> There's a big difference between writing code to solve problems and being
> a software engineer. Designing, coding, and compiling is only 40% of the
> battle. Hopefully you're also spending some time planning and testing.
To me planning and testing are such obviously important parts of any
development process that I didn't mention them in my original post,
assuming (I hope correctly) that most everyone on this list is as
old-fashioned as I am ;>)
> We recently interviewed an electronics engineering graduate who didn't
know
> the difference between NPN and PNP transistors !! What do they teach kids
> these days ??
I repeat, the "faculty" of the "institution" which issued this fellow a
degree should be tarred, feathered, and run out of town on a rail.
Glen
0/0
From: Gene Buckle <geneb(a)deltasoft.com>
>What is so bloody difficult about this? It's not like I'm asking
>questions about quantum mechanics or something. Keerist.
No one knows or has seen one.
>I had two questions. The one about the SCSI interface has been answered
>to my satisfaction. The second however is not only the easiest of the
>two, but seeming the most difficult to answer.
Seriously it's been about 14 years since I've seen one. It's length was
long enough to reach with neat folds. Don't know the exact number of
inches. However that FDC controller worked fine with 24 inches of
cable for another project.
>If you happen to own a Series 100 box, I'd be most appriciative if you'd
>open it up and let me know what you find out about the cable. If you
>_don't_ own one of these things, I don't want to hear from you. Period.
I bet you dont turn up many Series 100 boxes.
>I know how to build my own cables. I know what connectors are required.
>(I even have a purpose built connector compression tool!) I'm trying to
>restore this machine to _factory_ condition. Smart ass comments about
>making my own cables (while simultaneously alluding that I don't know
>what the fsck I'm doing) is not only not wanted, but just pisses me off.
>If I'd wanted to make a generic, "will do the job" cable, I wouldn't
>have asked about the original one!
The Ampro cab was not common as most bought the dry board being
cheaper and use whatever package was handy or one that met their
spec. Even my Ampro LB manuals do not specify a length!
>FYI, the machine is built for CP/M. It runs CP/M quite happily. Even
>with ZCPR3.
We know that too. I run on here.
Allison
From: Ben Franchuk <bfranchuk(a)jetnet.ab.ca>
>I know that. That is why the early memory chips had separate in/out pins
>to emulate core memory. Everything nowadays tends to be 8 bits with
>tri-state
That might have been a factor but not a requirement. The reason back
then
was it simplified the timing and construction of the chip.
>I/O ( Grumbles here as he has a 12 bit computer and has to waste 4 bits
>out of 16 ) >BTW - some say the best way to cook a fish is in the
dishwasher.
Use four bit wide cache parts then. Or if it's EPROM do three bytes wide
and
select the odd(right 12 bits) or even half(left 12bits). There are other
schemes
to reuse loose bits!
Allison
-----Original Message-----
From: Ben Franchuk <bfranchuk(a)jetnet.ab.ca>
>
>After reading the core memory stuff I realized the what I was thinking
>about was dynamic memory (16k chips) for a 2 mhz 8080 in byte. The
>reason
>I got confused was it used the split memory cycle read-write as core
>memory.
Core is one of the few destructive readout memories used. So every
read has a following write to restore the data, often between data
read there will be a modify cycle which means new data written back.
Allison
Suggestion,
Pict the articles on the basic of historic design point like
the Core article or some significant hardware or software.
Usually that kind of article was a feature.
Allison
-----Original Message-----
From: John Allain <allain(a)panix.com>
To: classiccmp(a)classiccmp.org <classiccmp(a)classiccmp.org>
Date: Sunday, December 16, 2001 1:42 PM
Subject: Re: [PDP8-Lovers] how to clean a PDP8/A, dishwasher?
>> Unfortunately these scans aren't from *my* bytes...my collection
>> doesn't go back that far, as I was 7 years old in 1976...so I can't go
>> scan more of 'em. :-(
>
>I guess I can pony up and put at least one jpegged BYTE article up
>for people to see, perhaps one a month. I have no great ideas on
>how to pick the titles, however. Some kind of democratic thing?
>
>John A.
>
>
From: Richard Erlacher <edick(a)idcomm.com>
>I'm curious about something here ... Was this particular system ever
observed to
>run as configured? I've never owned a prepackaged AMPRO box/system, so
I have
>no experience with them on which to base any guesses. With a SCSI drive
and no
>bridge, the thing should either run or not, though, and if it doesn't
there's
>probably a good reason.
Over the years I've seen two and added hard disks to them (purchased
without).
If you have the right controller and drive it's an assemble the peices
project.
Those included some of the early SCSI drives.
>I'm not in a good situation to help you out with this, at the moment,
but I
>could send you a bridge controller of the sort that the firmware is
supposed to
>recognize on its own. With that you can test the Little Board side of
things.
>The Little Board supports the Xebec controller as well as a number of
models
>from OMTI, ADAPTEC, and others. You'd have to attach an
ST506/412-interfaced
>drive that works properly at the other end, though. The Adaptec models
I've got
>are capable of either RLL or MFM, depending on which model you use and,
of
If you match the known configs it's pretty straightforward, if not you
need to go
into the bios and set up the config by hand. The latter is a bit less
obvious as
the Ampro BIOS for hard disk was done in three layers, one for the
physical
SCSI driver, then the SCSI protocal for the controller target and then
hard disk
interface. There are several added tables not normally seen in a CP/M
bios.
Those tables allow for things like assigning logical unit 3 (whatever it
may be) as drive A:. As BIOS for CP/M go it's very sophisticated.
Allison
>> > From: Gene Buckle <geneb(a)deltasoft.com>
>> Funny, the Ampro docs list all these various controllers and whatnot.
>> The software also _asks_ what controller it is going to talk to. The
>> SCSI controller is part of the LB. What would the software consider
the
>> onboard controller to be?
NCR5380! But that's half the picture. The other half is the SCSI
controller on the other end of the cable. What I meant by PC logic
is that in PCs you plug in an adaptec controller or whoever and you
need a driver specific to that controller. The AMPRO-LB has the
onboard NCR5380 which is the controller and the software specific
to that is in both the bootrom and CP/M bios.
The other peice of that that does not conform to the PC standard
is the other end of the 50pin SCSI cable is the "drive" and back then
the drive was MFM (ST225 typically) with a Bridge adaptor to go
>from SCSI to MFM and that could be done with a board made by
Xybec, Adaptec or Western Digital. Now, those bridge boards did
not behave exactly the same like modern SCSI drives so the "driver"
had an install/init that tweeked the CP/M bios to conform as well as
partition the disk for CP/M use. The CP/M Bios (and bootrom for
that fact) do not query the drive for it's config like PCs. It must be
hard coded into tables in the bios. The bridge controllers require
this info to be pushed into them before they can access the
physical drive. This is in contrast to SCSI drives that have the
tables integral to themselves.
FYI: CP/M supported only 8mb per logical drive so a ST225(20mb)
was usually partitioned into three logical drives (for ampro that would
be A, C and D). Drive B is reserved for floppy as is pseudo drive E.
Drive A(physical drive 0) is always the bootable drive and can be
either floppy or hard disk. The tables for logicla drives are limited
to 16 logical drives (A through P) by CP/M itself and if memory
serves only 10 logical hard disk partitions in the Ampro HD
CP/M bios. So any drive larger than 80mb cannot be fully
accessed. There are ways around the latter problem but
that's a different issue.
>What hard-drive are you using? If it is a SCSI drive? Is it set to
>ID=0? If so, choose either Seagate or Maxtor and give that a try.
I used a 45mb fujitsu. The boot roms expect it to be ID=0.
>> > formatting. If you dont mind hacking Z80 code you can go frm
>> > an AMPRO LB with SCSI to a smaller SCSI drive (64mb or less,
>> > or the rest will be unused).
>>
>> I've got a 20MB SCSI drive I'm trying to use. The bootrom rev tells
me
>> that it's got support for the SCSI controller. What coding has to be
>> done?
CP/M bios which is not the boot rom. The reason for that is the Drive
tables and the controller tables in the install(HDINIT) and format(HDFMT)
will not work for all but a limited subset of possible drive/controller
setups.
The only SCSI drive directly supported was the ST157N and maybe the
ST296N might work.
>If you have the proper version of the HD Utilities, you should not need
>to do any.
> - don
Only if you have a supported controller and drive, otherwise they
do not help you much.
Allison
On December 14, Chad Fernandez wrote:
> The Post Office still uses Unisys stuff, although, I think they may be
> switching to IBM. I saw the Portage (next city over) PO had new looking
> IBM stuff a few weeks back.
>
> Unisys had/has a govt. division of the company.
The post office in Beltsville, MD just switched over to that
Unisys stuff maybe 2 years ago...
-Dave
--
Dave McGuire
St. Petersburg, FL "Less talk. More synthohol." --Lt. Worf
From: Dave McGuire <mcguire(a)neurotica.com>
>> Not likely as the sense voltages are quite small and the slice time
>> has to just right.
>
> Well I wouldn't expect to connect it *directly* to the PIC...some
>analog jellybeans would seem appropriate.
Look at the byte article and see why something like a PIC adds little
to the task.
Allison
> I wonder if it would be possible (and practical) to use a
>microcontroller, perhaps a PIC, to act as a core controller. Use the
>A/D and D/A hardware to handle the drive and sense stuff, and do all
>the timing in firmware...making it easily tweakable.
Not likely as the sense voltages are quite small and the slice time
has to just right.
> That would be something I'd be up for trying...if I can find a chunk
>of core of low enough density to trace the wiring in. There are some
>nice low-density planes on eBay right now, but they are priced WAY too
>high in my opinion.
No need to trace, they are quite regular, you can ohm them. You still
need core drivers, sense amps and a dozens of diodes for current
steering.
Allison
