Jump to content
Level42

The mT RAM investigation thread. Show yours, good or bad....

Recommended Posts

USA also. Bought mine from someone on comp.sys.atari.8bit about 25 years ago. I don't use it much as I'm not a big fan of the keyboard.

  • Like 1

Share this post


Link to post
Share on other sites

Do most 130XEs have a different brand chip at upper left? Is there a technical reason for it, such as bus termination? Has this been discussed elsewhere?

  • Like 1

Share this post


Link to post
Share on other sites

Do most 130XEs have a different brand chip at upper left? Is there a technical reason for it, such as bus termination? Has this been discussed elsewhere?

 

Level42 started a thread about that possibility before I think. So far as people have shown, it appears that quite a number of 130XE's do, though we really don't have a solid reason why.

  • Like 1

Share this post


Link to post
Share on other sites

There are two things that need to go away in an Atari 8 bit. MT ram and Wang Capacitors. The suspect Wang's are usually in 1050's. Both are ticking time bombs.

  • Like 1

Share this post


Link to post
Share on other sites

I'm going to guess that it has to do with address bus termination. The upper left chip is farthest from Freddie, which originates the RAM address bus. An EE rule is that the far end of a bus should have some extra load to reduce the reflection of signal edges. Maybe the Micron RAMs have less input load than other brands (I haven't researched the datasheets).

  • Like 3

Share this post


Link to post
Share on other sites

I looked at three datasheets and saw no difference in input leakage current between Micron, Motorola, and NEC.

Share this post


Link to post
Share on other sites

Yeah I started a thread about it and one or two people thought I was talking nonsense and that it was just "random" but to me, lots of pictures and my own experience with machines on my desk with ONLY the top left position filled with another brand AND my own experimentation, trying to run a good working mT RAM chip which works fine in any other location constantly failed to work in the top left one..led me to the conclusion that it was a very deliberate action to put other brand chips on that position, so Atari knew about issues with mT RAM already during production yet chose to keep using them.....it really makes me wonder who figured this solution out and how he did it.....or knew the reason behind it.

 

http://atariage.com/forums/topic/271274-130xe-ram-what-configuration-do-you-have/?hl=+mt%20+ram

  • Like 1

Share this post


Link to post
Share on other sites

it's both, asked and answered... address bus load and tolerance of power hits at start up... not to mention when a power supply goes out of spec. I was deliberate. No one listens...

I never learn because I keep typing.. :)

Share this post


Link to post
Share on other sites

of course it was deliberate, the single position where it happens is proof enough.

 

Funnily I found back one picture that has all -20 Samsung RAMs and also a different brand one on the top left position in that thread that I had forgotten about...

 

I'm sure it's something like you say, but it would take some decent measuring/analyses to see what's really going on. I recently saw a video where someone analyzed something with a HP device that was much like a "regular" (digital) logic analyser, but it could also record the signals in analogue format and you could compare them.....don't think anyone has such a device here, I can imagine even now they are pretty costly.

 

I have a Fluke 9110 system that might find something, but I never got it to work with the A8, but I later realized that the regular 6502 pod naturally wouldn't work in a XL or XE because of the extra Sally functionality missing. It should work with an 800 though.

 

Anyway,....I left that thread as is.....this thread is about finding a pattern....please keep those 800XL/130XE RAM configurations coming. I realize that most of us don't take pictures from bad RAMs we replace, I regret very much so not doing either....I even had NOS mT RAM chips here that were bad from the start.

Mmmm.....did I keep some of those just for the record....let me dig around a bit :)

Share this post


Link to post
Share on other sites

I have a pile of them that were good years ago, that were piggy backed for mem upgrades that never got installed, I wonder if they died in storage.. not even being used... you never know....

 

*edit* as luck would have it, I did find one of the zip lock baggies with what looks to be a 512k memory upgrade ready to go....all mt's

should I attempt an install?

Edited by _The Doctor__
  • Like 1

Share this post


Link to post
Share on other sites

About this single different chip, think I know the explanation. It might be factory fix for some known bug.

RAM chips are able to keep data stored up to 15sec. without refreshing. And there is no procedure for cleaning up/zeroing memory that Atari performs on booting. So - if you power off the computer, then turn it on immediately - it is likely that Atari just hangs up due to some residues of data still existing in RAM.

A working solution is to replace single memory chip with another with different access time. For instance - one chip 200ns vs all the rest with 150ns. Or just put in one chip made by another manufacturer - the one replaced will be almost identical as the rest, but almost is enough to do the trick.

Source: Jerzy Sobola homepage http://dereatari.republika.pl/serw.htm
and this topic on atari.area board (use Google to translate) http://www.atari.org.pl/forum/viewtopic.php?id=13612

  • Like 3

Share this post


Link to post
Share on other sites

About this single different chip, think I know the explanation. It might be factory fix for some known bug.

 

Yes, I think we all agree on that by now :)

 

If your theory is true, then why do 400/800 and XL's have all the same RAMs ?

 

Also....what does a faster acces time on one RAM do to make the others (or actually, all of them) loose their contents faster after power off ?

Edited by Level42

Share this post


Link to post
Share on other sites

Oh and.....in my experience it is actually XE's that need the longest power down time before being able to boot up "clean" again.....and XE's is where the "one different RAM is found, not the others"

.

 

Further, I tried to actually run a 130XE with a know working mT in any other position but It would fail when put on the top left position......even after having been powered off for minutes,

Edited by Level42

Share this post


Link to post
Share on other sites

Hello there,

 

Not sure this'll be of any use to you but here is a picture of the RAM in my BAD 800XL

 

All mT 433. Not sure how many are bad yet.

 

Geoff.

 

post-62928-0-63540800-1515524454_thumb.jpg

  • Like 1

Share this post


Link to post
Share on other sites

About this single different chip, think I know the explanation. It might be factory fix for some known bug.

 

RAM chips are able to keep data stored up to 15sec. without refreshing. And there is no procedure for cleaning up/zeroing memory that Atari performs on booting. So - if you power off the computer, then turn it on immediately - it is likely that Atari just hangs up due to some residues of data still existing in RAM.

 

A working solution is to replace single memory chip with another with different access time. For instance - one chip 200ns vs all the rest with 150ns. Or just put in one chip made by another manufacturer - the one replaced will be almost identical as the rest, but almost is enough to do the trick.

 

Source: Jerzy Sobola homepage http://dereatari.republika.pl/serw.htm

and this topic on atari.area board (use Google to translate) http://www.atari.org.pl/forum/viewtopic.php?id=13612

There would be no way of knowing the data pattern would be garbage at start up and the Atari doesn't care (for the most part), you could not ensure what changes and what doesn't! This theory would keep the Freezer and other devices from determining their own warm/cold boot capability... slightly different refresh rate has nothing to do with 8 bit and 7 bit refresh schemes... and theory's of corrupting a frame intentionally to force a cold start are ill advised.

Edited by _The Doctor__
  • Like 1

Share this post


Link to post
Share on other sites

Some science here: http://citp.princeton.edu/research/memory/

Atari found out, that DRAMs from particular series "remember" bits for a long time after the power is off. And if you switch power off and on to force cold start, the computer boots with garbage on screen, then hangs up. So they developed a fix: mix each set of DRAMs with a single chip from another batch, the "fast forgetting" one, and the assembled computer passes tests OK! It was easy and cheap, so they did it.

Can't see anything unbelievable in such a story. It is recommended to add one or two different chips when replacing defective RAM in XL/XE's, it's common practice actually. And I didn't make it up, it came from guys good at electronics.

  • Like 2

Share this post


Link to post
Share on other sites

This didn't manifest until late 2007 to 2008.... it only worked on relatively cool chips... and in fact.. you could freeze a chip and preserve the memory contents..... while I was unable to open the page the paper would point to... I am aware of what it will say... this is an attack vector on encryption... but it is pointless as the attacker must already have physical control of the computer and the memory.... it's only importance is to pulling keys off of lets says cable and or satellite system chips that might have one time installed battery backed up keys in them, and most of us who have had to recover such keys don't freeze the chips... we simply supply power to the chip as we move it...

 

Trust me it would not take 11-15 seconds to fully clear an XE if what you state was fact.. because it already has one odd chip...and the specs for timing et all match up..

 

I have already told you the reason for this chip. If Tramiel could have made the damn things any cheaper he would have... you have heard what actual engineers have said... applying discoveries on chips a decade later or more to try and explain this is the crazy part...

 

did you even read the papers.. I just did... and nope it did not empirically eliminate the bits in memory while the strategy was considered. I have learned nothing new and feel slightly deflated about that. It would have sure made the Atari an even bigger pioneer than before.. sadly not the case... hey what about the other memory chip configurations in the 130XE's... or 65's XL or XE... it just doesn't carry through..

 

trust us memory remanence was not the issue!

 

This change was to prevent death of the memory and was as stated at the end of the bus for dual purpose.... or you might assume we would replicate the odd chip in each bank of the 130XE's two rows of ram if it were a remanence issue.... we could have bled the banks in other ways mind you with far less effort or cost....

Edited by _The Doctor__

Share this post


Link to post
Share on other sites

In my experience, the machines most prone to prolonged DRAM retention are XLs (600XL and 800XL). I had an 800XL in the other week which couldn't complete a power-off, power-on cycle for some 3-5 seconds without the BASIC prompt immediately appearing (precisely as per the previously referenced Sobola article). I can't say I ever observed such severe symptoms on the XE machines. Just what I'm seeing in practice...

Edited by flashjazzcat

Share this post


Link to post
Share on other sites

dram retention of 4 seconds is not uncommon as the board draws down, the XE can take as much as 11 to 15 second for a clean cold boot depending on the caps and odds and ends on the board. The odd ram chip clearly doesn't change that condition. But it does address the other issues mentioned time and time again...

Edited by _The Doctor__

Share this post


Link to post
Share on other sites

The odd ram chip clearly doesn't change that condition.

I wouldn't know one way or the other, but the assertion in the quoted article is that it does. The idea appears to be that you only need one bit to clear more quickly for the OS magic bytes to become invalid (even if all seven other bits retain their data for several seconds) and thereby force a cold boot. One DRAM would accomplish this if it actually possesses the supposed properties.

  • Like 1

Share this post


Link to post
Share on other sites

While I've never experienced data retention problems

outside the 64 x 4 machines such as XEGS and last

run of 130XE with 4 ram chips total. They will

retain for 20 seconds, the XL always does a coldstart

boot at one quarter second power off time total here.

I had always assumed it was due to the superior silicon

used to make the modern 64 x 4 chips and of course I

can't possibly prove or know anyway.

 

What I'm reading is outside decades of my experience,

so I'm having one of those moments of disbelief.

 

And just now I do recall the issue also occurred with

some ram upgrades as a side note. Some did retain

while others did not and by now I've completely

forgotten who the good guy was and which was the one

you never wanted to install. This alone put me off

of installing a ram upgrade for some 5 years. Just

about long enough to forget the issue entirely.

When I had money burning a hole in my pocket I got

a 1 meg Newell which did not do it and have forgotten

all about ram upgrades doing this to an otherwise

well running machine before the upgrade was done to

it until now.

Share this post


Link to post
Share on other sites

If that were the case the ram chip would require a faster refresh rate than the other ram as well as a faster bleed rate... it of course would have to be in tolerance to still operate... I can't remember datasheets, scopes or analyzers supporting that.. I've got the odd chip out XE sitting here and it takes forever to get a cold boot flipping the power switch... so all theory aside. I'll stick with what was designed and leave the other benefit if any to happenstance. Might I also add that failing intermittent ram does not cause the Atari to reboot over and over endlessly, as it depends on where the memory corrupts for the magic calculation... to be honest I've been done with this for ages... I know exactly why it was done originally, and if there is a side benefit to it all the better. I don't see it with the four 130XE NTSC 16 chip boards I have here... although my 2 stacked mt chip upgraded 320XE's do cycle faster...

 

I will re visit this at some point and check the scribbling if any that I find. Who knows maybe it was a triple purpose... forgotten and not applied as time progressed.. Anything is possible.

 

Maybe I should look at this slow as sh*t 130XE and see what I find.... my plan for it was an u1m and not worry about it any longer.... I guess I'll check to see how much power remains from the timer and reset circuit as well as the ram rail / refresh lines... although the horse is long dead why not...

Edited by _The Doctor__

Share this post


Link to post
Share on other sites

Funny thing.... is during the fail to cold boot and then it returns to basic.... the memory contents are corrupted and sometimes basic ends up freezing... that should not happen according to the theory... it should come back as invalid magic number and reboot... it doesn't.. so the corruption must not be in the magic calculation area... why am I futzing about with this? Is this a form off ocd? Do I carry this out to freezing cold bench and or do I carry all that crap into the house.... well that answers that.. the wife says neither... No more crap is coming back in and it's too cold to go out. This will probably wait till spring. All 8 of us will get some answers then, laugh about it and wonder why we did it. Although the cold would increase the time I get to observe this... and the wife just smacked me upside the head... she's reading what I'm typing.. she just doesn't understand...

Edited by _The Doctor__

Share this post


Link to post
Share on other sites

I just found a hand full of these RAM chips. With the exception of a couple, there are 16 each of 8504 or 8505. All have been desoldered (is that a word?).

So I'm assuming that those are from upgrades. One other chip is dated 434 and the other is 43? because there is a nice round hole in the top center of the

chip where it looks like something inside was rapidly vaporized. I guess I kept it cause it looks cool...

 

DavidMil

  • Like 1

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...