• Please review our updated Terms and Rules here

Apple //e wikipedia page

I

iz8dwf

Veteran Member
Joined
Mar 20, 2017
Messages
812
Location
Italy
Hi all,
I'm not exactly an Apple expert, but I surely have a long enough record of repairs (starting in the middle 1980s).
Now, the Apple //e wikipedia page:

https://en.wikipedia.org/wiki/Apple_IIe

currently states:

"The IIe also switched to using newer single-voltage 4164 DRAM chips instead of the unreliable triple-voltage 4116 DRAM in the II/II+"

I think the switch was purely dictated by the chip "density" of 64kbit VS 16kbit and not even the triple supply VS single supply issue, since the Apple //e also had the same PSU as the previous models. I'm quite sure the 4116 isn't unreliable per se, the power supply must be indeed more reliable if one has 4116 DRAMs, but that's another issue entirely.
Facts: I repaired several (> 10 so far) computers that use 4116 (PET 2001N, PET universal boards, ZX spectrum, Apple ][, arcade PCBs etc...) and the current total number of failed 4116 I have found is 3 and this is an unfair number too:
One 4116 was split in half from the beginning, probably because of bad handling during shipping.
One 4116 failed on the same data line of an already failed 4116, that might be because of the already failed chip driving the data line when not selected.
So, in my experience, 4116 are indeed at least as reliable as the much denser 4164 (that I substituted far often, but on a larger sample of repairs too).
What do others think?
Frank IZ8DWF
 
I consider my experience quite anecdotal, rather than scientific. But I find what you have found: 4116 seems much more reliable than 4164. I don't think I've ever replaced a 4116. I've replaced dozens of proven-bad 4164s over the years.
 
KC9UDX said:
I consider my experience quite anecdotal, rather than scientific. But I find what you have found: 4116 seems much more reliable than 4164. I don't think I've ever replaced a 4116. I've replaced dozens of proven-bad 4164s over the years.
Click to expand...

Experience is experience. It might not be statistically "large enough", but it's still valid as "single data points in statistics".
I still don't believe that calling the 4116 "unreliable" is scientific, unless who wrote that can point to proven facts that contradict my (and yours) "few data points".

Frank
 
Either way, "unreliable" is a bit of an opinion.

I'm sure the Apple IIe moved to the 4164 for a variety of reasons. Such as smaller space used, availability, and cost. If there were any reliability issues, they were probably secondary considerations. And of course, 4164 simply was single voltage. The way it is worded makes it sound like there is a direct link between the voltages and reliability.

Although the voltages could result in bigger bangs and more smoke when a chip did fail. :D
 
SomeGuy said:
Either way, "unreliable" is a bit of an opinion.
Click to expand...

Unreliable is never an opinion.
Most of us know that MOS technology produced some unreliable ICs, since a large part of them failed in computers where all the rest of logic chips (not produced by MOS technology) haven't failed or have failed with a much lower rate. Not all MOS chips are unreliable: 6502 are usually very reliable, on the other side there're the 7501 and 8501....
Micron Technology 4164 DRAMs are another example, they have a much larger failure rate with respect to all other 4164 DRAMs of the same era, thus they are unreliable.
4116 do need tree supply rails (and they can be damaged if -5V is missing or applied in the wrong sequence), that doesn't mean they are "unreliable", it means the designer must take this into account. All 4116 based computers that I've repaired didn't have any high DRAM failure rate, 4164 have a worst failure ratio (entirely thanks to mT ones).
Of course the Apple //e redesign would benefit of the denser 64kbit DRAMs for a lot of reasons (even easier refresh circuitry could justify that alone), I just think its not fair to call the triple supply 16kbit DRAMs "unreliable" as there seems to me that there's no proven record of this unreliability.

Frank
 
The early 16K DRAMs were nothing to write home about. In my case, my experience was with the Intel part (2117). The yield on those was quite low, to the extent that Intel offered 2109 8K chips, where one section contained errors. A number suffixed to the part number indicated which half of the chip was to be used. I still have a few of those.

My take is that the 4116 probably had the same issue, such that the ones that made it out into the field were more rigorously tested than their 64K brethren.

I also suspect that there were a lot more 64K DRAMs manufactured than 16K ones.
 
I think when it comes to reliability, with cmos or early MOS technology, it has an awful lot to do with how the chips are handled. Much of the electrostatic damage is not initially evident and they have a habit, if damaged, of dying early in service. So an individual's experience of their reliability turns out to relate to other peoples involvement and handling of the IC's before they were fitted to a pcb, Probably, if you could get them right out of the factory, with no third party handling, treated them with extreme precautions, they would seem more reliable !

I read a remark from a engineer from Silicon Valley in the early 1970's, they used thousands of 74 series TTL's around that time, he said they found 1 in one hundred were faulty from new. In my experience that is not the case, its far less and I have used a few thousand of them. TTL's of course are substantially more robust than cmos to electrostatic damage. I also think as the years went by manufacturers got better with their internal clamping diodes to protect the inputs of cmos IC's. Also, I have yet to find even one faulty 54 series TTL new old stock IC.

Once a cmos IC is wired in circuit, its a lot safer, but its what happens to it on its way from the factory to the pcb it lives in, that probably largely determines the supposed reliability of the part, unless the circuit it is in pushes it past its design parameters.
 
Another aspect is the packaging. Early 16K DRAMs were almost universally CerDIP, with later, cheaper ones being plastic DIP. My guess is that the ceramic ones have fewer problems wrt longevity.
 
You must log in or register to reply here.
Back
Top