[rescue] Slightly OT: ?Bad Cap Saga

[Geoffrey S. Mendelson]

On Thu, Aug 21, 2008 at 02:53:59AM -0400, der Mouse wrote:

>> At what point does it become impossible to source the chips?
>
>Well, strictly, never; it jsut becomes uneconomic.

It depends upon what you mean by unbeconomic. At some point, it's not
just making something that isn't made, but you have to go back and
re-engineer the device, replace old technology with new and so on.

What has kept things going for so long is that TTL (transistor-transistor
logic) has been used since the 1970's and is basicly unchanged. But what 
about all of those other logic families that are long gone, such as RTL,
DTL, etc?

Things also took a giant leap away from compatibility in the last few
years as you can no longer manufacture or sell lead based solder in
most of the world. Old soldered chips can still be sold, but if you had
to make a batch of new ones, you have to use a different material and that
changes all sorts of things. The new solder melts at a higher temperature,
has a habit of growing "whiskers" and shorting out, and so on.

What's going to happen when TTL is replaced, which it eventually will be?


>I don't believe that.  What I believe _is_ the truth is that they would
>require substantial hackery, probably needing both hardware and
>software hackery.

But then it's not the same. Old software won't work with the new hacks and
so on. If you are going to hack that much, why not hack a control board
that fits in a modern computer, or use one of the single board computers
sold for process control or experimentation?


>I have no idea, since I don't even use, much less hack on, either.  I'm
>fairly sure NetBSD does (or at least easily could); they call them
>"bounce buffers" and they use them for very analogous but more common
>situations, cases where a DMA device can access only part of the
>system's RAM - this is not as unusual as situation as you seem to think
>it is.

It's not unsual, but it's become less and less. 

>
>> Could they still be written?  If so, by whom?
>
>Any good kernel hacker for the relevant OS, of course.

Unfortunately most of them won't. They either have no interest in what
you want, or are busy doing other things. Remember the device support in
the 0.9 era Linux kernels? It was limited mostly to devices that people
interested in writing device drivers had on hand. If a card cost $10 and
could be bought at any computer store, someone would write a driver for it.

If it cost $1,000 and had to be ordered from the manufacturer, there were
no drivers for it until someone needed the card, and paid for it.

Considering the wide availability of process control (aka embeded) 
microprocessor combinations, (processor, RAM, ROM and ports) on cheap
chips with well known programing specs and toolkits, it hardly makes
sense for anyone who wants to experiment to try to get working a port
which has been all but abanonded.

You can buy all sorts of boards, and they are IMHO a lot cheaper than getting
someone to put a parallel port back on a PC, find someone willing to write the
drivers for Vista. If you are going to use Linux, or a BSD variant, it's
very likely you can already find a version for any development board you
want, or easily port it.

As you said before at some point it becomes uneconomic. You want your
cheap PC motherboard to include the port, which not enough people want
to pay for. So you either buy an expensive (in single quantity) embedded
board, or pay a premium for that port to be added to a PC.

Unfortunately for all of us, the bottom line, as it were, is if it does not
run with Vista, it's a niche item. Even the Mac, with it's less than 15%
market share is a niche item, a big niche, but a niche never the less.
Since modern Mac's run windows, they are more of a combination product.

Linux, BSD, etc don't even really count, because no one has ever made a
real inroad on the desktop with them. I've run linux since before RedHat
existed, and UNIX and Macs even before that, so I'm not just saying it. 
They have done well in the server market, but that's a niche anyway.

I'm also a lot more familar with device driver development than you give me
credit for, I've been writing and patching what you would call device drivers
probably before most of the people on this list were born (late 1960's).

Geoff.

-- 
Geoffrey S. Mendelson, Jerusalem, Israel gsm at mendelson.com  N3OWJ/4X1GM