128K: bad RAM repair

[Mac128]

In addition, having two chips fail is seemingly quite rare as well
One other question - where is the Shut Down command for System 1.1?

Actually failed RAM chips is relatively common. Others have rejected my theory, but I find this to be the most common on Apple-branded chips. Have a stack of bad boards with these chips to repair sometime. It's called bit-rot.

And there is no shut down command on 1.1. So be careful that you quit everything on the disk, given that it is stuck and you will not be able to eject before you turn it off, which could result in disk corruption.

 

[tomlee59]

Very interesting about the Apple-branded DRAMs exhibiting unusually high incidence of bit rot. I've not seen that myself, but your data certainly trumps my non-data. I'm just trying to figure out how two could fail together (assuming that this is what happened), or what type of failure of one DRAM could cause the death of another.

It is true that US memory manufacturers didn't do such a great job of manufacturing and testing DRAMs back in the 80s (which is why we got our backsides kicked by the Japanese). Perhaps you're seeing the longer-term consequences of poor QA/QC.

In any case, I'm glad that the OP got it working -- it's nice when a compact Mac is saved from the scrap heap!

 

[Byrd]

What's even more puzzling is that going off the documents saved to the floppy drive, this 128K was used for nigh on fifteen years before apparently failing - two DRAMs together, or something along those lines.

Hoping to test out the Imagewriter I printer next.

JB

 

[Mac128]

Byrd, this is pure speculation, and certainly others may know more than I, but my guess is the 512K was functioning properly when it was mothballed, in favor of a new computer. It was right around 2000 where the Internet had progressed to a point where many of my 68K Macs were rendered useless, by more sophisticated websites. In particular BBS and other text-based sites, the kind on which stock quotes could be found, had begun to be discontinued.

But bit rot strikes me as something that happens when the RAM sits idle. Though since I don't know what causes it, I can't be sure, but there are a number of stories here and elsewhere of someone putting their working 128K/512K into storage for only a year or so, to find a failed RAM error code upon attempting to put it back into service. So a Mac that sat idle for over a decade would certainly not be uncommon to develop bit rot.

Tom Lee may certainly be onto something with Quality Control. In my experience there is no particular set of chips that fail. Granted the fact it is usually two, not one, or three or more, is certainly strange. But it does always seem to be the Apple variety. What has been mentioned before is that Apple bought their chips from many manufacturers and screened their logo onto them. If this is the case, I would still suggest that I am seeing boards from one or more manufacturers with poor QC. In which case, it is not all Apple-branded chips but Apple chips are the most likely to experience bit rot failure.

 

[Gorgonops]

bit rot strikes me as something that happens when the RAM sits idle...

I'm not saying this to start a fight, so... don't freak out.

I'm somewhat curious where you found this definition of "Bit Rot", IE, how this nickname applies to physical failure of a DRAM. The standard definition of "Bit Rot" refers to corruption of *data* stored in "static" memory technologies, most often EPROM/Flash chips, but also SRAMs storing data for long periods without refresh. In all these cases it describes the data loss that results when the stored charge denoting a bit "leaks" enough to change a "1" to a "0". The chip itself may not actually be damaged but the data has become corrupted and the only way to fix it is to write a new copy of it back to it. (A nontrivial process in the case of an EPROM.)

(Reference: A description of Bit Rot in EPROMS)

The term is also applied to data corruption on disk, tape, and other magnetic storage devices due either to mutual interactions of the magnetized bits causing it "erase itself" over time, or because the media itself literally degrades, or "rots". (The physical degradation definition also applies to, for instance, data stored on aging CD and DVD media.)

"Bit Rot" is a great term to apply to DRAM chips failing over time, but again, I'm curious if there's actually some electrical/chemical/technical phenomenon you're describing with it. (A link would be great.) Most of the references I can find on aging effects on semiconductor memories suggest that said devices have a very flat degradation curve once they're past an initial "breaking-in" period. An implication derived from that is that it's possible the memories that "fail" in Macintosh 128ks haven't actually "failed", but were simply marginal in terms of output response to begin with. If that were the case then perhaps there is something else in the support circuitry on the Mac motherboard that *is* more prone to aging that is now failing to work with said marginal devices. (Top candidates would probably be filter capacitors or parts of the power supply. Capacitors in particular are something that seem to suffer from periods of idleness.) It would be an interesting experiment to try the "bad" chips in something else that takes 64k DRAMs, say an old PC expansion memory card, and see if they fail RAM tests there as well. (It's also possible that aging in other components is subjecting the mediocre RAM chips to out-of-spec voltage surges and "killing" them, in which case it may only a matter of time before the more robust replacements or other chips on the board start to fail.)

 

[tomlee59]

Well, I can invent scenarios that could explain the observations; whether they're correct is a different matter altogether.

For example, suppose these DRAMs aren't truly hermetic. Over time, moisture could seep in and corrode things (say, a bond wire or bondpad). Or if gold bondwires were used, it's possible for the dreaded "purple plague" to work its nefarious magic.

These problems would show up over time in powered or unpowered units but if poor hermiticity were the problem, storage in a damp environment would certainly be a bad thing. An operating machine would also be a hot machine, perhaps keeping moisture somewhat under control.

Without some failure analysis on the dead chips, one can only speculate. But mechanisms that could plausibly explain a couple of dead DRAMs in one machine do exist.

 

[Gorgonops]

Corrosion of the bonding wires hadn't occurred to me... that could well be a possibility. (Although I suppose you'd have to *very carefully* crack open the case of a dead DRAM and look at it under a microscope to really say if that's what is going on.) If it *is* some sort of internal corrosion that's killing these units it would seem that "Chip Rot" might a better name for it than "Bit Rot", given the conflict with the widely-used definition for the latter.

(IE, "Bit Rot" describes the data loss from aging or physical degradation, rather than being a technical term for a specific type of physical degradation process affecting the hardware. The only way the standard definition of "Bit Rot" applies to DRAM is the corruption of data in a "live system" due to soft errors, IE, bits flipped due to static discharges, alpha particles, cosmic rays, etc. Since DRAM doesn't retain data when powered off by design it cannot *itself* suffer from "Bit Rot".)

Anyway, seriously, I was curious if the term was being used specifically here to denote some manufacturing or process weakness widespread in chips of this age, and if there were a reference one could point to when "diagnosing" a failure of this type as "Bit Rot". Otherwise it doesn't seem like it's useful to diagnose these chip failures as such.

(I vaguely remember a term similar to "chip rot" being used in an article I read *many moons ago* in an 80's computer magazine to describe an age-related failure in early PROM chips, PROMs being programmable memories which store bits essentially by "blowing fuses" instead of retaining a charge like an EPROM. PROMs were widely used in early S-100 and other low production volume computer systems, and I *seem* to recall that the problem was traced to a "Tin Whiskers"-type phenomenon inside the packaging... but it's been far too long for me to remember specifically.)

 

[Dennis Nedry]

Another perfectly valid possibility is that it's a not-too-unlikely coincidence. If one chip could fail completely on its own in a 10-year window, another chip could also fail completely on its own. Simply square the probability of the first one failing and you have the probability of both independently failing.

 

[H3NRY]

Multiple chip failures are fairly common. What kills one is likely to kill another from the same batch, ie. they are likely to have the same bad spot on the die as the mask is stepped across the wafer. As for dying in storage, it's common for the analog section to drift as aluminum capacitors lose their oxide dielectric when not carrying a charge. They need time under bias to reform the oxide and for leakage to decrease. That's why old tube radios are usually gradually awakened on a variac. Anyhow, it's pretty likely that +5V is a bit lower than it used to be, and those 2 chips need a full 5 Volts to work. Like an overclocker raising his RAM voltage to run his memory faster than it's rated speed. I can think of a couple more similar explanations, but I have noticed the same effect, and the Apple-branded RAM failed fairly commonly when it was new. Most of it was made by Micron, and their early chips were flaky. Flaky chips are good when they are potato chips, but not the silicon variety. (Micron was founded by an Idaho potato millionaire.)

 

[Dennis Nedry]

My math is slightly off in my previous post.

 

[Gorgonops]

Most of it was made by Micron, and their early chips were flaky. Flaky chips are good when they are potato chips, but not the silicon variety. (Micron was founded by an Idaho potato millionaire.)

Hey now, electronic television was invented by a farm boy from Idaho (not to mention some other cool things, like the only fusion reactor that qualifies as a science fair experiment), so no hating on Idaho when it comes to technology.

 

[techknight]

I thought baird invented TV.

 

[~Coxy]

He invented TV first perhaps, but his invention was a mechanical version that was pretty temperamental and complicated.

 

[H3NRY]

Nothing against Idaho. Like any startup, Micron had teething problems. Potatoes to RAM still seems like an odd segue, but Micron Semiconductors certainly made it.

 

[trag]

Nothing against Idaho. Like any startup, Micron had teething problems. Potatoes to RAM still seems like an odd segue, but Micron Semiconductors certainly made it.

Is Micron the memory company the same Micron that made the Xceed video cards?

 

[H3NRY]

Yep. Micron went looking for places to sell their RAM, and video cards back then were nothing but lots of RAM and a DAC/CLUT chip.

 

[QuicksilverMac2001]

Well, I must have really lucked up. My 128k was used at least relatively thoroughly by every owner (including one who wore out a pair of capacitors on the analog board and messed up the yoke connector solder joints (which I had to wait to fix, but I would occassionally turn the 128k on despite the issue because something told me I needed to) and today I find this, which perfectly explains why I knew I needed to turn it on fairly often even with it's issues.

This topic explains it because my 128k has all Apple logo encrusted 4164 RAM chips and to my knowledge has never Sad Mac'ed due to a ROM or even RAM failure. Now I have excuses to both get a Kensington System Saver and use it more both!

And because of that, now I need money! D'oh! I'm a starving (figuratively, but I don't have a job though) college student, where money for non-school expenses is hard to come by.

With the serial number decoder, I can make an educated guess that my 128k was made on either September 30th or October 1st (depending on whether the Fremont factory was open on Sundays or if Apple gave the factory workers Sunday off.)

My serial number is F4410PMM0001, the 803rd Macintosh made in the 41st week of 1984 in Fremont, California USA, and despite the Apple chips (since it's previous owners used it like I want and now need to do!) it still boots up perfectly every time so far!

I don't mean to offend anyone, I am just so excited that my 128k can be kept from developing serious problems by using it, which I wasn't expecting.

My 128k works like a jewel, what more can you want in a computer...

Thanks all,

Alexander Harris