Compact Mac theory

[Apostrophe]

I have a theory that compact Macs, along with all computers, can continue to work no matter what, even if they fail. What I'm saying is, no matter how many vital chips cease to be available, they can still be obtained.

Here's the base of my theory: what has been made by mankind can be repaired and/or made again by mankind. Discontinued dates mean nothing. It was made once; it can be made again. As people master brand-new technology, as they become so familiar with it that they can work it in their sleep, then not only is mastering 20-year-old technology possible, but it's included in the package! All new technology is the old technology, just with more added onto it. But we can agree that whether right off the assembly line or 25 years old, all computers have a CPU (or to expand on that, a motherboard); they have a hard drive (or some sort of system disk); they have an operating system; and they have capacitors and chips that make up their 'brain.'

So the question is: what does this have to do with obtaining parts that are no longer available? What has been made by mankind can be repaired and/or made again by mankind. Yes, I mean that we can quite literally make the chips, capacitors, or anything else that is needed. Because now, with all this advancement in computers, it should be positively easy to make a processor that was introduced 20 years ago. You'll need the knowledge on what you need to make one, as I'm sure that some sort of special equipment is needed, but with all of this new technology and all these new developments, how hard can it be to make a chip that is relatively 'simpler' than what is made today? And, of course, if companies deleted the designs for these chips, what better solution than to simply open it up, carefully examine it, and make a copy? If a company has discontinued something, it doesn't at all mean that all hope to get it is lost. And I have no doubt that some people in this forum have the knowledge on these chips and processors, about how they're made and what is needed to make them.

Of course, I can't test this theory, nor will I be able to until a malfunctioning part is really and truly unavailable for replacement. Because then it wouldn't be a theory, it would be fact. And that instance won't happen for a while, since not only do Macs last, but most parts are still available for replacement from companies and websites here and there, such as capacitors or RAM.

But we all know that it won't always be that way, and that even now there are parts in a compact Mac that are really hard, if not impossible, to come by. And when that happens, it'll be time to see whether or not this theory was well-founded.

-Apostrophe

 

[MacMan]

Technically this theory is correct but realistically, re-manufacturing very obsolete computer parts would be difficult. The majority of technology is driven by the same force that drives pretty much everything in the modern world - money. No modern company is going to be interested in making parts for older machines if there is going to be no profit made from it - they're only interested in themselves. However, fortunately we will always have enthusiasts who have enough drive to keep these machines alive and help their fellow enthusiasts by supplying replacement parts for little or no financial profit.

So yes, it would technically be possible to produce components for 20+ year old compact Macs if someone or a group of people possessed the right equipment, knowhow and were interested in preservation rather than personal financial gain.

An interesting expample of this is vintage cars. In recent years, certainly in the UK, a few vintage vehicle clubs have started producing replacement parts for cars from the 1920's and 1930's. These parts are made in the traditional way and are sold to make a small profit for the club but primarily to help the members to keep their cars running. Some have gone as far as improving on the original design of a part by using a different material for example. Many good, original replacement parts have vanished so vintage car enthusiasts are going to be relying on these modern parts more and more. I reckon that this is how it will happen with vintage computers too in the future. Who knows, perhaps even the 68KMLA will have people producing 68000 processors or 30 pin SIMMs for their fellow Mac enthusiasts.

 

[Quadraman]

The problem is that there will always be a finite supply of parts available which makes each liberation VERY important. Every machine that goes to the crusher today means fewer repair parts for the future, which is why I never pass up any machine that comes my way. I remember reading in the vintage video game community that certain parts of the Mattel Intellivision system are becoming extremely difficult to find. Eventually we'll all end up with a bunch of dead vintage Macs with not enough repair parts to go around.

 

[Apostrophe]

Exactly. There won't be replacement parts to go around. But when I say make them, I mean quite literally sitting down at a table, pulling out some solder, plastic, a laser...everything needed to make a chip. And, using the information on how to make it, put it together piece by piece. Which is why, according to my theory, we can't run out of Mac parts, since all we have to do is get the equipment as needed, and especially the desire to make these compacts work, not financial gain, as MacMan said. And there are people like that in the world, apart from me and most people in this forum. Despite all the greed and personal interest, there are people who will spend money and effort getting these compacts to work.

 

[Quadraman]

You wouldn't be able to set up a chip foundry in your basement. You need absolute sterility without a speck of dust anywhere and need to wear a bunny suit to keep contamination from your person down. You would have to farm the job out to a real chip foundry and in small quantities you wouldn't get a very good per unit price, if you could get someone to take the job at all.

 

[luddite]

I mean quite literally sitting down at a table, pulling out some solder, plastic, a laser...everything needed to make a chip.

I think you underestimate the complexity of chip manufacture. You're unlikely to have much success making silicon wafers in your kitchen, and photolithography on such a microscopic scale would be a bit problematic. The investment required would be prohibitive.

It may be possible to find a manufacturer capable of and willing to do small quantities, but then you have a few legal hurdles ahead of you.

 

[Apostrophe]

Well, yeah...I know it won't be easy...but my whole point is, with the right effort and resources, it can be done. Yes, maybe the conditions have to be perfect, but it can all be set up. Quadraman, I actually could set up a chip foundry in my basement, but if I were to do that I would have to make it absolutely sterile, as you said.

And yes, small amounts won't be a problem, because mass manufacturing chips would be overdoing it, since unfortunately not many compact Macs are around anymore. Everyone has a PC today, but I only know of 5 compacts in my entire city! (Which isn't bad, but still, PC's outnumber them 1,000,000 to 1)

So yes, there will be 'hurdles' to get past, but if you have the will, effort, and determination to get these parts made, then it can be done.

-Apostrophe

 

[Quadraman]

Well, yeah...I know it won't be easy...but my whole point is, with the right effort and resources, it can be done. Yes, maybe the conditions have to be perfect, but it can all be set up. Quadraman, I actually could set up a chip foundry in my basement, but if I were to do that I would have to make it absolutely sterile, as you said.
And yes, small amounts won't be a problem, because mass manufacturing chips would be overdoing it, since unfortunately not many compact Macs are around anymore. Everyone has a PC today, but I only know of 5 compacts in my entire city! (Which isn't bad, but still, PC's outnumber them 1,000,000 to 1)

So yes, there will be 'hurdles' to get past, but if you have the will, effort, and determination to get these parts made, then it can be done.

-Apostrophe

You do realize that running a cleanroom isn't cheap, right? Most are pressurized like an airliner cabin to prevent outside air from getting in. You also need an airlock with an air shower to blow particles off the clothing of anyone who enters along with air filtration systems down to .5 microns.

 

[Apostrophe]

What I just posted was a theory. A general theory that won't be able to be tested for a long time. So yes, there'll be all sorts of problems, esp. involving money. But my whole point is, whether you set up your own (expensive) chip foundry or else have a company do it for you, the whole idea of my theory is that the conditions can be replicated to make the appropriate chips, processors, or whatever parts you need replicated at that time.

But as I said, this theory won't be able to be tested for a long time, since the main component failures in compacts are the ones that can be replaced with ease and little money, such as capacitors. Then, if the hard-to-find chips need to be replaced, then we can know exactly the kind of the problems it could present and deal with it likewise.

I guess you could buy the chip company.

-Apostrophe

 

[JDW]

Well, someone please put this theory to practical use and bring us a SE/30 grayscale video setup for under $100. :b&w:

 

[II2II]

I'm going to say, no it will not be practical. At least for almost anything from the 16-bit era onwards.

Ideal case: you can get a datasheet on a chip and, assuming that the datasheet is correct and that there are no bugs in the chip, you can reimplement the chip. You do the chip foundary bit, or you use a programmable chip like an AVR to emulate the original chip. No huge fuss.

The next group of chips are things, like EPROMs, that are programmable. You can get datasheets here, for the chip. You may be able to get developer documentation on some of those programmed chips in some machines, but very few chips will be fully documented byte by byte. Which may be important if programmers used undocumented entry points. So if the ROM fails and you don't have access to another ROM (or a copy of its contents prior to the fact), you are SOL. As added fun, there is no particular reason why every used address in a ROM (or any address for that matter) has to be addressable for the CPU. So an image of the ROM made for emulators may not help.

Yet another group, well, they are programmable and have state information stored on the chip. (If I recall correctly, PALs and GALs fit in this category.) So it has the same issues as those EEPROMs and you cannot relaibly make a copy of that chip. At least, not without cutting open the chip -- and I don't know what you can determine in that case.

A final group that I can think of are programmable chips with internal security mechanisms. Mechanisms where, if a bit is set, it becomes impossible to read the contents of the chip. I've heard of this sort of thing for PICs and AVRs, but I don't know how common it is with older chips.

 

[Scott Baret]

This is exactly why I take in old Macs I come across. Yes, I have several unused Classics and SEs sitting around but they could prove to be valuable if a) one of my in-use ones dies or B) someone else needs a part that only I have.

If I get a marginally useful Mac (like a Classic I got from a friend last year that had a flaky analog board) I will usually let it sit until I need a particular part. For example, I may pull the floppy from that one or strip it of its RAM if I need either of them.

Don't forget that hard drives fail too. Especially old ones. When you get a new Mac in, make sure you use a good drive wiping program on the drive to sort out the bad blocks--the "Erase Disk" command won't cut the mustard.

 

[tomlee59]

What I just posted was a theory.

Well, of course you are correct in principle -- given enough time and money, you can certainly replicate the state of technology c. 1986. But that's not the question, unless you are engaging in purely academic musing. The real question is one of cost. The paradox of advances in technology is that building old chips in a new fab is not directly feasible. The recipes (and dimensions) have changed so much that it is impractical to undertake such a task for under, say, 10 million dollars. That's how much the first chip would cost; the rest would be free. But you have to pay for the first one to get the second.

The best option would be to arrange a deal with the few remaining legacy fabs. Many military contracts specify very long term supply guarantees, so there are a few old fabs still in operation, but they are disappearing rapidly. Get your orders in now!

 

[Bunsen]

FPGAs FTW!

 

[tomlee59]

FPGAs FTW!

I'm waiting for someone to put the Mac Plus on a Xilinx FPGA. Now *that* would be fun!

 

[trag]

What I just posted was a theory. But my whole point is, whether you set up your own (expensive) chip foundry or else have a company do it for you, the whole idea of my theory is that the conditions can be replicated to make the appropriate chips, processors, or whatever parts you need replicated at that time.

Your theory boils down to, "Anything made once can be made again.". Which is self-evidently true, but not necessarily useful.

A bigger problem than clean room conditions, yadda, yadda, yadda, is the toxic chemicals involved in chip manufacturing (e.g. hydroflouric acid).

A smaller problem (for a billionaire) is simple economies of scale. The cost to tool up to make a chip is huge. In the millions, generally. If you're only making 1000 copies of the chip, or that is the total demand, then congratulations. You've just produced chips which cost several thousand dollars each.

 

[ealex79]

FPGAs FTW!

I'm waiting for someone to put the Mac Plus on a Xilinx FPGA. Now *that* would be fun!

People did that for the AMIGA not that particular thing, but just an AMIGA on a chip.

 

[trag]

Well, someone please put this theory to practical use and bring us a SE/30 grayscale video setup for under $100. :b&w:

Hmmm. Not terribly likely. Not because of the design obstacle, although that is certainly holding everyone back, but the economics of such a project make the under $100 part fairly unlikely.

Let's assume a run of 200 units, which is enough to get the low-hanging economies of scale and still be within the realm of possible sales volume. Also, this card will do an external color monitor as well as an internal gray scale modification, because, really, it adds next to nothing to the cost and complexity.

If it can be done in a 4-layer circuit board, then the PC boards will cost about $20 each. The FPGA will be $10 - $20 each. The 120 pin Euro-DIN connector will be $4 each. Some kind of Flash on board (for the firmware) is $1. The memory will be...really hard to predict. There are a lot of choices and the market keeps changing. Let's call it $5. The D to A converters will add another $6 - $12 and then the handful of other components adds another $10. That's (taking midpoint costs) $63 just in materials and may be a low estimate (or high if the builder lucks into cheap lots on Ebay). So up front, the maker would be investing $12,600, which he or she will want back.

Oh, but that's just for the frame buffer video card. You still need the gray-scale adapter yoke board. I have no idea what those cost to produce because I do not know the cost of that connector for the rear of the CRT. I would guess in the neighborhood of $20 in materials. And we haven't allocated anything for wiring harnesses which could add another $10 easy but possibly $20. Those molex connectors really add up. And have you ever priced wire?!? It's shockingly expensive.

So, you're looking at $18,600 up front with a material cost of $93 per setup.

Now you could cut that initial investment by paying a lot more for each setup. The circuit boards can be had for $55 each in small quantites (like 4 at a time). So now the total material costs might be something more like $150 per board, but the initial investment could be closer to $500 - $1000 for materials, prototypes, etc.

Now, if you have $18,600 sunk into 200 of something up front, which you must sell for at least $100 each in order to cover your costs, but you're not certain that there are even 100 people who will want one, how do you price it?

Or do you make the $150 version a few at a time? If so, how do you price them?

This example pretty well illustrates the frustrating economics of all these projects.

Even my IIfx 16MB SIMM project would have been uneconomical if I had not found incredibly cheap 72 pin SIMMs from which I could steal the memory chips and had a $500 off coupon at 4pcb.com. If I had had to buy the memory chips loose I could not have afforded to make them and noone would have been able to afford to buy them.

And even then, I had to design a special board which could run off of 16M X 4 chips because SIMMs with 16M X 1 chips are relatively rare and expensive.

And that solution meant hours of extra effort recovering and cleaning the memory chips. That was okay while I was unemployed, because I had the time. Now that I'm happily slinging electrons again, it wouldn't be practicle to make that time for money trade.