Amiga 600 Ressurection - Cap damage, RAM errors

[telemonster]

Hello all!



With the change in social activities I decided to take a crack at a long overdue project. Years ago a friend gave me his Amiga 600, and I cherished it. It started to smell bad, and I thought it was one of the rubber feet -- it turned gooey. As everyone (except me) knew, capacitor electrolyte damage.



I investigated it, and decided it was too tough to deal with so put it aside. I planned to mail it off to have someone else repair it, but never made it to FedEx.


So now I'm taking a good hearty crack at it.


Current state on workbench, black screen never wakes up.


There is a good amount of damage from the leaky caps.


Currently I've replaced most caps, except some are still missing. I do not have C821 or C822, which are 470uF between GND and the +12v and -12v rails.

I am also missing C239 which should just tie to the RF modulator, and C214 which I think is mono audio to the RF modulation.



I have one bad VIA thru the board over on the right side that I believe is audio, so shouldn't matter (cap 303 maybe, or 304.)


I get clock on pin 15 of the CPU, and reset is high on pin 20 of the CPU.


So after black screen woes, I cut a DiagROM chip and dropped it in. I had great fun with it trying to repair an Amiga 500, and it was a help though I never got it working.


Soooo very interestingly and right off the bat I get:


- Test of writing word $FFFF to $400 OK
- Test of writing word $00FF to $400 FAILED
- Test of writing word $FF00 to $400 OK
- Test of writing word $0000 to $400 FAILED
- Test of writing byte (even) $ff to $400 failed
- Test of writing byte (odd) $ff to $400 OK


Then tons of RAM errors which are obvious.


The system worked fine before the cap damage. Is there some kind of chip select or bank select that might be at play? The motherboard has two RAM ICs on it. I haven't prodded the bus or anything yet.


It repeats similar errors every boot.



Any thoughts?






----


Two notes - my troubleshooting wouldn't have been possible so far without two awesome resources:
- http://www.amigapcb.org/ (Amazing visual guide to the PCB, mine is a different revision but it's close enough!)
- www.amigawiki.org - Has schematics

[telemonster]

Started to do more digging.


Clipped in caps to replace the two 47uF that sit on the +12/-12 rails. No different.


Looked through the board more in the damaged areas. Damage is mostly around power supply in/ RF out, the middle where the two +12/-12 caps sit, and audio sections.


Looked at schematics a lot. Agnus handles DRAM refresh.


Checked the clocks from agnust, resistor to left of agnus hmming at 28MHz., then some 7.xMHz from the resistors above agnus (and some clocking slower.)


Checked for activity in and out of U21 and U22, the two that sit on address bus between the Mot 68K and Agnus. Nothing stuck high or low.


Anyone understand the Amiga RAM layout enough versus the diagrom output below?

Garbage before this text was binary numbers: 1, 2, 4, 8, 16, 32, 64, 128, 240, 15, 170 and 85
To help you find biterrors to paula. Now starting normal startuptext etc
Amiga DiagROM V1.2 - 13-Aug-19  - By John (Chucky/The Gang) Hertell


Testing if serial loopbackadapter is installed: <> NOT DETECTED
- Parallel Code $ff - Start of ROM, CPU Seems somewhat alive
    Checking status of mousebuttons for different startups:

    Set all Interrupt enablebits (INTENA $dff09a) to Disabled: Done
    Set all Interrupt requestbits (INTREQ $dff09c) to Disabled: Done
    Set all DMA enablebits (DMACON $dff096) to Disabled: Done

Testing if OVL is working: OK
- Parallel Code $fe - Test UDS/LDS line
  - Test of writing word $FFFF to $400 OK
  - Test of writing word $00FF to $400 FAILED
  - Test of writing word $FF00 to $400 OK
  - Test of writing word $0000 to $400 FAILED
  - Test of writing byte (even) $ff to $400 FAILED
  - Test of writing byte (odd) $ff to $401 OK
- Parallel Code $fd - Start of chipmemdetection

Addr $00000400
       Write: $AAAAAAAA  10101010101010101010101010101010
        Read: $FFAAFFAA  11111111101010101111111110101010  FAILED

       Write: $55555555  01010101010101010101010101010101
        Read: $FF55FF55  11111111010101011111111101010101  FAILED

       Write: $F0F0F0F0  11110000111100001111000011110000
        Read: $FFF6FFF6  11111111111101101111111111110110  FAILED

       Write: $0F0F0F0F  00001111000011110000111100001111
        Read: $FFFFFFFF  11111111111111111111111111111111  FAILED

       Write: $0F0FF0F0  00001111000011111111000011110000
        Read: $FFF6FFF6  11111111111101101111111111110110  FAILED

       Write: $00000000  00000000000000000000000000000000
        Read: $FFFEFFFE  11111111111111101111111111111110  FAILED
Addr $00008400
       Write: $AAAAAAAA  10101010101010101010101010101010
        Read: $FFAAFFAA  11111111101010101111111110101010  FAILED

       Write: $55555555  01010101010101010101010101010101
        Read: $FF55FF55  11111111010101011111111101010101  FAILED

       Write: $F0F0F0F0  11110000111100001111000011110000
        Read: $FFF6FFF6  11111111111101101111111111110110  FAILED

       Write: $0F0F0F0F  00001111000011110000111100001111
        Read: $FFFFFFFF  11111111111111111111111111111111  FAILED

       Write: $0F0FF0F0  00001111000011111111000011110000
        Read: $FFF6FFF6  11111111111101101111111111110110  FAILED

       Write: $00000000  00000000000000000000000000000000
        Read: $FFFEFFFE  11111111111111101111111111111110  FAILED
Addr $00010400
       Write: $AAAAAAAA  10101010101010101010101010101010
        Read: $FFAAFFAA  11111111101010101111111110101010  FAILED

       Write: $55555555  01010101010101010101010101010101
        Read: $FF55FF55  11111111010101011111111101010101  FAILED

       Write: $F0F0F0F0  11110000111100001111000011110000
        Read: $FFF6FFF6  11111111111101101111111111110110  FAILED

       Write: $0F0F0F0F  00001111000011110000111100001111
        Read: $FFFFFFFF  11111111111111111111111111111111  FAILED

Looks like everything is mostly getting stuck on.


Thoughts?

[solarmon]

A common issue with the A600 is a problem with the reset circuit caused by cap damage.

Usually, cap C612 has leaked and stops the reset circuit from working, because of itself being degraded or it has caused corrosion to surrounding components and traces.

Remove it with appropriate tools and methods and it might boot up again.

[telemonster]

Originally Posted by solarmon

A common issue with the A600 is a problem with the reset circuit caused by cap damage.

Usually, cap C612 has leaked and stops the reset circuit from working, because of itself being degraded or it has caused corrosion to surrounding components and traces.

Remove it with appropriate tools and methods and it might boot up again.

Thanks for the reply. The reset circuit would prevent the computer from ever starting right? In my case the DiagRom is starting, it's just filled with RAM errors. Seems like errors either writing or reading data,


The original black screen I think was due to the RAM errors being so deep? Thought I could be wrong. With diagrom in place I do see red and green out of the video circuitry so it seems like that is working.

[telemonster]

Replaced C307, no change

Checked data signals from 68K, all of them show activity. Nothing odd (like, nothing is held high or low.)

Sniffing around the damage of the +12/-12 caps, I tried a jumper to carry the CCK signal over the rough area, but made no difference. There is a pull up resistor as well but that is for the XCLK signal that feeds out the video DB23. All of them test okay with a multimeter but I was thinking at high speed data maybe it would cause noise. Nothing there.

[Pyromania]

Maybe this machine is beyond repair? Chipset could be used as a donor to a new A600 PCB perhaps?

[telemonster]

Originally Posted by Pyromania

Maybe this machine is beyond repair? Chipset could be used as a donor to a new A600 PCB perhaps?

The more I work on it the more it seems not that bad damage wise. In the middle most of the damage is concentrated on the two pads, near the power input I managed to work around any damage and then the audio stuff is TBD (I'm a huge game music fan so the audio will get fixed.)


The problem with the RAM errors seems like maybe there is a control line issue somewhere that is off, and I don't know enough about the low level architecture stuff to know what's up.


The A600 PCB was never remade correct? I would be all for trying to migrate the chips if there was a replacement PCB that was pre-populated (pick and place) with all the common stuff and maybe sockets for the main ICs or pads, but wouldn't want to try to populate all the SMD stuff on a blank PCB. I'd give a A3000 repro a go that is all thru hole but the 600 is compact.


The bottom of mine looks flawless. I will take some photos and post them.


It's a challenge. I like bringing things back from the dead. I just hope none of the custom chips are bad.


I had a 286 10mhz Yamaha midi laptop that it took me 3 years to figure out :-)

[psoma]

Nah, if you can get diagrom running... you can repair it.

It's time to get out the multimeter and trace/check the DRD and A lines (data and address) between the 68000 and Agnus and then Agnus / transceivers (U21 & U22) / RAM.

Looking at your symptoms, my gut feel is a bad transceiver... but trace it out first and make sure you have continuity at the key points.

[telemonster]

Psoma - Thanks! Will take that on tonight! I checked activity on the transceivers and saw data flow in and out (nothing being held high/low) but that doesn't mean much. I should trace out the connection to them as well to make sure that area is clean.

Here are some images of my Amiga 600:

https://imgur.com/a/YFa0IOF

Thanks

[r.cade]

Get that corrosion neutralized and cleaned up and then take some more pics of that area above the Agnus. That doesn't look too good to me and you have have some bad traces and vias.

[telemonster]

Originally Posted by r.cade

Get that corrosion neutralized and cleaned up and then take some more pics of that area above the Agnus. That doesn't look too good to me and you have have some bad traces and vias.

Okay, hit it with some white vinegar just to be safe -- then rinsed it off. On my A2500 when I was taking care of the battery damage I did white vinegar then baking soda then cleaned it all off, but this A600 -- the caps have been removed for around 3 years now (My project queue is deep between old computers, music stuff and arcade/pinball stuff.)


I have multimetered through all of the VIAs that I could tell (using http://www.amigapcb.org/, which is a lifesaver for this) and all of them above the agnus seem to be okay. I will double check those pull down resistors that are on the clock lines and stuff in a bit.


I just added a jumper wire on the back of the board between two of the VIAs just in case the through hole was generating noise. This is the one on the CCK line.



The other two that have corrosion (but multimeter okay.. but could be adding resistance) are:


R253 - Pull up between VCC and XCLK, which is destined for the DB23 video connector


The other VIA that is pretty corroded is the CCK signal again. Last night I used two meter probes tied together and jumpered across that one to see if it would pass the RAM test. No go there.


This is fun! Much thanks for the help everyone.

[solarmon]

Originally Posted by telemonster

Thanks for the reply. The reset circuit would prevent the computer from ever starting right? In my case the DiagRom is starting, it's just filled with RAM errors. Seems like errors either writing or reading data,


The original black screen I think was due to the RAM errors being so deep? Thought I could be wrong. With diagrom in place I do see red and green out of the video circuitry so it seems like that is working.

If it was memory issues then it is normally a green screen, with a normal Kickstart chip in. You mentioned that it has a black screen. I would check the reset circuit to rule it out, if you haven't already done so

[cpiac64]

verify with amigapcb connection from first ram to second

verify cas and ras

[psoma]

So diagrom will run in ANY chip mem (or FastMem) - if you have access to an expansion board, throwing it in might get diagrom to launch so you can add a few other vectors to your diagnostics...

[telemonster]

Originally Posted by solarmon

If it was memory issues then it is normally a green screen, with a normal Kickstart chip in. You mentioned that it has a black screen. I would check the reset circuit to rule it out, if you haven't already done so

I looked at it a little bit. I see the reset pin on the M68K high when the system is running.


DiagROM comes up and runs through memory test, would the reset circuit prevent this?



Note - DiagROM immediately starts seeing errors, look at the log above. I wonder if the memory function is too weak for Kickstart to green screen?


There is definitely corrosion in the upper right where some of the reset circuit is. I can double check it tonight. I was thinking that if the reset circuit was bad the DiagROM wouldn't be able to kick off but I could be totally wrong.



Okay -- looking at the schematic, the 555 timer that drives that power on Reset (saw another thread about this) is tied to pin 63 on U5 / Gayle, that is different than the IC reset line that I checked. I will check this tonight! Maybe this is something that is Kickstart specific.



Thanks, will report!

[telemonster]

Originally Posted by psoma

So diagrom will run in ANY chip mem (or FastMem) - if you have access to an expansion board, throwing it in might get diagrom to launch so you can add a few other vectors to your diagnostics...

Unfortunately I only have the base system. It was stock when my friend gave it to me 10+ years ago.


To the other post regarding RAS and CAS, I will poke at that again tonight. I can verify traffic, I think I saw activity at the RAM ICs when I checked before but I will do so again! Thanks

[demolition]

Originally Posted by telemonster

I was thinking that if the reset circuit was bad the DiagROM wouldn't be able to kick off but I could be totally wrong.

That is correct. That you get output from diagrom means that the reset circuit is fine and the cpu is running code from the ROM. The typical error I see in acid-damaged A600s are broken vias. They tend to get hit by the acid so if you can see some that look dull, check that they conduct as they should.
The errors that you see tells me that some memory control signal is missing so check that you can follow RAS/CAS/RW from Agnus to each RAM chip. It could also be the A/D-bus since a write to the wrong address could also get results like that.

And btw, vinegar is acidic just like the electrolyte that leaks from the capacitors, so it is actually not very useful in neutralising that. Vinegar is suitable for NiCd battery (as used with RTCs) leakage since that would be an alkali. No matter if you use vinegar or not, always finish by rinsing it with plenty of dimineralised water to push it towards a neutral pH.

[telemonster]

Originally Posted by demolition

That is correct. That you get output from diagrom means that the reset circuit is fine and the cpu is running code from the ROM. The typical error I see in acid-damaged A600s are broken vias. They tend to get hit by the acid so if you can see some that look dull, check that they conduct as they should.
The errors that you see tells me that some memory control signal is missing so check that you can follow RAS/CAS/RW from Agnus to each RAM chip. It could also be the A/D-bus since a write to the wrong address could also get results like that.

And btw, vinegar is acidic just like the electrolyte that leaks from the capacitors, so it is actually not very useful in neutralising that. Vinegar is suitable for NiCd battery (as used with RTCs) leakage since that would be an alkali. No matter if you use vinegar or not, always finish by rinsing it with plenty of dimineralised water to push it towards a neutral pH.

Much thanks for the info, will flush the board!



Okay back at it.

At work before I got home I had an epiphany that maybe it's VIA damage near cap C612. I replaced it with an aluminum SMD but didn't look at the VIAs closely since it was in pretty good condition. After checking them all, it all looks pretty mint in that area.


Using the resistor clump (R151A-H and R152C) I easily see data on the pins that handle row and column addressing on the RAM chips. I see data on the address and data pins on the RAM chips as well.

Since most of the ram errors seem like everything is stuck in a high state, I checked some of the grounds on Agnus and they're all fine.

A friend called me today talking about some arcade hardware stuff, and while I was giving him advice on troubleshooting a MK2 and a Wrestlemania, it dawned on me I should try my own advice. I started to measure the 5v on the board, around the board. Mostly using the noise filtering caps near most ICs.

I stumbled into cap C3 which was reading low, 4.7vdc or so. It feeds power to U3 (Paula.) But it looks like that sits past a 1ohm resistor so this might be on purpose. Still investigating this one.

[telemonster]

Originally Posted by telemonster

I stumbled into cap C3 which was reading low, 4.7vdc or so. It feeds power to U3 (Paula.) But it looks like that sits past a 1ohm resistor so this might be on purpose. Still investigating this one.

Looks like that is on purpose,

[psoma]

Hah, yeah, I thought I stumbled onto something with that one too but found the same thing. It's really handy to have a working A600 nearby to sanity check