Amiga 3000 SCSI no longer working

[Lianna]

Hi,

I've recently got back in to the hobby again and after putting a Pistorm in my A500+ I've been wanting to upgrade my 3000 with a ZuluSCSI (old, noisy harddrives truly terrify me). However, dusting off my 3000 I discovered my SCSI is no longer working.

As soon as I connect anything but just the ribbon cable to the SCSI bus the SCSI LED comes on on startup and it won't recover from that. Whether I connect the old 240MB harddrive, the ZuluSCSI, or the official Commodore external terminator on the DB25 connector (just has some resistor packs inside) either alone or in any combination the system won't boot. I've of course tried with and without termination set, different id's, etc.

It boots fine (off the gotek) if nothing is connected to the SCSI bus, and everything else appears to work just fine (went through what I could test using the amiga test kit).

System:
* A3000UX with ROM tower, 40.68 roms
* INT2 mod is done
* Stock is 030/25 with FPU and MMU, just 2MB chip
* Cyberstorm MkII 060/50 with 128MB 60ns RAM
* AM33C93A-16PC SCSI chip
* All other Amiga chips are also latest revisions etc

Things I've checked:
* Leakages: no, I recapped the Amiga, and replaced the battery with a coin cell in 2016.
* I initially tried the bare bones system without the Cyberstorm, adding that back to the system made no difference.
* I've put a new lithium coin cell in (it wasn't keeping its time), and I made sure to set the SCSI prefs to something sensible. So, it should not be amnesia.
* I've checked pin 25 on the external connector, that has 5V. (The ZuluSCSI LED also lights up, also indicating it it is getting power off the bus).
* I have another SCSI chip (WD33C93A-PL, marked PROTO) around, swapped that out ... made no difference. Can't remember why I have that chip and whether it was working when I put it in the box many years ago (may have been used in a GVP or such, and back then I did sometimes upgrade chips to latest or replace broken ones). I have no other means of testing either chip.

I've not been able to find anything else to try beyond perhaps replacing the diodes for the SCSI termination power, but it is getting power so not sure about the merit of that.

I'm at a loss on what to try next to try and get this SCSI going again. So does anyone have any suggestions on what to try or what the problem might be? (I do have a scope, but I'm not an electronics expert at all ... I mostly just know how to solder).

[Matt_H]

The main thing I can think of is amnesia. You said you set the SCSI prefs, but did you also go through the amnesia procedure, i.e., fully resetting the NVRAM? Once upon a time my 3000 wouldn't cooperate until I did that.

Got another ribbon cable you can try? Is it mounted correctly, i.e., not misaligned on the motherboard connector?

[thebajaguy]

Hung LED = likely termination problem.

Mobo may or may not have 3 removable SIP terminators, and your HD, at the other end of the cable (SCSI bus) also needs to have termination enabled.

If there are none on the mobo, there needs to be an external DB25 termination module (does the same duty). The rule is that termination needs to be at the 2 most distant points of the SCSI bus.

Do turn off the Sync setting in the CMOS bits if on. It makes it more sensitive to bus termination irregularities. (speculation) The ZuluSCSI might not like the Sync negotiation query - there are devices that hang when the (typically) at-boot-scan query is made.

the 33C93A almost always has the PROTO label on the silkscreen. -04 was the common chip shiped. The -08 was the last version, and was the best one to have (C= never was good at working around the earlier bugs in it with some equally odd devices' firmware not helping maters).

Maybe locate an active SCSI termination module - they are better than passive at times.

Good luck.

[Lianna]

Originally Posted by Matt_H

The main thing I can think of is amnesia. You said you set the SCSI prefs, but did you also go through the amnesia procedure, i.e., fully resetting the NVRAM? Once upon a time my 3000 wouldn't cooperate until I did that.

I used http://m68k.aminet.net/package/disk/misc/SCSI-Prefs originally as that has a GUI and SetBatt is lacking documentation on how to use it (can't find a "procedure" online either). Not afraid of the shell, just afraid of undocumented stuff.



Now also used SetBatt which did complain about amnesia, no longer does (after "setbatt amnesia"). Managed to turn sync off, fast sync off, and set timeouts to long.


It made no difference.

Originally Posted by Matt_H

Got another ribbon cable you can try? Is it mounted correctly, i.e., not misaligned on the motherboard connector?

I've tried two different cables yeah, neither working. Both used to work as far as I can recall (the longer one absolutely did). It's definitely not misaligned on the connector. No bent or missing pins.

[Lianna]

Originally Posted by thebajaguy

Hung LED = likely termination problem.

Mobo may or may not have 3 removable SIP terminators, and your HD, at the other end of the cable (SCSI bus) also needs to have termination enabled.

Yeah, I know a hung LED means it is likely a termination problem, and I've definitely had to resolve such issues in the past. Always managed to, until last night.

Originally Posted by thebajaguy

If there are none on the mobo, there needs to be an external DB25 termination module (does the same duty). The rule is that termination needs to be at the 2 most distant points of the SCSI bus.

The resistor arrays on the motherboard are not populated. I've not yet tried populating them, but given that plugging in the external terminator on the DB25 connector locks it up all the same I don't think that's the solution either, and it would stop the external port from being usable. It's one of the few things left to try that I haven't yet ... though in the past I've never needed to either (passive external terminator has always worked fine).

Originally Posted by thebajaguy

Do turn off the Sync setting in the CMOS bits if on. It makes it more sensitive to bus termination irregularities.

Yeah tried that, doesn't help.

Originally Posted by thebajaguy

(speculation) The ZuluSCSI might not like the Sync negotiation query - there are devices that hang when the (typically) at-boot-scan query is made.

Sure, though just connecting a terminator or harddrive on the bus (no ZuluSCSI), properly terminated or not ... always a hung LED. Attaching anything to this bus causes a hung LED. Can't get to the Early Startup Screen when this happens.

Originally Posted by thebajaguy

the 33C93A almost always has the PROTO label on the silkscreen. -04 was the common chip shiped. The -08 was the last version, and was the best one to have (C= never was good at working around the earlier bugs in it with some equally odd devices' firmware not helping maters).

Maybe locate an active SCSI termination module - they are better than passive at times.

If using an active terminator is my last resort then I'll try finding one, but it used to work fine in the past.

[MrFluffy]

I realize you have moved on from wondering if its the zulu causing the hangs and now think it could be hardware, but Ive recently dealt with a scsi2sd (4.x series, older, but the device the zulu was inspired from), and on that, if the partitions on the sd card didn't match the config applied to the scsi2sd firmware and the card itself was ok, it hung the scsi bus during discovery like you describe. Its also sensitive to sd card quality, as is well known there's a lot of fake sd and micro sd cards out there, only 2 of the 5 I tried from my stash were good in it.



I also pulled a old scsi hard disk out of storage to eliminate the bus being the issue, and it turned out all my old scsi hard disks don't work at all either after being in storage for years on a shelf and did the same thing even playing with termination etc, I ended up proving the scsi cards in both machines were ok with a external tape drive and a active terminator on its external housing then going back in deeper to the scsi2sd config and card formatting.



Being able to pop the sd card out and put it in something else to transfer or take a dd of the entire card for backup is really nice and is worth the grief though. And no singing ancient old hard disk sounding like a jet engine during takeoff.

[Lianna]

It's been a while but haven't had the opportunity to look at it (limited space taken up with modding the A500+) and I was waiting for newer roms and the 3.2 OS. But, it's working now, somehow, and I've managed to get 3.2 installed. One of my long cables is definitely not working, but 2 others including one I tried before are working. It even works with and without termination, including with the external terminator that locked things up before, or both with and without the resistor packs on the motherboard. Absolutely baffling.

Thank you all for your input.

[mousehouse]

Congratulations! Nicest machine ever in my opinion - despite the quirks.

[Lianna]

Originally Posted by mousehouse

Congratulations! Nicest machine ever in my opinion - despite the quirks.

They're horrible to work in, the taking it apart to get to anything, all the metal bits, metal scraping taking it apart or put back together, finding the right screws ... it's a faff. And when it doesn't work, like with this thread, it gets a bit tedious and frustrating, especially on the hardware side.



But yes, I have loved the A3000 ever since I've had it. Can't recall since when, it was donated to me as part of payment for some freelance contracting I did back then in the Netherlands (I'm Dutch) and I was told it was one of the first A3000's in the Netherlands, given or supplied to them by Commodore directly.



I've considered selling it too, but just can't bring myself to parting with it even though I don't have a particular purpose for it.

[jrobbins70]

Sorry that I'm late to the party here...

Anytime my a3000 won't boot and the SCSI LED is on, it's a termination problem. Sometimes something really dumb like I have an external scsi cable connected out the back of my a3000 that's either connected to nothing (loose) or connected to a device that's turned off. I just disconnect the external cable and it boots again.

Anyhoo, just adding to this thread for someone in the future who might be chasing termination issues.

[Vypr]

I've got the exact same issue but this time it's not a termination issue.
Same symptoms. Black screen and Solid LED light.
Nothing plugged nto external SCSI, resistor packs in place (and swapped for new ones just in case)
Internal SCSI device has internal termination, but have toggled this off and put an Active Terminator on the end of the cable. Also tried several cables that work fine on other machines. Nada.
The only thing I have found that works is to plug a processor card like an A3630 or A3660. Once it has that plugged in everything works fine.

[Vypr]

Originally Posted by Boing-Ball

So did you swap the SCSI chip?

Yes, and all the custom chips except for Amber and the Super D_MAC. I doubt they would be the ones ot go faulty, if they had it wouldn't work with a different processor card.

[jrobbins70]

Originally Posted by Vypr

I've got the exact same issue but this time it's not a termination issue.
Same symptoms. Black screen and Solid LED light.
Nothing plugged nto external SCSI, resistor packs in place (and swapped for new ones just in case)
Internal SCSI device has internal termination, but have toggled this off and put an Active Terminator on the end of the cable. Also tried several cables that work fine on other machines. Nada.
The only thing I have found that works is to plug a processor card like an A3630 or A3660. Once it has that plugged in everything works fine.

Anything else that could be affecting the situation such as an accelerator card installed (improperly) and/or the int/ext jumpers not set correctly? My a3000 will black screen with SCSI LED on if I use my 040 board set incorrectly.

Just a thought.

J

[mech]

Sockets,sockets sockets, in many unexplained cases like this where it begins working out of the blue is likely bad contact of chips in sockets. I recommend tearing the machine down,use a quality contact cleaner and remove and reseat the chips(not just the scsi chip,all of them). It sounds crazy but it usually fixes problematic A3000's like this.