16/6000 MMU question
steve scrivano
man page for the Xenix configuration kit ( a separate software package
for those that want to build their own kernel), the following statement
is made:
Installing more than 1 meg of memory requires the 6000
MMU (26-6020).
What is a 26-6020? Is this the stock number for the Model 6000 68000 card,
a component that went on the board to expand memory management, or?
I have the old catalogs for Tandy and this number was never mentioned.
There is a: 26-6022 Model 6000 with 15 meg drive
26-6021 Model 6000 with 2 floppy drives
26-6019 Model 6000 Memory kit
26-6014 upgrade kit to a Model 6000 which includes the 68000 and
memory board.
Notice how the 26-6020 stock # is NOT in the series of hardware stock #'s
mentioned in the catalogs, however, the number series seems to fit in
somewhere that would leave one to believe that this was an item that had
been around from the very moment the Tandy 6000 was sold.
Of course the best way to sell a product is not to tell anyone that you
have it for sale. Nobody told me about the product and I have owned these
machines since 1985.
Tandy was well known for putting out special bulletins containing stock
numbers for upgrades for their products that were never seen by the general
public and only mailed to registered owners of these machines. Software
packages like the kernel configuration kit and this MMU are two examples
of items not to be mentioned in any product catalog and it really makes it
frustrating when doing historical research of a product line.
Frank Durda if you are reading this, maybe you are the only left that can
answer this question?? What the heck is a 26-6020 MMU in layman's terms?
--
Steve Scrivano
sscr...@nyx.cs.du.edu
F. Barry Mulligan
memory management unit in Jan '89. Stock number AX-0245 was a modified 68000
CPU board, available only on special order through your local repair center.
You surrendered an 8MHz CPU board and $324.99 ($384.99 if it wasn't working)
and got the modified board in exchange. It was not available without a swap,
and a 'pal chip' 6MHz board was not acceptable.
In a further note he reported the rumor that there were 100 of the
modified boards, with the exchanged boards being recycled through the system
as they came in.
At a guess, the 26-6020 catalog nbr was assigned before they knew how
the MMU was to be implemented.
/* barry /&
Frank Durda IV
[0]What is a 26-6020? Is this the stock number for the Model 6000 68000 card,
[0]a component that went on the board to expand memory management, or?
It is a small board that plugs into the 68000 CPU board in place of the
68000 chip, then the 68000 chip plugs into the 26-6020 board.
[0]Notice how the 26-6020 stock # is NOT in the series of hardware stock #'s
[0]mentioned in the catalogs, however, the number series seems to fit in
[0]somewhere that would leave one to believe that this was an item that had
[0]been around from the very moment the Tandy 6000 was sold.
That's because they only built 150-300 of them because the guy in charge
decided that was the more than he would sell and just enough to cover
"the development" costs. Read on to find out about those "development
costs".
The MMU extension was developed by four employees at Tandy in their
spare time after Tandy turned down the project. (Reason: it isn't PC
compatible(TM).) When these people went to the VP of R&D for a release
so they could sell their MMU themselves (they already had working printed
circuit boards and kernel changes at this point), the VP demanded to know
why Tandy wasn't offering the product itself. The lower-level director
had to do an about-face and embrace the project he had denied earlier.
So the request to privatize the MMU was denied, and Tandy sold the product
themselves. The developers ate the cost of the boards they had fabricated
and were not supposed to make anymore. Tandy took the design, made
a few changes to simplify manufacturing and off it went. To hide some
other expenses, a portion of the development costs of XENIX 3.2 were
burdened into these 300 boards, inflating their price further.
Keep in mind that at this point the buyer for the 6000 was also the
buyer for the 1000 line and he was out with two shovels trying to bury
the 6000. He had already killed all 6000 advertising. If he could get
away with it, the 6000 wouldn't even appear in the catalog. He also
cancelled dozens of software stock numbers in an effort starve the 6000
out of existance. His superior, the "boy genuis" of Tandy marketing was
also trying to kill anything that wasn't DOS or DESKMATE, so there was no
court of appeal. (The boy genius also ran the VIS project. Any questions?)
When I went to buy my 6020 MMU board, I had to go to the parts depot to
get it, after calling four or five people to find out where it was.
The big reason why it was so hard to get the MMU was that Tandy marketing
and technical support (not R&D) decided it was *too* complicated for users
to install the MMU themselves and so the customer had to turn in their
68000 CPU board and get one for exchange that had the MMU already
installed.
There was one cut and three or four jumps that had to be done on
the CPU card to support the MMU. This brought the extra MMU signals up the
68000 socket on some pins the 68000 chip didn't use. There were no flying
wires in the design.
So you had to order the CPU+MMU board set, unless you knew how to
get just the MMU from the repair depot.
I believe I have copies of the installation instructions we wrote for
the customer (never published) and the ones that the technical support
did for the repair depot to use in modifying boards.
The MMU itself consists of a thru-hole socket, a header to plug into
the 68000 CPU socket on the CPU board, and four or five 74xX ICs, plus
one 7nsec PAL. That's it. Tandy sold it for around $300. The
material cost was under $30. And don't forget, Tandy didn't do
much development on it since it had already been done by other people
on their own time!
This MMU expansion would address up to 4Meg of RAM by adding two bits
to the base and limit registers and comparators. The changes were
put into the XENIX 3.2.0 kernel and they lay dormant until more than
1 Meg of contiguous memory is detected.
We actually put four 1 Meg boards in a machine to verify it worked, but
found the buss signals were so noisy that I believe the documents went
out saying only three cards were supported. There was another vendor
with a memory board that had 2Meg per board, and two of those worked fine.
Above four meg could still be used for buffers or a RAM drive.
[0]Of course the best way to sell a product is not to tell anyone that you
[0]have it for sale. Nobody told me about the product and I have owned these
[0]machines since 1985.
Well, it was certainly discussed in the Tandy owners group (I forget
the name now) and Bob Snapp went on and on about it. All the big
accounts certainly knew about it.
[0]Frank Durda if you are reading this, maybe you are the only left that can
[0]answer this question?? What the heck is a 26-6020 MMU in layman's terms?
Was that OK? :-) If someone *really* wants a schematic, it is certainly
possible to dead-bug one onto the CPU board. It won't be pretty, but
it probably will work. But you are certainly on your own on this!
Frank Durda IV <uhc...@nemesis.lonestar.org>|"The Knights who say "LETNi"
or uhclem%nem...@trsvax.ast.com (Internet)| demand... A SEGMENT REGISTER!!!"
...letni!rwsys!nemesis!uhclem |"A what?"
...decvax!trsvax.fw.ast.com!nemesis!uhclem |"LETNi! LETNi! LETNi!" - 1983