Amiga A2620 troubles -> no autoboot from HDD?

[HanJammer]

Hi,
new member here, so a quick hello to all users of EAB.

First post, and I have question already...

I bought A2620 for one my A2000s - card works OK, passes RAM test, CPU works OK. Problem is it won't boot AmigaOS from HDD. It doesn't even see the HDDs as boot devices (SCSI HDD on A2091 and IDE HDDs on Buddha).
When I switch back to 68000 via right-click powerup menu it will boot just fine from the HDD. Controller cards are visible in the SysInfo when using A2620.
Kickstart is 3.1. Machine has 2MB of Fast on A2620, 2MB of Fast on A2091 and 4MB of Fast on 3-State memory card (more memory is populated on the card, but I can't use it after installing the A2620). It seems A2620 has latest Rev 7 ROM (but I'm not 100% sure it does).

I know A2620 has a jumper to allow Amiga Unix autoboot, but I don't think it's necessary for Amiga OS.

What I'm missing here?

[Jope]

The A2091 should definitely be compatible. Have you tried it with the HD controllers one by one?

Just insert the turbo and one controller at a time and see what happens.

[Matt_H]

+1 for Jope's suggestion.

Other ideas:
The SCSI controller is still showing up in 020 mode but the volumes are not. Is the SCSI bus known to be stable? e.g., properly terminated, all devices jumpered correctly, 2091 jumpered correctly. In other words, I'm wondering if there's a bus fault that's only triggered in 020 mode.

Might also be worth toggling the Unix jumper in case the system is getting confused. Unix won't boot in 000 mode but if you're in 020 mode and the system is expecting to boot into Unix and there is no Unix... maybe it doesn't know to fall back to AmigaOS? I've never understood the exact mechanism of how that jumper worked, but if its methodology is to disable partitions in the RDB with an FFS DosType (thus leaving only Unix partitions accessible), that could explain your symptoms.

[HanJammer]

Originally Posted by Matt_H

+1 for Jope's suggestion.

Other ideas:
The SCSI controller is still showing up in 020 mode but the volumes are not. Is the SCSI bus known to be stable? e.g., properly terminated, all devices jumpered correctly, 2091 jumpered correctly. In other words, I'm wondering if there's a bus fault that's only triggered in 020 mode.

I think it's properly terminated, and didn't gave me any problems in the past.
I also have another SCSI controller (in fact two - GVP one and 2090). But the problem is the IDE drive volumes don't show up in the boot menu as well...

Originally Posted by Matt_H

Might also be worth toggling the Unix jumper in case the system is getting confused. Unix won't boot in 000 mode but if you're in 020 mode and the system is expecting to boot into Unix and there is no Unix... maybe it doesn't know to fall back to AmigaOS? I've never understood the exact mechanism of how that jumper worked, but if its methodology is to disable partitions in the RDB with an FFS DosType (thus leaving only Unix partitions accessible), that could explain your symptoms.

I will try yours and Jope's suggestions tomorrow.

I also have another A2000 (in fact two of them) if it may help...

PS: At one point in time (few months ago) I had 8 or 9 A2000 - LOL? 3 still remain but one of them has to go...

[HanJammer]

OK, played around some more today:

First A2000 (the one I described earlier)
I removed the IDE controller and the 3-state memory card but it didn't helped. Same issue remained. No HDD boot once the accelerator was enabled. Also with J304 jumper on the card would drop to the CPU selection screen right after the power on and the mouse wouldn't work as well.

Second A2000 (which I bought some time ago in near mint condition with GVP HC8+ and Vortex GoldenGate bridgeboard):

So I moved the A2620 into the second A2000. Machine wouldn't boot at first, but it was likely due to memory conflict with GVP HC8+ (I had 4MB of Fast installed on this card and the A2620 has 2MB). Perhaps same issue as described here: http://eab.abime.net/showthread.php?t=78829 so I was aware of the problem.
So I switched it for A2091 from the first machine and it started working right of the bat. Machine is booting from the HDD and the accelerator works as well. I moved the 3-State Fast RAM card into this machine. Everything works fine. 2MB of Fast on A2091, 2MB of Fast on A2620 and 4MB of Fast on 3-State card are all visible.

Cards installed in the second Amiga:
https://imgur.com/zAEQZM5

The Workbench running off the HDD with A2620:
https://imgur.com/O6ZQqNc

I populated the GVP HC8+ with SIMMs and Installed it in the first A2000 and it works just fine. Perhaps it's also a better SCSI controller than A2091? Isn't it?

One thing I forgot to mention - First A2000 when I bought it in 2020 had some problems with Zorro slots not working which I'm pretty sure I fixed. One of the TTL LS245 logic chips between the CPU socket and Zorro slots is broken - I traced it using motherboard blueprints and detected it's bad using logic analyzer. Being lazy I piggybacked it with a working chip a year ago or so and Zorro slots were working fine ever since (I didn't had any problems with the Zorro cards in this machine - A2091, Buddha IDE and 3-State memory card all were working beautifully). But maybe A2620 is more sensitive to timings or some issues the piggybacking is causing? Or maybe something else is broken? But again - it's a bit strange that the A2620 is working fine in this machine (and the Zorro cards were visible and working as well - because Fast RAM and it's tests were working fine with the A2620 enabled) but it just wouldn't see any HDD as bootable device... What do you think?

Here's the logic chip I mentioned BTW :
https://imgur.com/E16mNXy

Also: I tried to install Vortex GoldenGate in the First A2000 and it wouldn't work - machine wouldn't even boot with the card installed... The card requires a CPU adapter which I didn't moved from the second machine but it's just a single capacitor between to CPU legs and does it really matter that much? Hopefully I will be able to use this bridgeboard in the second machine which now has A2620 installed... But at the same time I want to install Amiga Unix on this machine... So I would prefer to use it in the First one

And finally: I recently noticed that the first A2000 shows some intermittent instability - especially when I try to switch the Kickstart 1.3 on. Immediatelly after that switching back to 3.1 also causes this issue. Then after rebooting it again it is rock-stable. It was like this before the configuration changes I made today... I have no clue what can cause it (the RAM test in the Amiga Diagnostic Kit doesn't show any problems).

Anyway - I keep the First A2000 for experimenting with stuff and for some mods which are modern. I already have Indivision ECS (which I hate) in this machine, perhaps I will install PiStorm in it (or maybe Terrible Fire?). I also have modern rev. Buddha IDE and use CF card and a Gotek for getting files in and out easier. And the Second A2000 I want to keep period correct...

[Jope]

Piggy backing is not a good long term solution, as the broken chip can still drive the signals to the wrong direction. You can use it to find the fault, but then the broken chip should be removed and replaced with a working one.

It's difficult to say whether it is the source of the problem, but definitely something that you should look into fixing in any case.

[HanJammer]

Just to let you know (and for future reference in case somebody stumbles upon similar problems in the future) - I fixed the problem desoldering faulty 74SN245 and soldering a new one in it's place. HDD boot works without problems now and it seems that other 'quirks' have been cured as well.