jvas Posted February 23, 2015 Share Posted February 23, 2015 When I turned the computer on it went to the memory test screen with the following result: Does anybody have any idea how to trace the error? It has the 512KB memory upgrade from MegaHZ (removed it but no change) and a 32in1 OS from AtariMAX. The 32in1 works fine, I can change the OS during boot. It also had a dual pokey upgrade which was removed too. I measured the current of all the soldered memory chips, and it is 4.9V. Thanks for your help in advance! Cheer,s Jozsi Quote Link to comment Share on other sites More sharing options...
adam242 Posted February 23, 2015 Share Posted February 23, 2015 I once repaired an 800XL with memory issues by piggybacking a known good RAM chip onto each chip, one at a time, until the machine would pass the memory test. I then unsoldered that bad chip, installed a socket with the good RAM, and all was well. Quote Link to comment Share on other sites More sharing options...
+bob1200xl Posted February 23, 2015 Share Posted February 23, 2015 Those are Micron memory chips. They have not been reliable, I'm afraid. Are any of them hotter than the others? Start there. Does the first memory row run OK? (4 rows - what is that?) Bob Quote Link to comment Share on other sites More sharing options...
jvas Posted February 23, 2015 Author Share Posted February 23, 2015 Those are Micron memory chips. They have not been reliable, I'm afraid. Are any of them hotter than the others? Start there. Does the first memory row run OK? (4 rows - what is that?) Bob Hi, No difference in temperature. All of them are cold. The first row is OK (green) + the 1 st square in the 2nd row. I don't know why there are 4 rows. Maybe the memory upgrade? Cheers, Jozsi Quote Link to comment Share on other sites More sharing options...
BillC Posted February 23, 2015 Share Posted February 23, 2015 As Bob already posted, the system has Micron RAM(MT4264) which is prone to failure. I would recommend simply replacing them all and checking again. As for 64 squares for the memory test, I would think it's a feature of whichever OS are you using for the diagnostics. I have a CSS US+ OS which completely tests a vintage Newell 1MB(total 1088k, this isn't included in the Atarimax 32-in-1) 800XL, I don't know if it will work with a U1MB or MEGA-HZ 512K upgrade. Quote Link to comment Share on other sites More sharing options...
1050 Posted February 24, 2015 Share Posted February 24, 2015 Hi, No difference in temperature. All of them are cold. The first row is OK (green) + the 1 st square in the 2nd row. I don't know why there are 4 rows. Maybe the memory upgrade? Cheers, Jozsi They should be all quite warm after working ten minutes or so, it's only the really HOT ones that can raise a blister that indicate it's a shorted out RAM chip. These can only be fixed by replacement - the system will NOT boot until fixed. If they are still cold after ten minutes it's also a sign that they are out to lunch forever more - these are the ones that can be piggy backed with good ram and will show a pass then. They also need replaced, but this failure mode of doing nothing allows the piggy backed to pick up the slack as a temporary diagnostic situation. All RAM squares should be showing red with any single RAM chip doing nothing. Row structure in that test leads nowhere, you can't use it to find a bad chip. Every RAM chip is involved in every single memory location so the arrangement of the system can't possibly be used to hunt down only the bad chips. Despite the squares that very often lead users to think otherwise - it's always just wishful thinking and wrong. Your RAM is working, you wouldn't see that screen if it were bad since the image is stored in RAM and you are looking at it. What the squares do represent is certain block sizes retaining information written to them, so first row green says first quarter of memory is retaining data. Except that they can't test the stack or page zero or, or, or and the issue becomes one of just how the test was written and exact locations with block sizes the squares represent. A matter for experienced code disassembly and with little reward for doing so, few even bother to try. Four rows may be from MegHZ OS? Each OS will have it's own self test code (offset 1000 thru 17FF, always remapped to 5000 thru 57FF) and you should be able to see three rows with one of them out of 32 possibles. Four rows tells me you have custom code running for the self test and it may not be properly de-bugged to work on your system even when it IS working perfectly. Also possible that the custom code requires the memory upgrade to be working and it's not. Especially now that it has been removed in the meantime. It's like a big bucket of worms, you want to ignore it as much as possible, if at all possible. The self test is a known stinker because of so many issues with too few real solutions for those issues - it's there primarily for the eye candy aspect in the first place. Choose your battles wisely. Quote Link to comment Share on other sites More sharing options...
Kyle22 Posted February 24, 2015 Share Posted February 24, 2015 If you are talking about the four blocks on the lower part of the screen, these test the extended RAM. If you have ANY red blocks, I would replace (and socket) ALL of the RAM chips. Quote Link to comment Share on other sites More sharing options...
jvas Posted February 24, 2015 Author Share Posted February 24, 2015 Progress report: All the 8 chips had been piggy backed. Unfortunately no change. The memory check is the same. Quote Link to comment Share on other sites More sharing options...
blacka013 Posted February 24, 2015 Share Posted February 24, 2015 As has been said, unsolder the ram in the motherboard, solder in sockets and fit new ram. These micron ram chips fail for any number of reasons. DO NOT replace with micron chips. Quote Link to comment Share on other sites More sharing options...
+bob1200xl Posted February 24, 2015 Share Posted February 24, 2015 I don't recognize the OS - can you boot the normal 800XL OS? Motherboard looks strange, also. Is that a Freddie chip in the lower left? If it is, that's where a lot of your address lines get processed. Bob Quote Link to comment Share on other sites More sharing options...
Rybags Posted February 25, 2015 Share Posted February 25, 2015 Weird... maybe that 4 row memory test occurs if the older XL OS fails the initial memory test right at the start, ie doesn't detect anything and leaves the memory size at 00 ? Could be something to test under emulation... Quote Link to comment Share on other sites More sharing options...
1050 Posted February 25, 2015 Share Posted February 25, 2015 Thanks Kyle22 for the screen shot of the highest version of the Atari 130XE OS self test. Those four 'extra' blocks are only testing the stock XE extended banks of E3, E7, EB, and EF. I get the same screen when testing my 1 meg newell and no other banks are tested. This same self test code is in the XEGS machine but since it doesn't have extended memory, the test starts over when the extended memory test fails to find any defaulting to the standard three rows as per a normal self test screen. So someone has re-written the self test for this as yet un-named OS to do a larger job of extended memory testing. I'm thinking they didn't try to use it on this non-working memory upgrade though. Perhaps the upgrade memory red blocks actually start two blocks into the second row in this custom code? Which OS is the 32 in 1 set for? Have you tried any other OS choices? If yes, do you still get this same screen with four rows? If so, it appears that the 32 in 1 is NOT working, do you remember which OS it was selected for in the beginning? Rybags, IIRC the fourth row has to be set up by code changes over a stock version to even allow for it to be on the screen. So my opinion is that the self test code is custom. Bob1200xl, I do believe we are looking at a freddie 800XL, talk about rare. End of the day, I'm thinking you need to box it up along with removed items and see if the installer can find the problem, they have vastly more experience with their product than I for one example will ever have. My hunch is that the memory upgrade at the very least needs to go back to be tested for faults. Quote Link to comment Share on other sites More sharing options...
jvas Posted February 25, 2015 Author Share Posted February 25, 2015 As has been said, unsolder the ram in the motherboard, solder in sockets and fit new ram. These micron ram chips fail for any number of reasons. DO NOT replace with micron chips. Will do, but I don't have none at hand. Need to buy somewhere... Quote Link to comment Share on other sites More sharing options...
jvas Posted February 25, 2015 Author Share Posted February 25, 2015 I don't recognize the OS - can you boot the normal 800XL OS? Motherboard looks strange, also. Is that a Freddie chip in the lower left? If it is, that's where a lot of your address lines get processed. Bob It is "XL/XE (REV C)" according to the OS selection screen. Which one you consider "normal"? "XL/XE (REV A)" does the same memory test Yes, there is a Freddy. This MB worked previously when I put it away. Quote Link to comment Share on other sites More sharing options...
jvas Posted February 25, 2015 Author Share Posted February 25, 2015 Replacing/socketing the memory chips solved the problem. Thanks for your help! Quote Link to comment Share on other sites More sharing options...
Kyle22 Posted March 3, 2015 Share Posted March 3, 2015 Thanks 1050, now, looking back at the pic, I see what he meant about the extra row of blocks, making 4 total. The only thing my mind could picture was the extended RAM test, which I posted a pic of. I have never seen that OS, but it would be nice to have a dump of it Quote Link to comment Share on other sites More sharing options...
1050 Posted March 5, 2015 Share Posted March 5, 2015 I had the same evil plan but wasn't going to be so obvious about it until I had it nailed into a cage at my place. Unfortunately, jvas doesn't seem to understand that he can't possibly be looking at that screen AND be using different selectable OSes at the same time - it just doesn't work that way. He doesn't yet realize that the 32 in one is stuck on one OS for some reason or he isn't operating it right - I have no clue as to which. But this may be the ONLY reason I have seen for getting a 32 in one so far... It may even boil down to the memory upgrade having a good deal more talent than the typical upgrade and it commandeers ALL self tests to it's own version? That might be something to have too. And in which case jvas is an absolute genius for ignoring my rattlings entirely !! It's been interesting all around. Thanks back. Quote Link to comment Share on other sites More sharing options...
+Larry Posted March 5, 2015 Share Posted March 5, 2015 Thanks 1050, now, looking back at the pic, I see what he meant about the extra row of blocks, making 4 total. The only thing my mind could picture was the extended RAM test, which I posted a pic of. I have never seen that OS, but it would be nice to have a dump of it Yes it would. OK, let's start looking... where did you find that screen shot? It looks like the regular Atari memtest with those ext. blocks added, so if it is stock Atari, it would likely be in a very late model OS like Rev. D. (?) This isn't the same thing, but I always liked this one -- similar "look and feel." -Larry Radon Memtest.zip Quote Link to comment Share on other sites More sharing options...
jvas Posted March 6, 2015 Author Share Posted March 6, 2015 I had the same evil plan but wasn't going to be so obvious about it until I had it nailed into a cage at my place. Unfortunately, jvas doesn't seem to understand that he can't possibly be looking at that screen AND be using different selectable OSes at the same time - it just doesn't work that way. He doesn't yet realize that the 32 in one is stuck on one OS for some reason or he isn't operating it right - I have no clue as to which. But this may be the ONLY reason I have seen for getting a 32 in one so far... It may even boil down to the memory upgrade having a good deal more talent than the typical upgrade and it commandeers ALL self tests to it's own version? That might be something to have too. And in which case jvas is an absolute genius for ignoring my rattlings entirely !! It's been interesting all around. Thanks back. I'm not a genius. I simply knew, that previously the computer worked fine, but now boots into the ram checking test (nothing has changed in between). I haven't checked the memory test in the working computer. First I thought that by "reading" those squares, the faulty component can be identified. But I learnt, that reading and writing to/from every memory location, all the eight memory chips are involved. The 32in1 OS is very easy to use. Every time the OS selection screen appears, it definitely indicates which one is active. It is really hard to make a mistake. There are memory tester programs coming with the memory upgrade which can test the extended memory, and now they reports no problem. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.