SE/30 Boot Exception

[tsillay]

Hi All

Firstly, congrats on a brilliant site. There is a mine of information here.

Secondly, I need some help with my trusty old 30.

It failed with a bus error, and upon restart throws chimes of doom and 0000000F 00000033. My son was in the middle of a shufflepuck match and is mighty upset...

Here's my train of thought.

- It seems to be completing memory test, as screen greys out, it waits a requisit amount of time appropriate to the memory fitted before throwing error

- It isn't initialising ADB, SCSI or Screen as the mouse pointer doesnt appear and can't init PRAM

- I'm deducing that $#33 is error 51, unserviceable slot interrupt.

- Although it has an ethernet card, removing this makes no difference.

- I've pulled pram battery to kill off any likely corruption in there.

- Video decrlom seems ok, I've swapped with a spare

- caps have been done, board cleaned, et al. there's no visible nasties, I've checked traces around anywhere caps lived.

So:::: Plan is to throw the scope across the interrupt lines around via2, see if I can catch something moving there.

What sage advice can you share?

- Is my logic around 'unservicable slot interrupt' correct?

- What is the actual boot check sequence, eg, does it do video before adb before scsi therefore i'm potentially chasing an onboard video/slotE problem not something later in the sequence?

- Could something related to the ethernet card be stuck in the pram and pulling the battery hasn't got it clear?

- Has anyone seen this error before in a situation where no expansion cards exist?

Thanks in advance

Tim

 

[uniserver]

did you wash the board? then dry it off with the air compressor.

there might be cap goo lurking under some of the I/C legs.

also make sure to remove and re-insall any removable I/C's including PAL and VROM.

Another thing you can do really quick is get a meter on PSU Voltages make sere they are kosher

also, what kind of caps did you install?

 

[tsillay]

Yep, gave it a good clean with aerosol PCB cleaner + a decent blasting from the compressor.

Used D and C package SMD tantalums for the recap and gave all the removeables a good reseating along the way.

Voltage looked good at 5.02

 

[tsillay]

finally got some time to poke around with the scope.

Sure enough, it's firing IRQ6, the Video interrupt, then throwing the exception.

So I'm speculating it's getting to the point where it clears vsyncen, which lets the vsync signal though UI6 pal, which turns into IRQ6, which goes through the GLUchip, VIA2 throws the IRQ line, but the vsync interrupt handler isn't there...

From what I can figure, the vsync handler also deals with reading ADB and moving the mouse pointer which explains the lack of pointer or ability to reset pram from the keyboard.. ADB circuits seem just fine, polling away cheerfully until you hit a key...

Only sign of unusual signals I've found are some stray noise on D22 once the exception is thrown..

I guess it's diagnostic mode from here and knockup a serial I/F to something..??

Has anyone had a situation where the PAL's go 'soft' in the '30 and the timing goes astray??

 

[tsillay]

Writing things down helps sometimes.

I was wondering if it would boot if I could somehow disable IRQ6.

I pulled vsyncen (pin 5) out of UI6 and hey presto, it's away.

Of course with nothing to drive the vsync manager there's no mouse movement or keyboard action but it boots into system 755 cheerfully now from SCSI....

Very odd indeed...

 

[uniserver]

interesting…

i think you are hot on the trail.

I have a hunch there is a solder bridge some where with your cap job.

not to sound mundane or anything.

 

[tsillay]

Usually I would concur thoroughly. However this one gave up the ghost prior to the cap job. It was a nice clean board but the sound was going and figured I'd recap before the troubleshooting started. It could mean there's some as yet undetected cap goo evil lurking (also mundane).

The other interesting thing is I discovered I can reconnect vsyncen after we're all booted and the keyboard works. Mouse button also works, pointer didnt seem to move but I was walking out the door to the office so I'll explore more tonight...

I wonder if the default vertical refresh manager gets patched out during the 755 system load, and there's a funny in the one that loads from rom initially....

 

[uniserver]

well i'm thinking the cap job went good.

but as the board heated up afterwards is when it started messing up.

maybe slight expansion after it was on for a bit caused a short?

somewhere?

 

[tsillay]

Harumph

So with vsyncen pulled, it boots wonderfully into 755 from internal SCSI. ADB works, keyboard works, mouse button works but mouse won't move.

With a soft reboot, you can then pram zap, extensions off etc.

Reconnecting vsyncen once it's booted makes no difference, it doesnt chuck the slot exception, it doesnt make the mouse work again... hmmm...

I can see the vertical sync IRQ(6) gets all the way through the pals, via2 and to the GLU as VIAIRQ2...

The VIAIRQ2 signal looks quite dirty to me, there's a volt of noise on it. By comparison, VIAIRQ1 looks clean as a whistle.

Does someone have the ability to look at VIAIRQ2 on a working SE30 for me?

Does anyone know....

- Does the vsync interrupt handler get loaded at boot time from the on board video declrom as part of the whole slotE psedoslot thang?

- Or does it get loaded from the main romsim?

 

[uniserver]

you should check this and make sure this filter is good.

i have had this thing die on me and screw up the keyboard and mouse.

Causing other strange boot-hang type issues as well.

 

[tsillay]

Okey Dokey.

What's inside it?

Is this the famous Bournes Filter full of C's and R's?

 

[uniserver]

i think its something like that.

 

[tsillay]

Got back to this recently. I convinced myself that the interrupts were working correctly and I have a slot manager kind of problem. Through some distorted logic I decided that putting a cheap PDS iisi video card in would force slot manager and the vertical blanking manager to sit up and take notice of something different...

Long story short, I got the damn thing to boot and bugger me, the SE30 onboard video (slotE) appears in 4 different slots. (E,A,6,2). It forces the poor old thing to think there's 4 screens attached, which has some predictable nasties when you move the mouse pointer or windows off the screen edge...

So it's got to be an address decoding problem for the video declrom.. I just can't find out where the problem is.... yet.... I've swapped all the PAL's and anything else that's socketed...

Who can tell me why microbug only works in 24 bit mode on the 30? The address translation to figure out where the slot memory space lives is seriously messing with my head...

I've ordered one of Dougs romsimm programmers, so I'll put a si rom in there and see what happens..

 

[tsillay]

dah. it's not just the onboard declrom address decode... A PDS video expansion card also appears in more than one slot....