DKB 2632 crash

[SpeedGeek]

The DKB2632 memory is extended memory which configures above 24 bit address space so the 8MB 24 bit/Zorro2 limit is NOT affected by this board! The most likely problem is a voltage drop on the +5 rail when both boards are installed.

Unfortunately, both the DKB2632 and A2630 have some design bugs which make them voltage and heat intolerant (as well as an overclocking nightmare).

... and since, I don't like repeating myself:

http://www.amibay.com/showthread.php?t=33672

[Michele31415]

Interesting thread.  I guess it's time to break out the DVM and check the +5 v. rail.  After all, the power supply is nearly 30 years old.  While we're at it, does anyone know what the current draw of the Spectrum board is?  I've been unable to find that anywhere.

Right now, I'm running the 2630, 2632, 2091 with a 105 meg. Quantum hard drive, the Deneb, the X-Surf, a 2 gig. SCSI drive, an 8 gig. thumb drive in the internal Deneb slot, a CD drive, the 3.5" floppy, and a 5.25" floppy.  Too much?

[Michele31415]

OK, I checked the voltages with my DVM and got this:

+12 v = 12.06
+5 v. = 5.17
-5 v. = -4.89
-12 v. = -12.34

Just for laughs I also scoped each one and they're all flat as a rail.

This was with all of the cards inserted, including the Spectrum.  I let the machine boot up completely and the voltages remained steady.  I think the positive voltages are OK, but I'm a bit concerned about the negatives.  OTOH, I don't know what the tolerences are for these in the 2000.

Are there any voltage adjustments in the Amiga power supply?

[SpeedGeek]

Where did you measure the +5 volt supply? There is significant voltage drop from the PSU connector to mobo to expansion slot. Try measuring it on the A2630 or DKB2632 boards. There should be a small VR in the A2000 PSU you can tweak but if you've got good voltage readings that won't be necessary.

The second problem is the DKB2632 design bugs which result in poor EMI/RFI noise immunity. They are more complicated to solve but you could try moving your expansion boards to the left most slots and rerouting the SCSI cable to see if that helps.

[Michele31415]

Hmm - I hadn't considered that.  I measured the + voltages off an unused Molex connector and the negative voltages right on the mobo connector.  It's still on my bench, so I suppose I can try to pin down (heh heh) some +5 v. point on the 2630 and measure again.

But I'm really thinking now that it's more likely a software issue.  My experience with power problems has been that they cause random crashes even when the machine is just sitting there idling.  This crash only happens when I give the 2632 command and the Spectrum board is in place.

UPDATE:

I measured the +5 v. on the 2632 from a conveniently located bypass cap and got 5.01.  I also checked on the Spectrum board too and got 5.07.

[SpeedGeek]

I have found that the 2632 program has a resident module which seems to be incompatible with 3.x but it only fails on reboot. Its possible (but unlikely) that this resident is not compatible with some older RTG software.

You can test this by booting with no startup-sequence via bootmenu and then run 2632 from the Shell. You could also use 2632 -r to add memory without the resident module.

If that doesn't help try 2632 -m0 which lowers the priority of 2632 memory to the same as auto-config fast memory. It won't be accessed until the auto-config fast memory is used up. This will help determine if the DKB2632 memory is failing itself.

[Michele31415]

Thanks for the ideas.  The first thing I tried was the no startup-sequence boot with the Spectrum board installed.  I gave the 2632 command from the CLI prompt and it instantly locked up the machine.

Next I tried 2632 -r after doing a normal boot.  Instant lock-up.

After rebooting again, I tried 2632 -m0.  Same thing.

Then I tried putting the 2632 command in the startup-sequence right at the beginning.  This made the machine not boot at all (never displayed anything other than a black screen).

So I pulled the Spectrum board and rebooted.  But now it still wouldn't boot.  That's as far as I've gotten today.  Is this perhaps a clue though?  

Even with the Spectrum board out, the 2632 command only works if given from a shell after the startup-sequence has run.  Putting it at the top of the s-s as DKB recommends causes the machine to freeze.  Sort of one step forward and two steps back.  

[TjLaZer]

Have you tried different SIMM modules and also checked the jumpers to make sure you have the right settings set for the sizes?  I had issues like that and I had to do a lot of testing to make some of my 16MB SIMMs work.

[SpeedGeek]

I would suggest posting your s-s but if the 2632 command fails with no s-s then you clearly have problems with the DKB2632 or the SIMMs themself. You can try the 2632 -d command which will give some you info about the memory config and bank size but won't fix anything.

The next step is to clean or replace the SIMMs and re-seat the socket mounted chips on the DKB2632. You can also try a memory tester program with out running the 2632 command.

[Michele31415]

"Have you tried different SIMM modules and also checked the jumpers to make sure you have the right settings set for the sizes?"

The 2632 was fully populated to 112 MB when I bought it.  I checked the jumpers and they're correct.  Changing SIMMs is not an option since I don't have any others.

"You can also try a memory tester program with out running the 2632 command."

OK, I ran memcheck (from Aminet memcheckBH.lha) and got this:

Testing 2632 memory (addresses $01000020-$07FFFFFF)...no errors found.
Testing expansion memory (addresses $00200020-$005FFFFF)...no errors found.
Testing chip memory (addresses $00000420-$001FFFFF)...no errors found.

Ran it two more times for good measure - same results.

I couldn't run memcheck without running the 2632 command first because then memcheck did not see the 2632 memory.  And of course, this is with the Spectrum board out since I can't give the 2632 command at all when it's in the system.

Running the 2632 -d command gives this:

    2632 Resident Memory Utility, Version 2.0

Address 01000000 Size 07000000 Priority 5    Name 2632 memory
Address 00200000 Size 00400000 Priority 0    Name expansion memory
Address 00000400 Size 001FFC00 Priority -10    Name chip memory

I still think it's odd that the 2632 command works fine when issued form a shell but crashes the machine when put in the s-s.  I have no stability problems from the 2632 when it's invoked from the shell.  I run in super hi-res, open bunches of windows, run AWeb, all kinds of stuff for hours without issue.

Would it be worth trying to figure out what line in the s-s file somehow magically lets the 2632 command work without crashing the machine?

I also still have to try the MMUlib idea.

UPDATE:

I installed MMUlib and much to my surprise, it completed normally.  I rebooted, but now what?  The 2632 RAM is not present automatically.  What do I have to do get the effect of the 2632 command?  I browsed the considerable MMUlib documentation which only made me even more confused.

[TjLaZer]

what happens when you COLD boot with the 2632 command in the SS?

[Michele31415]

"what happens when you COLD boot with the 2632 command in the SS?"

Nothing.  I hear a little bit of disk activity and then it just stops.  The screen never lights up.  I haven't tried this since installing MMUlib though.  Do I still need the 2632 command?

[SpeedGeek]

I never tried MMUlib with my DKB2632 but you should be able to edit the MMUconfig file to add non-autoconfig memory with out the 2632 command.

I noticed you have some additional auto-config devices in your Show config report. Is is possible you have some s-s drivers for theses devices which are not compatible with the extended DKB2632 memory?

[glitch]

I was thinking the same thing... perhaps a non-standard library in libs: somewhere too.  Booting off another device with a plain vanilla OS and issuing the commands from there?

[TjLaZer]

I have the 2632 command after SetPatch in my Startup-Sequence.

[cha05e90]

MMULib: You need to build a MMUConfig for your memory configuration. In MMULib/Install should be an ARexx scipt named "BuildMMUConfig" that can do this automatically for you

Please read the  MMU.guide in MMULIb/Docs - the section in that guide you need is called "Installation/Bonus feature for experts/For systems with non-autoconfiguring memory". I found it a very good and understandable read and since then I never looked back for the "2632" command.