a3000 scsi lockups

I'm working on an A3000 rebuild and it's almost complete, one problem remains.... the scsi drive/buss randomly locks up, locking up the entire system and requiring a CTRL+A+A to recover. Other than that it works perfectly.

I currently have OS3.1 with scsi.device 43.43 installed, with an 18GB Fujitsu disk that has a 1.5GB disk as the first partition to boot off. I am using SFS and have tried various settings for MASK and MAX TRANSFER but there is no difference.

I'm not on the computer at the moment but I can get more details if needed.... just asking if anyone has experienced any issues similar ? Also, I do have another 9GB disk that I will try during the week and see if the same issue exists.

Thanks.....

[SpeedGeek]

You should check for proper termination first. Some A3000s didn't include termination packs on the mobo. Probably, because you could get by without them using a single drive on a short cable in Asynch mode. But if you add more SCSI devices or try to enable Synch mode then the problems occur.

Your SCSI HDs probably have active termination which you should enable (for the last drive only) by setting various jumpers for TE and TP (Term Enable and Term Power).

Someone may also mention that you check the WD SCSI chip revision and upgrade to rev. 8 (or the AMD variant) but SCSI chip bugs tend to be over-rated by the guys selling SCSI chip upgrades. So this is probably your last option to resolve the problem.

AFAIK, Mask and Max. transfer are used only by FFS and not likely to affect PFS or SFS.

P.S. It might be a good idea to monitor the +5 Volt rail to see if the aging A3000 PSU can still hold it's own.

All good suggestions thanks, but I think I have most of them covered. The drive is actually in an external housing which has an auto terminate feature, there is a light that comes on at the back to signal that the chain has been terminated.

My chain looks like this;
A3000 external Db25 -> MiniHonda Db50 converter -> DB68 Wide Scsi Cable -> Housing unit -> UltraScsi Drive with SCA-2 connector

The drive does not appear to have any jumpers on it.

I have disabled SYNC mode, my understanding is that you really only get a benefit from it if there is more than one drive accessing the buss, then the devices can leave and join the buss.

I have changed the PSU with a relatively new PC one, I made a custom cable and installed a regulator to give the -12v needed.

I'm starting to think that I may have it some sort of compatibility bug between the drive and the SCSI chip, so maybe REV8 is a possibility, just gotta find one !

Any other suggestions.....??

[bdb]

Did you check diode D800? It should act as a "one-way" power (5V) gate, and if still not blown, the black band should face toward the motherboard, away from the external scsi connector.

http://www.amiga.org/forums/showthread.php?t=64847
http://www.wonkity.com/~wblock/SCSI/SCSIExamples.html

great thanks ! I will check the diode. I know for a fact that there are no terminating resistors on the controller. I have a ribbon cable and a terminator so I will try that. It gives me a way forward anyway.....

Anyone know where I can get some terminator resistor packs, and what sort are suitable for the A3000 ?

[ptyerman]

If it is a SCA drive as you mention then termination is your most likely culprit. I have several SCA hard drives and none of them work correctly on my A1200 or 2000 SCSI bus without ACTIVE termination. Leave all the termination jumpers in the off position but install a active terminator at the end of the chain, this sorts out most problems but not all. Some SCA drives require active termination on all 27 lines, there is adapters you can get that convert 50 or 68 pin SCSI to LVD/SCA with all 27 lines terminated. Just ordered one myself from the USA as they aren't all that common, I believe that should sort the problems I have had with SCA drives.

As I said the drive is in an external housing that has an active terminator built in. Infact a light comes on at the back that indicates that the termination is on. There are no terminating resistor in the SCSI host so I am using an active terminator attached to the end of a 50 pin ribbon cable... perhaps that could be connected directly to the board....

However I may be on to something, I have checked the diode D800 and it looks like the voltage there is only getting to about 3.9 volts, that may not be enough. I've checked the voltage coming out of the PSU and it all reads 5v, however I have not checked it under load.

Turns out there are some jumpers on the drive, just in a funny spot that I did not notice before, so I will go away and research those. I will also get some terminating resistors and try that.

Thanks all for your help so far, it really would be great if we can get to the bottom of this......

[bdb]

Termination "settles down" signal variance in a SCSI cable. Termination becomes more important as cable length increases and passive (resistor based) termination may be enough for a SCSI-2 connection. Active (powered) termination is always best and the bane of those used to SCSI-1 or 2 hardware, moving to SCSI-3 (which includes 68-pin U160/U320 and SCA 80-pin). The Motherboard based termination on the A3000 is adequate for short cable runs of SCSI-1 and maybe SCSI-2, but true, active termination will always dampen data signals. The setup should look like:

AT<--->{HDD}<--->[SCSI Controller]<--->{HDD}<--->{CDROM}<--->AT

Where AT stands for Active Termination. So in the above scenario, both ends of a cable have Active Termination, and the SCSI controller uses a connector; that leave only 2 connectors for drive out of a 5-connector cable.

Your A3000 may look like AT<--->{HDD}<--->[SCSI Controller]<--external-->[HDD+AT]

or lack the internal HDD. But regardless, the other side of the SCSI controller needs an Active Terminator on a cable (and check D800 to make sure it is functioning properly).

[Kin Hell]

If you run SCSI out of the A3000 on the external 25 pin header, the SCSI chain must be set according to what ever SCSI devices you have inside the A3000. Presuming just one internal SCSI drive, the drive must be set with Termination on. Pull the SCSI Resistor packs on the Mobo out (making the controller un-terminated) & then terminate the last device in the external SCSI Box. Keep Tape streamers, CDROM's & such inside the terminated SCSI chain for better results.

Seeing you are using converters also adds an element of concern. Mostly, single drives will work on un-powered converters but when you try to string several together, the 68Pin SCSI devices get confused with only 50 pins trying to control them.

Active Termination is for SCSI II onwards.
Passive Termination was more popular for SCSI I.

I Hope this might help you resolve the lock ups.

[alexh]

Which revision SCSI chip do you have? Early A3000's shipped with the -02 revision known to have bugs when you use more than one SCSI device. The -08 revision (latest) are still available for around €10

Google for :

WD33C93A-PL-00-08

[bdb]

Alexh is correct on the early Western Digital chips. A more modern replacement would be the AM33C93A-16PC. It corrects all the glitches of the WD chip. This is not a panacea to all A3000 SCSI issues.

[alexh]

I didn't know about that compatible model. They are MUCH easier to find.

http://www.ebay.co.uk/itm/400548216009

[Kin Hell]

Just re-read your 2nd post.....

Correct termination for no internal hard drives is to leave resistor packs on Mobo & terminate the last drive on the SCSI chain. The drive/s between the last external drive & SCSI Controller inside the miggy must be set with termination off.

I'm still concerned the adapters you are using, are also part of the problem.

[bdb]

Resistor packs are simple passive termination devices and as noted above are fine for SCSI 1 & 2 drives with short cables runs. The external drive in question is a SCSI 3 with active termination at its end. The passive termination at the controller may not be adequate in controlling signal fluctuations leading to the problems above.

I'm back.... I had left this project for a little while as I was getting frustrated and was having better luck with my other A2000 project.... anyway to answer all the questions....

- There are NO other devices, just the drive in the external housing.
- I have tried both ACTIVE and PASSIVE termination on the internal controller, the problem still happens.
- I am confident the cables are of good quality, excellent in fact. They are high grade enterprise quality cables that came off a HP midrange system.
- The SCSI chip is the original A3000 Proto-02 version

I knew about the rev 02/08 SCSI chip issue, but I didn't realise there was a modern replacement. So are we saying that the AM33C93A-16PC is a drop in replacement for the SCSI chip in the A3000 ??

[JimDrew]

With the A3000 you MUST have termination on the last device. The best way is an active terminator plugged into the external SCSI port. This is how my A3000 is setup.

[Jope]

It's best to also check that there's no termination on the motherboard if you want to fix it once and for all using external terminators on both ends of the bus.

Some A3000s also have a termination power problem. There is a diode for it on the motherboard and it was backwards in some revisions. If you can't measure term power from the SCSI bus with your multimeter (pin25 external db25, pin26 internal idc50).

Only have one device providing term power to the bus at a time.

[Kin Hell]

@ stevereaver

Any progress bud?

Well at this point I have tried every suggestion that has been made. Thank you to those that have made suggestions to fix this but it seems we have reached a point where new suggestions are just a repetition of ideas that have already been suggested.

I have tried every combination of ACTIVE and PASSIVE termination and I have checked the Diode, it is installed correctly but the voltage may be a little low, that is the only thing that I can see may be an issue.

I have ordered a couple of AM33C93A-16PC chips to replace the current PROTO scsi chip that is installed in there and I will report if that makes a difference.

I am fairly certain that the issue is on the A3000 somewhere as the drive/housing/cable combination works perfectly on my PC with Adpatec Card, and another drive/housing/cable combination that also works on the PC has the same issue on the A3000.

Once I swap the chip and test I will report back.

[Guho]

Originally Posted by stevereaver

I have ordered a couple of AM33C93A-16PC chips to replace the current PROTO scsi chip that is installed in there and I will report if that makes a difference. Once I swap the chip and test I will report back.

Any luck with the updated chips? Since you ordered a couple do you have any spares? I need one to upgrade from my current WD -06 in my A3000. I've ordered several but have gotten fakes where the black paint rubs off with rubbing alchohol.