Amiga.org
Amiga computer related discussion => Amiga Hardware Issues and discussion => Topic started by: servitus on February 27, 2014, 03:22:08 PM
-
Hi folks
... this is another sad A4000 story. I got it out of my cellar, opened it and recognized that most caps and battery have leaked. The last 3 weeks I have been soldering almost every evening. I replaced all the caps and most of the IC's surrounding the leaking caps. After that, I powered up and nothing happened... after reading some stories of others with similar problems, I re-soldered the video interface, all the solder joints on Lisa and some cold soldering joints on my A3640. All of a sudden, it booted to the disk-screen (with disk animation). I thought "wow, cool I got it" ... but now I'm facing the following strange behaviour:
Testing-Scenario 1:
Setup: Keyboard, Mouse, A3640 and Chip-/Fast-RAM
=> it boots to the disk-insert screen, then I insert Workbench 3.1, it loads the workbench and as soon as I double-click the workbench icon the system crashes and I get a red "software-failure" message:
Error: 8000 0004 / Task: 0780F1F4
Testing-Scenario 2:
Setup: Keyboard, Mouse, A3640 and only Chip-RAM
=> the screen goes dark-gray, light-gray, flickering white cpu-line across the screen, waiting for some seconds, then I get directly the following "software-failure":
Error: 8000 0004 / Task: 00003318
What might be the problem? I already tried another SIMM for Chip-Mem, same result. I measured voltage between GND and the 5V line and I got 4.91V with the setup described above. I already cleaned the CPU/MB connection and the power supply plugs on both sides. The CPU card is seated well. There should no short circuits exists between MB and chassis.
Any ideas how to proceed? Any help is highly appreciated, since I want to bring that lovely Amiga back to life :banana:
-
Possibly a Chip RAM problem...? If it works ok with Fast RAM (at least at first) then it'll be using less Chip RAM. Do a memory test on the Chip RAM (there are various memory test programs on Aminet) - and Fast RAM too for that matter. It's possible one of the addressing lines is dodgy.
Have you tried booting a game?
-
Unfortunately I don't have a reference machine to test the Chip RAM and the A4000 crashes as soon as I want to open any folder or program on the workbench... But I have plenty of other SIMMs which I'll try after work.
You're right, possibly the lines under the Chip-RAM sockets are dodgy. I'll solder out that socket to have a better look at them. But honestly said, the Amiga worked more or less stable before I began to change the IC's and caps - except it sometimes didn't identified all of the 16mb Fast-RAM - that was initially the reason why I began to replace all the elements... Which chips are involved in memory management?
I'll try to boot a game as soon as I am at home and will report what happens...
Btw: When I replaced the Poly-Switch (F175) near the parallel-/serial-interface, I pulled out both VIA-connections :-/ Do you know if these are connected to an intermediate level of the multi-layered mainboard? I successfully bridged the lines on the upper- and lower-side, but I'll have a big problem if there are more connections inbetween... :-o
-
I think you'll be ok on the VIA connections....
For the sockets, this page is your friend:
http://www.amigawiki.org/doku.php?id=en:service:schematics
The schematics on this page are simply awesome (and searchable). :) If it didn't always identify all the Fast RAM that sounds like something dodgy that could be a more serious problem.
-
The schematics on this page are simply awesome (and searchable).
Wow, those are amazing! Whoever spent the time to make those documents, great job! :drink:
-
basically I don't fear serious problems at all, as long as they are somehow solvable :-) I want that Amiga bring alive again..!
I realized that I can test the Fast-RAM SIMM's on my A1200 with Blizzard IV - I checked them with a couple of ram testing apps and they seem to be okay. I also try some other SIM-modules, same problem...
Games also don't work. They load until some point and then the system crashes again with some kind of red "software failure" message. What's the idea behind the "game-test" ?
I removed the "unrecognized" Fast-RAM for further testing.
Also I'm able to get into the early boot menu where all my devices are listed (even the hard-drive and the two external floppies which I once attached) DF0, DF2, DF3, HD0, HD1, HD2. What does "Disable CPU caches" mean by the way?
Someone any idea how to proceed ? Any measurement-based or mechanical approach? :rofl:
Btw: Thanks for the link to the schematics - I used them alot to identify which parts I need to order for replacement.
-
The purpose of the RAM test is less to check the RAM and more to check the connections to the RAM. Testing in another machine won't help here. :)
The purpose of a game test is just to rule out as many variables as possible (bad floppies etc.) - if a simple 512KB A500 game (Which is tested as normally working on an 040) fails, it'll give us a bit more idea. Still sounds like a bus error to me.
That thing about only 8MB RAM detected.... I presume you're fitting 4 4MB SIMMs? If it's supposed to be 16MB but only 8MB is detected, then something's squiffy with the RAM bus or connector, probably.
-
Kickstart ROM?
-
Thanks for the hint to the Kickstart-ROMs... I replaced the original ROMs from my A4000D with those of my A1200. Wow - now it boots within 1-2 sec to the boot-screen, but Fast-RAM is not recognized. Using the original ROM's, Fast-RAM is now recognized, but the system works very unstable and crashes all the times. Btw I replaced all the SIMM sockets with new ones...
Here's a quick summary of my two testing scenarios:
with A1200 ROM: very fast boot (1-2 sec), Fast-RAM not recognized (only Chip-RAM), system works stable
with original A4000 ROM: very slow boot (~1 min), Fast-RAM + Chip-RAM is recognized, system works very unstable
So my big question at the moment:
Are the Kickstart-ROMs v3.1 for the A1200 and the A4000 Desktop not the same :angry: ??
-
Kickstart ROM for A1200 is NOT the same as A4000.
Main differences:
scsi.device, battery clock support for A4000, ZORRO support (Super Buster) for A4000, onboard FAST RAM support for A4000.
And few more.
If A4000 crashes with A4000 ROM and working with A1200 ROM, it means that you have problems with stuff not presebt in A1200: FAST, RTC, Super Buster. If your Super Buster is socketed, please remove socket and solder Super Buster directly to the PCB. This socket tends to loose contact with the chip over the time.
-
How does it behave with all the Fast RAM removed? Does it work like with the A1200 or is still unstable?
-
@stachu100: many thanks for the valuable insights. so that explains my fast-ram issues in detail :-) since the original A4000 ROMs seem to have a defect (very long boot-time and problems addressing fast-ram), I'll try to replace them with new ones (I'll buy some from vesalia.de). Btw super-buster is not socketed and I already re-soldered all legs - thanks for the hint.
@spirantho: with all fast-ram removed and using the A1200 kickstart-roms, without any other devices except keyboard, mouse and chip-mem, the A4000 now works very stable again :insane:
summary: next step is to order new kickstart roms :-) It was a long and frustrating way until I got here - hope this finally solves my problems :-)
I will post my further findings as soon as I get the new roms... thank you so far for your help!
-
I'm 99,99% sure that ROMs are fime. You still have hardware problem on your mobo.
-
I agree, very unlikely that the ROMs are at fault. If there was a data corruption problem within them, this would be picked up by the checksum test during early boot.
Unfortunately I don't have a reference machine to test the Chip RAM and the A4000 crashes as soon as I want to open any folder or program on the workbench... But I have plenty of other SIMMs which I'll try after work.?
As others have pointed out, this behavour is consistent with memory issues. If you're able to boot without startup, run Memcheck (http://aminet.net/util/moni/MemCheck12.lha) from floppy, you may get results before it crashes due to memory issues.
Another problem which can cause this is clock synchronisation, which can happen if the A4000 is clocked from its own onboard oscillator while the external CPU board is running from its separate reference. Though often this causes total boot failure, or failure during early boot. With the A3640, both clock source jumpers on the A4000 main board must be set to EXT.
If there has been corrosion damage around U177 (RTC latch) this part should be replaced, else there can be issues with the latch causing corruption on the CPU address bus. Likewise, if U178 is damaged, this can cause CPU data bus issues. Though usually in this case, the machine just won't boot at all. If in doubt you can remove U177 and U178, the machine will boot fine without them. Use caution; if you had problems when removing F175 (below), then you're unlikely to remove these parts without causing more damage.
You're right, possibly the lines under the Chip-RAM sockets are dodgy. I'll solder out that socket to have a better look at them. But honestly said, the Amiga worked more or less stable before I began to change the IC's and caps - except it sometimes didn't identified all of the 16mb Fast-RAM - that was initially the reason why I began to replace all the elements... Which chips are involved in memory management?
In the A4000, half of U250/Bridgette forms the bidirectional bridge between chip memory and other devices on the chip memory bus. But it's unlikely this is the problem. Has there really been damage to the chip memory tracks? This is the U261 socket. There's nothing common in this area to cause much damage, aside from major leaking of C190, which is fairly rare, or worse than usual battery damage. If there is visible damage to any tracks, these should be checked and repaired accordingly. Most damaged tracks in this area will be picked up by the early chip memory test in ROM, which shows a green screen for chip memory failure. Though this test isn't very thorough, it only does a simple write/read check at 16kB intervals.
There won't be soldering issues on the chip memory socket, but broken retaining clips or very dirty socket contacts are a possibility. As others have said, you _MUST_ use a known good/working SIMM for chip memory, else you're chasing the unknown at this point.
The machine should run OK without fast memory. But any fast memory access problems are usually caused by open circuit _R_W to pin 11 of the fast memory sockets and/or corrosion damage to U891 and/or damaged tracks/vias around U891.
Btw: When I replaced the Poly-Switch (F175) near the parallel-/serial-interface, I pulled out both VIA-connections :-/ Do you know if these are connected to an intermediate level of the multi-layered mainboard? I successfully bridged the lines on the upper- and lower-side, but I'll have a big problem if there are more connections inbetween... :-o
Why did you remove this for? Anyway, one side of this connects to the internal +5V plane - specifically the side nearest U350. The A4000D is a four layer board, the internal two layers mostly carry GND and +5V. So measure it, and if the U350 side of F175 doesn't connect to +5V, then just run a small jumper on the bottom side of the board to the nearest convenient +5V source, which is practically everywhere. If this connection is damaged, you'll just have no +5V_USR, so essentially you'll lose mouse movement. The computer will boot fine without it.
If you can't resolve the issues yourself, feel free to contact me for a repair estimate.
http://amiga.serveftp.net
-
I meant how does it run with the A4000 ROMs and no fast RAM? Very unlikely indeed it's your ROMs at fault.
-
@Castellen: First of all, many thanks for your generous assistance! Your knowledge on the Amigas is really impressing. Following are my new findings and finally another question.
As others have pointed out, this behavour is consistent with memory issues.
Meanwhile I replaced all the SIMM sockets with new ones that feature those mettalic snap locks (here's an image of this area on my A4000 motherboard (http://nicolas.baumgardt.ch/amiga/amiga-4000D.jpg)). The original SIMM sockets didn't snap correctly anymore since they were weared out. Then I ran Memcheck from floppy and tested Chip- as well Fast-RAM in different combinations and all the tests succeeded.
With the A3640, both clock source jumpers on the A4000 main board must be set to EXT.
These jumpers are and already have been set correctly.
Has there really been damage to the chip memory tracks? This is the U261 socket. There's nothing common in this area to cause much damage, aside from major leaking of C190, which is fairly rare, or worse than usual battery damage.
No, there was indeed no damage to those tracks. I have checked all of them with a continuity tester after the SIMM sockets were removed. However, at this point I want to mention, that the whole motherboard was heavily affected by leaking electrolyte and there was initially copper acetate all over the board (more accurately, wherever it had electrolyte capacitors). This also was the reason why I removed F175, the area around this element was pretty in a bad state.
The machine should run OK without fast memory. But any fast memory access problems are usually caused by open circuit _R_W to pin 11 of the fast memory sockets and/or corrosion damage to U891 and/or damaged tracks/vias around U891.
I replaced U891 and U177 in a very early state with new ones. I also checked all via's and lines in this area several times.
The A4000D is a four layer board, the internal two layers mostly carry GND and +5V. So measure it, ...
I measured +4.92V between GND and F175 without any devices attached (except keyboard and mouse). Is this within the valid range?
If there has been corrosion damage around U177 (RTC latch) this part should be replaced, else there can be issues with the latch causing corruption on the CPU address bus.
Finally I removed U177 without problems and started the machine... wow - the Fast-RAM problems and the crashes were gone :-) ! So the problem is that RP5C01 uses a direct bus-connection to the CPU and if this element fails, the CPU shows up with undefined behaviour, resulting in system crashes ?
The following two issues are still open:
1. When measuring CPU speed with SysInfo, it only reaches a max of 14.95 Mips instead of the ~18 Mips measured in the past. Any ideas?
2. The Amiga needs approx. 35 seconds from powering on until the boot menu (the one with the hand & disk) shows up. My Amiga 1200 does the same within 1-2 seconds. Is this normal for A4000's ?
(EDIT: answered by HammerD)
@all: many, many thanks for your help so far - really great community!
-
2. The Amiga needs approx. 35 seconds from powering on until the boot menu (the one with the hand & disk) shows up. My Amiga 1200 does the same within 1-2 seconds. Is this normal for A4000's ?
I believe the delay is normal due to a 30 second wait for IDE devices. You can buy IDE terminators from AmigaKit to eliminate this delay.
http://amigakit.leamancomputing.com/catalog/product_info.php?products_id=831
on my A4000 if I don't have anything connected to the IDE port I get this delay.
-
Thanks alot - these are appeasing news :-) But as soon as the baby runs correctly again, I will boot from IDE, so there's no need for a terminator in my case.
-
@Castellen: First of all, many thanks for your generous assistance! Your knowledge on the Amigas is really impressing.!
I should hope so, I do Amiga repairs as a job.
I replaced U891 and U177 in a very early state with new ones. I also checked all via's and lines in this area several times.
Sounds as though your memory issues have been resolved, but also check continuity between U215 pin 12 and pin 11 of any fast memory socket.
I measured +4.92V between GND and F175 without any devices attached (except keyboard and mouse). Is this within the valid range?
Yes, this is normal.
Finally I removed U177 without problems and started the machine... wow - the Fast-RAM problems and the crashes were gone :-) ! So the problem is that RP5C01 uses a direct bus-connection to the CPU and if this element fails, the CPU shows up with undefined behaviour, resulting in system crashes ?
Yes, that was exactly my point. U177 is also directly on the address bus and can cause problems if it's badly damaged. But you've replaced this already, so no problems there. I've written various repair guides (http://amiga.serveftp.net/RTCrepair.html) on the subject.
You can buy new RP5C01 (U178) from AmigaKit, also I carry them in stock (http://amiga.serveftp.net/inventory.xgi). Try and get the correct variant, the RP5C01A as used in the A3000 is practically the same, but the capacitive loading of the crystal oscillator is a bit different, so it'll run too slow. In saying that, the original 32.768kHz crystal will have aged to the point it'll be slow anyway, requiring recalibration. You just need to connect a high impedance (>=10M Ohm) oscilloscope or frequency counter to U178 pin 17 and adjust VC190 to get the correct frequency. Do this with the A4000 powered off as the oscillator runs fractionally slower while the RP5C01 is in standby mode.
The following two issues are still open:
1. When measuring CPU speed with SysInfo, it only reaches a max of 14.95 Mips instead of the ~18 Mips measured in the past. Any ideas?
I wouldn't get too beat up about that. System benchmarks are often seemingly random numbers or an approximate indication at best. There's unlikely to be any hardware fault that "makes the computer go a bit slower then usual". This kind of thing generally either works or it doesn't. Other factors include any other patches you happen to be running when you first recorded the value.
2. The Amiga needs approx. 35 seconds from powering on until the boot menu (the one with the hand & disk) shows up. My Amiga 1200 does the same within 1-2 seconds. Is this normal for A4000's ?
Yes, as answered already, OS v40.xx waits about 30 seconds to allow any hard drives to spin up and become available.
-
So here I am back again with some good news :). Thank you all again for your great assistance - you helped me alot so far - but the quest isn't done yet... I finally found some time to further study the problems on my A4000.
I should hope so, I do Amiga repairs as a job.
Believe me - I would already have sent my beloved motherboard to you for repairing if you wouldn't live that far away...
Sounds as though your memory issues have been resolved, but also check continuity between U215 pin 12 and pin 11 of any fast memory socket.
You can buy new RP5C01 (U178) from AmigaKit, also I carry them in stock (http://amiga.serveftp.net/inventory.xgi).
Okay I did and continuity is ensured. Meanwhile I bought a new RTC chip from amigakit and they sent me a recycled but working one. Btw thank you for your offer, I will come back to it for other parts - I decided to order it from amigakit since I additionally bought some other stuff they had in the store.
I booted the Amiga with the new RTC chip and everything seemed to work fine until I rebooted... Then I got yellow and black screens... I then thought, the "recycled" chip could possibly be wrong - but I then searched the Internet and found the hint with the diode between the transistor and pin 18 of the RTC chip. Wow - now the Amiga works very stable again and it's recognizing all memory all the time !!!
Yes, that was exactly my point. U177 is also directly on the address bus and can cause problems if it's badly damaged.
What is the cause that U177 fails? I mean the battery acid has damaged the pins, but the case of the RTC is intact. Is this maybe caused by a excess voltage due to a closed circuit or similiar?
I wouldn't get too beat up about that. System benchmarks are often seemingly random numbers or an approximate indication at best. There's unlikely to be any hardware fault that "makes the computer go a bit slower then usual". This kind of thing generally either works or it doesn't. Other factors include any other patches you happen to be running when you first recorded the value.
You were right (as usual ;-). With "68040.library v44.1" installed, the 3640 achieves 14.95 Mips in SysInfo v3.24 (http://aminet.net/package/util/moni/SysInfo) and with "68040.library v46.5" it outputs the usual 18.85 Mips.
But... as I said before, the quest is not completed yet - I also have some audio problems for which I'll open a new thread (http://www.amiga.org/forums/showthread.php?p=762546#post762546). It would be great if you could take a "short" look at it... :-)
Thank you very much for your help so far ... !
-
Believe me - I would already have sent my beloved motherboard to you for repairing if you wouldn't live that far away...!
It's no big deal, sending an A4000D main board doesn't cost a lot providing you use a standard surface or air mail service and not an international courier. I have hundreds of regular customers all over the world; UK, Europe, USA, Canada, etc.
the hint with the diode between the transistor and pin 18 of the RTC chip. Wow - now the Amiga works very stable again and it's recognizing all memory all the time
That diode just provides a more direct 5V source to the RTC. What can happen is if the battery is very depleted, it causes excess load on the power supply to U178, which does not operate correctly with a low supply voltage. This can cause the system to freeze during boot.
Pleased that's fixed it for you.
What is the cause that U177 fails? I mean the battery acid has damaged the pins, but the case of the RTC is intact. Is this maybe caused by a excess voltage due to a closed circuit or similiar?
I don't quite understand your question. U177 damage is almost always due to battery corrosion. If the device appears visibly corroded then it should usually be replaced. It can happen that it operates OK for a while, but can fail in the future, even after it has been "cleaned up".
the quest is not completed yet - I also have some audio problems for which I'll open a new thread. It would be great if you could take a "short" look at it
No problem, but have a look at the information I've already written on the subject (http://amiga.serveftp.net/audio_repair.html).