Amiga 3000 boot issues

[scuzzb494]

Hi

The Amiga 3000 hard drive finally died. Seems I lost more in the process. I now get the floppy boot options for1.3 and 2.0x and no hard drives. There are two SCSI ancient drives. The rear one gives out a nasty click and refuses to boot. Anyway, if I click either 1.3 or 2.0x I need the special A3000 boot disk which I install and 1.3 boots no problem but 2.04 refuses to boot. I get the 2.0ROM boot screen and the disk starts to load. It goes past the part where it tries to load drives and onto the next section and then halts. When I try to load 1.3 with the 2.04ROM it halts at the point where it says no clock found. I removed the battery ages ago.

So the machine will run 1.3 but not 2.04. I even tried using an external drive.

The computer has an AMAX II+ card and suspecting this I bypassed the floppy connection to the AMAX. The second Conner drive is a MAC drive formatted hard drive and I don't want to muck that hard drive up.

So why is the computer refusing to boot. I did ask nicely and pressed down on the ROM chips. SysInfo under 1.3 shows everything other than the hard drives and does a speed test. So odd.

[Matt_H]

Are you trying to boot a "new" hard drive? I've found that the 1.4 ROMs have very specific/unusual behavior when it comes to booting from floppy vs. hard drive.

First, the drive must have the correct partition layout. The 1.3 partition must have the device name (in HDToolbox) as WB_1.3: and the 2.0 partition must have the name WB_2.x:. The file Devs:Kickstart must exist on each partition in order for the 1.4 ROMs to boot the respective OS version.

The 1.3 partition will be hidden if booting 2.x and the 2.x partition will be hidden if booting 1.3. I semi-remember the partitions being hidden in different ways if you load SuperKickstart from floppy instead of hard disk.

All this is done automatically by the original A3000 Install disk. If you don't have one (it's different from the SuperKickstart disk) it's probably easiest to do this prep on another machine. Remember to also put an OS2.x FFS in the RDB as DOS\01 and set the partitions to match.

I think the reason you're getting the clock error under 1.3 is because the A3000 clock hardware is different from the A500/A2000 clock hardware and the 1.3 SetClock command doesn't know how to handle it. The A3000 Install disk comes with a replacement SetClock for 1.3 that understands the A3000 hardware.

[Pat the Cat]

If the hard drive has packed up then the onboard ROMs are just giving you the initial "Load Kickstart from floppy" options. I guess that means the machine has a ROM tower?

And also, the Kickstart 2 floppy disk has a read error and the Kiskstart loader can't read it in to memory.

This is not too bad, but it looks like your mechanical hard drive, after many years of spinning, has bit the dust.
v
Make a careful note of the ID the hard drive is set too. That would be my first step.

Next, you have to setup a replacement hard drive or scsi2sd or Rascsi. That has to be partitioned just so, in that the contents of the floppy Kickstart disks onto the relevant "Kickstart only" areas. With the right scsi ID number so the motherboard ROMs can detect it and load the Kickstart file into fast memory.

I can't remember whether the A3000 had any non-volatile RAM to store scsi information in. Might have been the A3000T. If the battery backed up RAM stored the settings that might be a problem. I don't think so though, otherwise it wouldn't have ever booted from when you took the thing out.

I'll have a delve and see what I can come up with that's more specific.

Pretty sure size of the hard drive isn't an issue. Could be wrong on that, but usual 4 gig full size 2 gig full compatibiity on partition sizes.

Break out of the boot with ctrl-c and edit, comment out the setclock on the WB 1.3 floppy. Then you need a compatible HD Toolbox to set up the new hard drive.

[Pat the Cat]

Are you trying to boot a "new" hard drive? I've found that the 1.4 ROMs have very specific/unusual behavior when it comes to booting from floppy vs. hard drive.

First, the drive must have the correct partition layout. The 1.3 partition must have the device name (in HDToolbox) as WB_1.3: and the 2.0 partition must have the name WB_2.x:. The file Devs:Kickstart must exist on each partition in order for the 1.4 ROMs to boot the respective OS version.

The 1.3 partition will be hidden if booting 2.x and the 2.x partition will be hidden if booting 1.3. I semi-remember the partitions being hidden in different ways if you load SuperKickstart from floppy instead of hard disk.

Very indepth "how to" in part 1

http://amigax1000.blogspot.com/2019/08/my-amiga-3000-build-part-1.html

Part 3 of the same blog contains the file system type, it's usual FFs mask 0X444F5301 but odd version number (Earlier than 2, V35. 22412 bytes long.

The kickstart partitions are seperate from the Workbench partitions (detais on Part 3 again). God knows why, but I guess it keeps the Kickstart partitions from being taken out by software causes.

Estimated mean time between failure, less than 2 years. The previous hard drive did very very well.

EDIT: Is it confirmed that the dead drive is the drive that had the hidden Kickstart partitions? It might not be so. You could try them both seperately with correct cable termination.

[scuzzb494]

Thanks for the reply which I will study....

Couple of things. There is no ROM tower. Just the two ROM chips. The A3000 Kickstart is working fine I can get to the boot screen no problem. I have tried other copies of the disk and they work fine. The problem starts when the computer tries to boot into 2.04 off the floppy disk. Again I have tried several versions of 2.04 and even the 1.3 WB off the 2.04 boot but doesn't work. There is an issue with loading 2.04. I can load 1.3 from the 1.3 boot screen.

The one drive has a failed motor and the other I think is still working. I may swap them over and see. For now I would be happy being able to boot into the Workbench. There is another issue and that is with the screen when I try to play a game from the 2.04 boot. Here is the blog I just posted on the website.

https://www.scuzzscink.com/amiga/scuzzblog_february21/scuzzblogdfebruary21_2001.htm

I'll read what has been posted thus far and will take another stab at it later

[Matt_H]

The problem starts when the computer tries to boot into 2.04 off the floppy disk. Again I have tried several versions of 2.04 and even the 1.3 WB off the 2.04 boot but doesn't work. There is an issue with loading 2.04.

On a copy of the 2.04 disk, add Set Echo On as the first line of the startup-sequence. Then you'll be able to see where it's getting stuck.

There is another issue and that is with the screen when I try to play a game from the 2.04 boot.

For now, I would attribute this to an 030/cache/OS2.0 incompatibility as opposed to a diagnostic indicator. That may change as we learn more, though.

[scuzzb494]

Sadly the little green corrosion around the Denise is probably the cause of the graphical issues. I was greeted today with a pink screen backdrop and garbled text. I removed the Denise and stripped back to the motherboard and when done I managed to go from pink backdrop to grey and white. And the mouse is hopping over the selection icon for floppy disk and I have had this before and it was the Denise that was broken. Whether this happened at the same time the hard drive popped or earlier don't know. There are two issues at play and I will swap the Denise tomorrow and see how I get on. The SCSI needs another drive.

I had disconnected the SCSI completelyfrom the board and was able to access the early boot but the floppy refused to get past that point in the boot process.  When I rebooted I was in pink screen mode and then grey sadly. She is falling over fast. Hopefully start climbing out the hole tomorrow. Or not. Its fun either way.

[Pat the Cat]

Grey is not so bad on an A3000. Usually means lack of access to boot ROM.

I thought the image for WB_2.x was bigger than the 1.3 image?

If so, the reason why one works and the other didn't would point to either the floppy corrupting the reading, or maybe a faulty fast RAM chip. those ZIP scRAM chips are HUGELY sensitive to static, most easily damaged component I've encountered on Amiga (even worse than CIA chips)

I guess if you had an adf of a super kickstart disk, you could try that on a Gotek? Faulty floppy drive is most likely explanation.

I guess if you got the Amax card and the Mac drive is OK, at least can plug that into different machine.

It could be that WB2.04 doesn't work with the earlier A3000 kickstart. But you did say that games didn't like it either for some reason, caused crashes.

A thought - have you checked the power rails? Perhaps the power supply is dying, taking out hardware with wobbly supplies.

[scuzzb494]

I swapped out the Denise and I am back to the correct graphical display and the mouse is not now hopping over the icons so the Denise fixed. I do sense that the computer may have had a leaky battery at some point dunno. Thats all working now.

I installed the ECHO in the start up and it boots and halts at this line ....

copy ENVARC: ram:env all quiet noreq

This is after the line Makedir ram:T

When I use my own A500Plus boot disk the machine crashes after it tries to create the RAM disk.

Interestingly the computer works fine in 1.3 Workbench. Just been playing Desert Strike on joystick. Also 1.3 has the RAM disk and I can see both CHIP and FAST with SysInfo. There isn't a jumper for loading RAM to the hard drive. Seems a longshot.. scratching my head cus all this happened when the hard drive stopped working.

[Pat the Cat]

Looking more like software problem, incompatible with later Workbench 2.04 floppies than hardware i guess.

Is Envarc: created or assigned at that point? You could try assigning it, maybe difference between your Kickstart ROM disk Oand a 500 Plus kickstart ROM.

Or you could try ctrl-c again and go for a quick n ditry boot;-

Binddrivers
LoadWb
Endcli

If you want to check hardware, memory, adf for a test kit can be download for that, should work in 1.3. It's zipped up.

https://github.com/keirf/Amiga-Stuff/releasest

[scuzzb494]

I have checked all disks I have of 2.04 [ originals ] and they are identical Part Number 367813-01. I also took them into DOpus and all file sizes are the same. They all also have identical startup sequences. 1985-1991. I am not aware of different versions of 2.04. All as Greg Donner Workbench guide http://www.gregdonner.org/workbench/wb_204.html.

The startup creates all the assigns but are prevented from doing so as it appears the process halts at the point previously noted. I suspect RAM failure.

The link given didn't work for me for the memory test.

This machine has the 2000 part numbers for ROM 390629-02 and 390630-02 and not the 3000 03 version. There are no jumpers at J180-181 to select compatibility mode with the 2000 so I am guessing by the time this motherboard was made the 2000 mode wasn't necessary and the chips are correct as there are two and the 2000 doesn't have the two chips.

The only other oddity was a noticable pin snapped off to PIN 3 on the AMAX. This would mirror the vacant pin header for the ribbon to the internal floppy. The pin isn't just removed as there is quite a bit of it left. I sense the break was deliberate. I bypassed the AMEX anyway going straight for the floppy. The 3000 has all its teeth... sorry all its pins on the floppy header unlike say the 1200 or maybe 2000 where 3 is missing.

Remember that this computer was primarily set up as a MAC emulator. The model of the floppy is the Chinon MODEL FB-354 RevE which is identical to the Amiga 500 where there is no Pin 3. So again I think the broken pin 3 is deliberate. The floppy loaded three disks of Desert Strike so I think its safe to say its being read OK.

I don't see a software error as I believe the various original Kickstart disks are authentic as is the Workbench and the ROMS. There may be an issue with the floppy and ribbon and the most likely is bad RAM/memory. I'm gonna go for memory. I've put the computer away now having got the Denise fixed and I'm giving up for now. I think there is little else to do until I can swap out the memory or atleast check. Not a today problem.

Thanks. I need to read my thick 3000 manuals and see how the machine deals with memory I think.

[Matt_H]

This machine has the 2000 part numbers for ROM 390629-02 and 390630-02 and not the 3000 03 version. There are no jumpers at J180-181 to select compatibility mode with the 2000 so I am guessing by the time this motherboard was made the 2000 mode wasn't necessary and the chips are correct as there are two and the 2000 doesn't have the two chips.

I don't think any 3000s ever shipped with the 2000 ROM configuration. I imagine it was something used during development. Vestiges of it were left on the early commercial motherboards but removed on later revisions. If you're getting the 1.4 Kickstart selection screen I think it's safe to assume you have the correct 1.4 ROMs.

There may be an issue with the floppy and ribbon and the most likely is bad RAM/memory. I'm gonna go for memory. I've put the computer away now having got the Denise fixed and I'm giving up for now. I think there is little else to do until I can swap out the memory or atleast check. Not a today problem.

I agree, it sounds like a RAM issue, or at least something having to do with chips rather than floppy drives. The line in the startup-sequence where the boot failed is entirely normal and noncomplicated, so if copying something to the RAM: filesystem causes a crash then RAM is a good place to begin checking for problems. (I thought at first it might have been hanging on BindDrivers and the AMAX board but that doesn't seem to be the case; for testing purposes, though, it's probably best to remove the AMAX.)

Some of the install disks for GVP's 030 accelerators come with a decent RAM check utility. I don't know if they'll work on an A3000, but it's an easy first thing to try. There's also the relatively new DiagROM. I don't know if it can be softkicked by the 1.4 ROMs (i.e., name it Devs:Kickstart or create a SuperKickstart floppy with it) but if it can it'll be an easy way to get very detailed diagnostic info.

In the meantime, I'd suggest cleaning/treating the motherboard on the assumption it's had a catastrophic battery leak. Isopropyl alcohol, etc.

[scuzzb494]

Like you I think this is a RAM issue. Maybe ROM also but I'm leaning to a busted chip. I'm about done for now. I cleaned up like you suggested and when I get some more time with the 3000 I will run a diagnostic on the RAM. I have some spares so may have a go. Workbench 1.3 is running fine which is strange to me but dunno, that's why I need to get my big manuals out for the 3000. Just happy I got the Denise back.

This was the latest blog below with that ECHO message and that chip replacement.

https://www.scuzzscink.com/amiga/scuzzblog_february21/scuzzblogdfebruary21_2201.htm

PS The busted hard drive I used doesn't look too bad. Sadly it has a massive cut across the back face. I bought it broken and for spares. Maybe time has come to rejuvinate the board given the rarity now. Sadly it now has a broken Denise and an ECS at that.

Thanks for the help.

[Pat the Cat]

Corrected link to Amiga test Kit download (1.6 latest at present);-

https://github.com/keirf/Amiga-Stuff/releases

Quite thorough comparisons of all Amiga Kickstart releases and memory maps (A3000 is 2nd column from right, rightmost is A4000);-

https://amigacoding.com/index.php?title=Amiga_memory_map

You can't mix static column ZIP chips and FP (fast page) ZIP chips (1MB x 4 size) so do check any spares to make sure they match the type of what is on there. It won't work. You to have google the part number and "datasheet" to ffind a pdf of an old datasheet that will tell you if it's static column or not (preferred on the A3000 because it's faster with an 030 in burst mode).

And DO earth yourself when touching these chips or you'll blow 'em.

My guess from memory on which one... if you look at the board with chip RAM beneath the fast RAM... the RAM builds up in layers of 2 chips on each row for each bank. The suspect chip would be on the top row furthest from the chip RAM, but I'm not sure on the left side or right side or which position exactly. Probably third from the left on a 12mb A3000, but that's a guess.

It depends how the ROM is mapped, to the very top and end of fast RAM or the beginning. Usually it's the end on an Amiga, hence my guess.

[Matt_H]

Another quick test: Try to fill up RAM: when you boot into 1.3 (i.e., just copy a bunch of files there).

The 3000 puts its SuperKickstart image into regular fast RAM (unlike the 1000 which has dedicated RAM for storing the Kickstart). Even though 1.3 appears to be working, it might simply be because the system isn't putting enough stuff into RAM to hit the "bad" sector. Under 2.0, in addition to a Kickstart that's double the size of 1.3, there's also the copying of ENVARC: to ENV: which doesn't happen under 1.3, plus more use of T:, etc. There's a lot more use of RAM (and the RAM disk), meaning the system is more likely to hit the bad sector.

Filling up RAM: under 1.3 might achieve the same result, helping to narrow down whether you're dealing with a RAM problem or a problem with another chip.