Adaptec SCSI Driver problem (?)

Before the problem:

Adaptec 2940 Ultra/Ultra-wide SCSI controller. Two hard drives
and CD-ROM on internal 50-pin Connector (narrow; 8-bit). External LS-120 SCSI drive (8-bit device) on an external 16-bit
bus (special SCSI cable 68-pin at PC and 50-pin at device).

Then had head crash. Original boot drive is toast. Bought a new
SCSI drive, a 68pin (wide; 16-bit) one. Now the SCSI controller has three strings:

Still on 50-pin cable: one hard drive and the CD-ROM.
External LS-120 setup unchanged.
On new 68-pin cable: the new replacement hard drive, the boot drive.

Using the Adaptec SCSI utility, the SCSI BIOS was set to boot from the LUN (6) of the new drive. Also, “fast” support was enabled.

Much of this works OK. At boot time, the SCSI BIOS is installed
successfully. It finds all SCSI devices. The new drive was C: and the other older drive was D: I’ve been able to use DOS FDISK to partition and format the new drive, so that it now has a primary dos partition and two logical drives in an extended partition. From a DOS diskette did a SYS C: and thus have a very minimal system on the new C-drive which boots OK.

Here’s the problem. In the good old days when all the SCSI devices were on the same narrow bus, I had a DOS bootable diskette which loaded drivers like ASPI8DOS.SYS from Adaptec’s EZSCSI 5.0. This was followed by other drivers like MSCDEX, and by booting from that diskette I was able to access the CD-Rom.

Now however, the ASPI8DOS.SYS fails and the system hangs during boot. So no access to the CD-ROM, which I need because I have all the data backed up from the original (crashed) drive and want to copy it to the new drive. Also, running DOSINST from the EZSCSI diskette does not work either. If I choose for it to scan for the controller, it hangs. If I choose to pick a controller, it does not give me the 2940 as a choice (which is a bit peculiar since it did so when all devices were on the narrow bus.)

All this might be because I now have a mixture of 16-bit and 8-bit devices. Some questions/thoughts:

1> Would updated EZSCSI drivers be needed?
2> Or maybe updated SCSI BIOS for the 2940
3> Does something need to be fixed in the CMOS bios?
4> Maybe the support for "fast" is causing a problem
with the LS-120.
5> Could I not disable the "fast" support and see what
happens? But I should be able to run the new drive
at the fast (16-bit) speed.

I have looked at the Adaptec web page, and nothing seems to address this directly. Any thoughts?

 

In the interim, take everything off the SCSI chain except the CD and the Boot drive to copy over the OS. Once that is done reconnect everything and see what happens. Personally if it were my network or PC and I needed all of the drives I would get a seperate controller for the narrow devices and keep them on one chain and have another for the wide devices.

 

I have the straight 50 pin 2940. When things get screwed up I just remove the card in device manager and reboot. Then I load the latest BIOS available from Adaptec. Seems to restore items that have gotten lost.

Your situation with the wide bus mixed with narrow is certainly more complex and it might not help in your case. I would certainly get the latest BIOS from Adaptec though.

There is a warning about the adapter cable having to have active termination of the High Byte: http://ask.adaptec.com/cgi-bin/adap...NhdF9sdmwxPTImcF9zb3J0X2J5PWRmbHQmcF9wYWdlPTE*&p_li= You might want to go through their database.

Rockn’s suggestion might be the easiest if you can’t get it sorted out. There is usually no problem with having more than one SCSI card. The Tekram DC395 is going on Pricewatch for about $45 and should serve your narrow drives just fine.

 

Thanks. The post on the Adaptec ASK has the following:

"Do not connect SCSI devices to all three connectors of an AHA-2940W or AHA-2940UW. "

It turns out that my card is indeed an AHA-
2940UW, and it indeed has devices on all three connectors. In addition, the post talks about the importance of terminating the High Byte when going from a 68pin to a 50pin. This is what the cable going from the 2940 to the narrow LS-120 drive is SUPPOSED to have, but I don't really know.

So, I will try removing the external LS-120 drive and see what happens. Will keep you posted. Also, the BIOS level of the 2940 is V1.25 (1996!) which is pretty backlevel; so an upgrade there shouldn't hurt. Longer term, separating the narrow and wide devices is probably a good idea.

One thing at a time, however .....

 

Well, simply disconnecting the cable to the LS-120 made the controller happy. ASPI8DOS loads OK now, and the new boot drive and the CD-ROM are both accessible, so I have been able to proceed.

I intend to ask Adaptec's ASK people why that model controller can't use all three of its connectors at once. Seems like I'll have to have a 2nd controller as a long term solution in order to have access to everything.

Thanks for your help

 

Looked more through ASK on the adaptec site, and they have the question and the answer thus:

Can I use all three connectors on my AHA-2940UW at the same time?

No. The internal 50 pin connector is a parallel branch of the lower 8 bits of the wide bus. Using all the connectors creates a "Y" connection instead of the point-to-point bus configuration required for proper operation.

The wide bus on the AHA-2940UW can be pictured as two parallel horizontal lines, the 'upper eight data bits, and the 'lower' eight data bits. The lower bus is tapped and routed to an internal 50 pin SCSI connector. If using an internal 68 pin hard drive at one end of the parallel lines, the other end of those lines must have a 68 pin terminated wide device on an external cable, an external non-wide device with an appropriate adapter to terminate the upper bus, be terminated on the host adapter (and have no other devices), OR have a non-wide device on the 50 pin internal connector with the host adapter termination set to
"high on/low off" or "Automatic". This maintains the point-to-point requirement of the SCSI bus.

 

Some lessons on SCSI and cable lengths.

In the above, I was able to proceed with setting up the new wide boot drive, but the LS-120 drive (which I wanted to have ultimately) was not working. It turns out that part of the reason for this was that the SCSI cable length exceeded the maximum allowed. Here is the guide:

If you have a FAST device , and if you have 3 or more devices, then the max cable length supported by the specification is 4.5 feet.

In my setup, once the wide boot drive was added, I could not use the 3rd connector on the 2940UW card. However, ASK (the Adaptec support data base) stated that in such a case, one could extend the internal narrow bus, and add the device to that.

So I bought a connector which converted from the internal 50pSCSI to the external SCSI-2 connector.
Then I connected the LS-120 drive and the internal narrow cable to it, thinking that this would work OK, since I was then using only two of the three connectors on the 2940 card.

Wrong! You have to add up all the cable lengths, and meet the 4.5 foot limitation. In my case:

· Internal wide cable was about 3 feet
· Internal narrow cable was about 4 feet
· External SCSI-2 cable was about 6 feet

for a total of thirteen feet! (Note in my original setup before the HD crash, I had no internal wide, an internal narrow of 4 feet, and an external 68p to 50p cable of 6 feet. This ten feet did not meet the limitation, but the setup still worked.)

The symptom of the cable length problems was the following: during boot, you’d see a message from the Adaptec SCSI BIOS identifying all the devices which it found on the bus. This was OK. However, the boot then hung, instead of proceeding with the message: “SCSI BIOS installed successfully!”

Adaptec support gave me the following work-around so that I could verify that it was a cable length problem instead of something else:

1> Press CTL-A during the boot in order to enter the SCSI Select Utility
2> Go into “Configure/View Host Adapter Settings” and “SCSI device configuration”.
3> For each device actually present, do the following:

- Set “Initiate WIDE Negotiation” to NO
- Set “Initiate Synchronous Negotiation” to NO
- Set “Max Synch Transfer Rate” to the lowest permitted (10 or 5 Mbps)

The result of this is to increase the allowed cable length under the SCSI specification to 3 meters (almost 10 feet). Of course, things are running slower, but this is only for a test. Doing this resulted in an OK boot in my case. Next on my agenda is to get some shorter cables.

 

Thanks for the feedback. I was hoping that the wide and narrow had separate cable length accumulations. Those restrictions just about eliminate the practicality of having an external chain and an internal fast HD on the same card without slowing the transfer rate to a degree that will really hobble the HD. 10Mb/sec is ridiculous for a fast SCSI HD.

SCSI2 has a cable length restriction of 6 meters without a SCSI HD attached, but reduces to 3 meters with a HD. I have 5 SCSI devices but none are HDs, so I can be pretty cavalier about my cables without problems. If you limited yourself to 9’ you should be able to go to 20Mb/sec. 2 internal 18” cables plus your external should work at 20Mb/sec.

If I ever hit the lottery and decide to go to SCSI HDs I’ll just use a separate card.