Dead Amiga 3000 - can someone test my Agnus please?

[spirantho]

Hi everybody,

I have here a not-very-well Amiga 3000.
Its screen likes to go any colour you want, as long as it's green. It's always green. Only green.

When I got this machine in, it started up - then after a few more boots while I tried reconnecting hardware, it just went back to the green screen (which is why I had it in to repair).

I've changed Agnus's socket (that was fun!). I've checked the Chip RAM... but I've found the primary culprit - the _RESET line. Not the _IORST line, as that's triggered by the _RESET line. Not the _CPURST line - that's fine too. Just the _RESET line. The pull-up resistors are good, there are no short circuits. The 7407 buffer and the 7432 OR gate (U712/U713) are verified (and changed just in case).
The only other chip on that bus is Agnus. Now, Agnus takes the _RESET line as an input, not an output, so my understanding is that if Agnus is holding the _RESET line active (low) then something's wrong.

The green screen, of course, is because Agnus's keeping the _RESET line active is making the _IORST line active too. This stops the chips like the CIA working, meaning the Kickstart ROM never pages out, so GARY never asserts the _RAMEN line (and it wouldn't matter anyway, because AGNUS is just twiddling her thumbs in a reset state). Because RAM is never enabled, the 68030 does a "memory test" without any memory, and kicks up a green screen.

So - my prime suspect is the only remaining chip on the _RESET line, which is AGNUS.
Can anyone test my 8372AB Agnus? Or even better could I borrow a known good one? Or does someone have one for sale?

Thanks for any help!

[bloodline]

Where are you based?

[spirantho]

Miles away from anywhere, in Aberystwyth, West Wales.

I actually just tried an A500's 8372A Agnus in the A3000 and it did exactly the same thing. Even though I'm not sure it'll work, it should do enough the same to suggest the Agnus is fine. I'll have to try the 8372AB in the A500 next.

But why would Agnus be holding the _RESET line low? It pulses up very briefly but then immediately goes low and stays low. Does Agnus control the _RESET line? I'm not aware of it doing so....

Very puzzled by this one. Any hints, anyone?

[bloodline]

Lol, I'm not heading to Wales anytime soon! :-)

Ok, it seems like there is a signal that Agnus is waiting for (and not getting) to continue the bootstrap. Anyone know the sequence for the ECS chipset?

-Edit- could Agnus be waiting for the chipram? Could that have failed?

[spirantho]

Thanks for the ideas - and (*looks out of the window at the weather*) I don't blame you.

I don't think it can be the chip RAM because Agnus is the chip RAM controller, and as long as the _RESET line is low, chip RAM is out of the picture entirely.

The _RESET line is high for 0.1us only - that's not even one 7MHz cycle - so it can't be something that Agnus needs to do, I don't think. Does Agnus need to be "turned on" somehow?
But this still doesn't make sense - the _RESET line is an input, not an output! A chip should never hold an input low itself.

Edit: For those that are wondering, the +5V line is held at 5.09V.

[bloodline]

A failed pull-up on that line somewhere, would manually pulling the signal high with a high resistance resistor cause any damage? Might be worth a go to see if that continues the boot?

[spirantho]

The pull-ups are fine. I removed the 470 ohm resistor bank RP701 and the 7407 and 7432 chips. Resistance was 1K to +5V, which is correct as RP610 is a 1K resistor bank on the other side of the daughter board socket.

It's crazy...!

Edit: Good news.. kind of. The Agnus is known good - I just stuck it in an ECS A500 and it booted.

Now, what can the A500 tell us about the boot sequence....

Edit 2:
Interestingly, there's a short pulse of _RESET high, and then the _RESET line is held low for half a second, then it goes high.
The problem is that a) This is a KS1.3 machine, and b) the _RESET line is shared on the A500, whereas the A3000 splits them into _CPURST, _RESET and _IORST.
Still interesting, though...

Edit 3:
I found the problem! Now I just need to fix it...

When I removed the backplane daughter board, it still gives a green screen. It used to give a yellow screen. But the important thing is that now the LED flashes when it reboots, like a Guru should! I found a partially shorted circuit between pin 53 (_RESET) and the pin 51 next to it on the backplane, which has a pull-DOWN resistor. When I fluxed and reflowed the solder joints on all the pin 53 and pin 51, the short is now complete! This is sending the _RESET line to ground all the time, which is stopping the Agnus from running, which is in turn stopping everything except the CPU running!

I'm not sure why it's now giving a green screen without the backplane, though - it should be yellow, I think... but I'll try putting back the original RAM chips. Maybe there's a fault in one of the (different) RAM chips I fitted. I hope so - don't want to re-do that Agnus socket! Next thing is to get the screen to go back yellow without the back plane, then I need to isolate the shorted pins, and then Hurrah!

[mechy]

Thanks for the ideas - and (*looks out of the window at the weather*) I don't blame you.

I don't think it can be the chip RAM because Agnus is the chip RAM controller, and as long as the _RESET line is low, chip RAM is out of the picture entirely.

The _RESET line is high for 0.1us only - that's not even one 7MHz cycle - so it can't be something that Agnus needs to do, I don't think. Does Agnus need to be "turned on" somehow?
But this still doesn't make sense - the _RESET line is an input, not an output! A chip should never hold an input low itself.

Edit: For those that are wondering, the +5V line is held at 5.09V.

I have repaired many green screen 3000's(mostly 3000t's) with bad chip ram, seems many of the toshiba zips used in the 3000T's went bad for some unknown reason.i am not sure this is the case with what you are seeing. If you had other colors etc. i would suspect the 3000's PALs. if memory serves the 3000 will boot with only 1Mb chip also.
Also, are you sure you didn't damage a trace to agnus when you replaced the socket? you might do a continuity check with a meter from the socket.

[spirantho]

Sorted!

The green screen was caused because this Amiga 3000 for some reason required all 2MB of chip RAM to be fitted. Not sure why, but there we go - I thought it would run with 1MB.
Once I fitted the other 1MB, the yellow screen came up again, which is correct without a Zorro backplane.

So, I attacked the backplane by desoldering all of the pin 53 and pin 51 pins on the connector sockets. Once I did this, pin 53 and pin 51 were no longer shorted, so I knew the fault was on one of the pins, not on one of the tracks. I then re-soldered pins 53 and 51 one by one until the short re-occured. I found that the offending pin was pin 53 on the socket next to the motherboard - then all I needed to do was to shove the soldering iron up between pins 53 and 51 by the edge connector. It melted the plastic bit of the connector slightly (I have a fine-tip iron, so not too bad) but it duly flowed the solder back onto each pin respectively! No short!

Plug it back in, and this is the result!



So: the moral of the story:

If you get a green screen on an Amiga 3000, it may not mean that the Chip RAM isn't working, it may mean that the entire chipset is prevented from working! You can tell this because the LED doesn't flash. Denise doesn't have a _RESET line, so will still work, hence the green screen - but without Agnus there's no LED flashing - just a green screen and then reboot, then green screen, then reboot.

If the _RESET line is held low (active) on the A3000, the CPU will continue running normally, because the _RESET line comes after the _CPURST line on the reset buffering circuit. This is how come the Kickstart ROM was executed perfectly even though a reset line was active, and how Ctrl+Amiga+Amiga (_KBRESET comes from Gary and comes even before _CPURST).

With no backplane inserted, the screen will go yellow if all Chip RAM is good, it will go green if not.

Maybe this information will help someone in future, who knows...?

Expect this A3000 to feature prominently on YouTube videos shortly....

Edit: Forgot to mention; thanks for the pointers, those who helped

[Plaz]

Nice troubleshooting. Just adding to your experience, my 3000 once suffered a similar problem because of solder problems at the daughter card slot. When I got the yellow screen to respond properly it was just a matter of getting all the remain bit back in to boot up.

Plaz

[mechy]

nice detective work there. Good to see another amiga alive and kicking. Looking forward to the videos.