A3000 and A4091
CyberDez
recently I've purchased an A3000D (rev. 9/03) in very good conditions
and I'm trying to squeeze from it the best performance possible through
a serie of upgrades I'm planning to do from now on.
This machine has already installed the Kickstart 3.1 (40.68 A3000 version).
The very first upgrade would been the replacement of the internal SCSI-I
chain with an external-only SCSI-II using a 4091 controller I pulled out
from my 4000D where it worked flawlessy. I know this card require the
Super Buster rev.11 to work so I transplanted this chip in the A3000 too.
Here begins the troubles.
Sometimes at power-up before the workbench is loaded a message about an
error reading a certain block appear (but not all the times).
During the intensive DiskSpeed test the system often lock-up while
testing the two biggest transfers, but sometimes it succeed and
sometimes not.
If I decide to use the workbench normally without perform any speed test
usually all goes on without problems.
Here are additional infos about my configuration:
- No SCSI devices attached to the standard A3000 controller, but
resistor packs are placed there as onboard termination.
- The A3000 SCSI chip is a WD33C93A-PL 00-04 PROTO
- The same 4091 with the same HD when placed in the A4000 works like a charm
- Tha partition parameters on the SCSI HD connected to the A4091 are
Mask 0x7fffffff and MaxTransfer 0xffffff. I've tried with Mask
0x7ffffffe and 0xffffffff too but with the same results.
- The attached HD is a Seagate ST34250N partitioned using the CyberSCSI
tool v 1.30 (the latest available). Using HDToolbox with this particular
drive lead to a system crash.
- The Fast memory is onboard only with Zip chips marked KM44C1000AZ-8
for a grand total of 8 Mb of Fast mem
- No accelerator card is mounted but testing the system with a Blizzard
4030 has show up no changes in reliability.
- Just for curiosity I've tried to attach this drive to the internal
A3000 SCSI and all works very well, no crashes, no lockups but the
maximum transfe rate is about 2.8 Mb/s only compared to the 5 Mb/s
obtained with the A4091.
- I've carefully cleaned and reseated all the sockets and the custom
chips in it.
I suspect that the SuperBuster 11 alone is not enough for a reliable use
of the A4091 inside an A3000 and other components should be updated:
the mind goes to Dmac and Ramsey that actually are in revision 02 and 04
respectively but are really they the cause of these lockups?
Anyone of you A3000 users online has ever tried to use the A4091 in your
computers? If yes, what have been the results?
Many thanks in advance to all who will help me.
CyberDez
Everett M. Greene
> recently I've purchased an A3000D (rev. 9/03) in very good conditions
> and I'm trying to squeeze from it the best performance possible through
> a serie of upgrades I'm planning to do from now on.
> This machine has already installed the Kickstart 3.1 (40.68 A3000 version).
> The very first upgrade would been the replacement of the internal SCSI-I
> chain with an external-only SCSI-II using a 4091 controller I pulled out
> from my 4000D where it worked flawlessy. I know this card require the
> Super Buster rev.11 to work so I transplanted this chip in the A3000 too.
> Here begins the troubles.
> Sometimes at power-up before the workbench is loaded a message about an
> error reading a certain block appear (but not all the times).
> During the intensive DiskSpeed test the system often lock-up while
> testing the two biggest transfers, but sometimes it succeed and
> sometimes not.
> If I decide to use the workbench normally without perform any speed test
> usually all goes on without problems.
>
> Here are additional infos about my configuration:
>
> - No SCSI devices attached to the standard A3000 controller, but
> resistor packs are placed there as onboard termination.
>
> - The A3000 SCSI chip is a WD33C93A-PL 00-04 PROTO
There's the most likely cause of your problem. The early
33C93 chips were buggy. Yours isn't even a production
version. It's been years since I worked with those chips,
so I can't tell you which version was the first to work
reliably.
Daniele Gratteri
> There's the most likely cause of your problem. The early
> 33C93 chips were buggy. Yours isn't even a production
> version. It's been years since I worked with those chips,
> so I can't tell you which version was the first to work
> reliably.
As far as I know they labelled as "PROTO" the very first production runs
for every revision of the chip. I had a -04 PROTO on a 2091 board and it
worked just fine. Also, if you better read the post, you'll see that the
A3000 SCSI controller works fine, he has problems with the added 4091 one.
--
___ __
/ __|___ Daniele Gratteri, Italian Commodore-Amiga user... ///
| / |__/ Nickname: FIAT1100D - ICQ: 53943994 Ritmo S75 __ ///
| \__|__\ Home page: http://www.gratteri.tk forever! \\\///
\___| E-MAIL: dan...@gratteri.tk ...since 1990 \///
CyberDez
> Also, if you better read the post, you'll see that the
> A3000 SCSI controller works fine, he has problems with the added 4091 one.
Hello Daniele,
yes you're right: the problems are with the A4091 (that works flawlessy
in a 4000D) not with the onboard A3000 Scsi controller.
Looking around the net about a possible solution I suspect that A4091
need a Ramsey -07 too, not only a SuperBuster 11, when used in a 3000D.
But as all serious Amiga user knows, this means a replacement for the
DMac too. I found that those two chips are quite rare (expecially the
Dmac -04) and expensive so I would buy them only if I'm sure they will
fix those problems with A4091.
Someone told me that if the onboard A3000 scsi chain is left empty some
malfunctioning occours in other parts of Amiga, the floppy drive for
example, maybe the Zorro DMA too?
Yes, I plan to upgrade the WD33C93 chip to a rev. -08 too but in the
main time I would fix (if possible) this issue between the A4091 and the
A3000D.
As additional info, the ROMs on the A4091 are the latest available:
rev. 40.13
Best regards and thank you for joining the thread :-)
CyberDez
wishbri...@hotmail.com
There are several reference files on the Web that explain this. In
short, the A4091 was never recommended for the A3000. It might work,
might not. Your right about the need for a Buster 11. The basic need
for the trio (Buster 11, Ramsey 07, DMAC 04) concerns the A3640. Here
is some of the excerpt, I edited it to answer your question, so please
read thru it carefully.
by Oliver Knorr, 1997-12-28
Kickstart 2.04 ROMs
These are 'normal' AmigaOS ROMs, that do not require booting a
Kickstart file.The newer A3000s were delivered with these ROMs. These
ROMs will work with the A3640, except if you have the combination of a
RAMSEY-04 chip (which almost any A3000 has) and static column Fast-RAM.
In this case, you need to exchange the SCRAM at least in bank 0 with
page mode RAM chips or (even better) upgrade to AmigaOS 3.1 ROMs. You
can not upgrade the RAMSEY to the new rev. 07 (which would also fix
this problem), because for this you would also need a DMAC-04, which
is not available anymore.
RAMSEY and DMAC
The RAMSEY and DMAC chips are important A3000 custom chips that, among
other things, implement part of the A3000's onboard DMA SCSI host
adapter. Nearly all A3000 were delivered with the revision combination
of DMAC-02 and RAMSEY-04. The problem with this combination is, the
DMAC-02 does not like 68040 boards like the A3640. There are both heat
and timing problems. If you are lucky, the tolerances of your DMAC-02
are wide enough to work together with the A3640. If you are not so
lucky, SCSI will not work correctly anymore with the A3640 installed.
To overcome this problem for the A3000T-040, Commodore designed new
revisions of these chips: RAMSEY-07 and DMAC-04. If your DMAC-02 has
problems with the A3640, all you would need to do is to exchange your
old RAMSEY and DMAC revisions with the newer ones. But in reality, you
can not buy a DMAC-04. In general, they are no longer available anymore
(except some single pieces that dealers might have in stock).
The RAMSEY-07 is available as a spare part (because it is also used in
the A4000T), but this will not help you. The problem is in the DMAC,
not in the RAMSEY and a RAMSEY-07 together with a DMAC-02 does not work
at all. You need either both chips in the old revisions, or both in the
new revisions. If you have an extremely old A3000, you might even have
a DMAC-01. This one runs very hot and has even more problems with '040
boards. I also heard of a RAMSEY-03, but don't know any difference to
RAMSEY-04.
So, if you have a DMAC that does not work with the A3640, you are
basically out of luck. What you could do in this case:
Buy another DMAC-02 (still available) and hope that one works with the
A3640 by chance. Replace the onboard SCSI host adapter by one on a
Zorro II or Zorro III plug in card. Forget about the A3640 and either
don't use an '040 board, or use one with a SCSI host adapter of its
own.
BUSTER
Revision 11 BUSTER implements Zorro-III DMA, so you can use the
Fastlane Z3 or A4091 boards, which is not possible with older BUSTERs.
The rev. 11 BUSTER is also
faster than previous versions and more compatible with different plug
in cards.
IF you read that all the way thru, save your money. A DMAC04 and
Ramsey07 will not cure your problem with the card, as the file suggests
a SCSI Zorro card. Some would say I answered you questions the long way
around, but my resource of technical info is usually darn excellent.
CyberDez
> There are several reference files on the Web that explain this. In
> short, the A4091 was never recommended for the A3000. It might work,
> might not. Your right about the need for a Buster 11. The basic need
> for the trio (Buster 11, Ramsey 07, DMAC 04) concerns the A3640. Here
> is some of the excerpt, I edited it to answer your question, so please
> read thru it carefully.
Thank you very very much, wishbringer.
After reading your post I'm more oriented now in purchasing a WD33C93-08
instead of a Dmac-04 and a Ramsey-07.
If the A4091 has been designed with the A4000 in mind there's no reason
to waste money and time in trying to made it working in the A3000.
There are other high performance ZIII Scsi Cards (like the Fastlane, for
example) or Scsi controllers integrated in accelerator cards that on
A3000 can do the same task of the A4091.
However, it would be nice if someone that has tried before this
experiment (either with or without succesful results) could post here
his/her experience.
Again, thank you for your precious infos and to all others who helped me.
Best regards,
CyberDez
wishbri...@hotmail.com
The WD33C93-08 only pertains to the A3000 motherboard, and certain SCSI
controller cards which also have the WD33C93 chips.
David Nelson
"CyberDez" <snake...@yahoo.it> wrote in message
news:44f75a83$0$47970$4faf...@reader3.news.tin.it...
> The very first upgrade would been the replacement of the internal SCSI-I
> chain with an external-only SCSI-II using a 4091 controller I pulled out
> from my 4000D where it worked flawlessy. I know this card require the
> Super Buster rev.11 to work so I transplanted this chip in the A3000 too.
> Here begins the troubles.
The problems you're having with the 4091 are similar to the problems I've
seen with a Fastlane Z3's SCSI on an A3000D. It might work, but usually
doesn't. Recently it was suggested to me to perform the INT2 hack to cure
the problem. I forget right now the technical reasons behind the suggestion.
I have not had enough free time to give it a try, but I wanted to throw it
out here. I do intend to try it.
-David