
"Nobody will ever need more than 64K."    "Nobody needs multitasking on a PC."