 |
Home page |
|
Documentation top level |
The basic features are:
A motherboard was shipped to people who bought them with a complete hardware description, full schematics, and PAL/GAL programming information. The first general distribution of pc532s included a special deal where all the chips were included in the purchace. Later distributions included only the motherboard and buyers banded together to get all the chips that were needed.
A owner of a pc532 designed a simple daughter board that plugs into one of the SCN2861 DUART slots and implements two Centronics parallel ports.
To our certain knowledge there were less than 200 pc532 mainboards made. There were only 187 pc532 out in the world as of 11 Jun 1996. (source "Matthias Pfaller" leo@dachau.marco.de). There are many less running NetBSD (a couple of dozen is my (jhs) guess). The only way to get one today is to find someone with one they don't want and to buy it from them.
There are two different ROM images. One for a 9600 baud console and one for a 19.2K baud console.
Get the latest INSTALL document, and try to follow the install procedure. Then, before you reboot, you need to get the boot information if you don't have the auto-boot ROM. (If you do have it, you don't need the next information.)
run bim
You will get something like:
---->/usr/src/HOME/phil
steelhead[1]# bim
Disk: disk 1 Type: SCSI
Physical Sector Size = 512
Disk Size = 1173930
partition type sector start length in sectors
a 4.2BSD 23980 383680
b swap 2180 21800
c unknown 0 1173930
d boot 2 2178
e 4.2BSD 451260 722670
f swap 407660 43600
Boot partition = d
Default boot image = 0
Boot Images: total of 8
(image address and size in sectors.)
Image address size load addr run addr name
0 0 32 0x3e6800 0x3e6820 /usr/mdec/boot
1 32 64 0x3bf000 0x3bf020 /usr/mdec/boot
From this you can easily calculate the disk address of the boot program
and where to load it and run it. So for the above stuff, the boot
partition starts at sector 2. The boot image is at offset
0 in the boot partition. So the boot program is at sector
2 and is 32 sectors long. You need to load it at address
0x3e6800 and run at 0x3e6820. Write that information
down for your system and you can then give the proper ROM monitor
commands to load the boot program and run it. All the autoboot monitor
does is to read that information from the disklabel and the boot images
table on the first two sectors of the disk.
Some original mail on the subject:
From: Phil Budne <budd@cs.bu.edu> > Philip Zimmerman found a little (6"x1"x3.25") Cabletron EA419 SCSI to > ethernet box, and I found documentation on how to speak to it. What does that box cost (aprox.) ? It doesn't look like this box is sold any more; However Philip Z. may have a line on two different sources. But in both cases it may boil down to "how much are you willing to pay". Since similar devices sell new for $200, I would hope it would be under $100. ======== From: Philip A Zimmermann <paz@world.std.com> To: Phil Budne <budd@cs.bu.edu> Date: Tue, 2 Jul 1996 00:55:33 -0400 (EDT) I'll look into sources. More may even pop up from the other subscribers out there; what I basically scared up, at Phil B.'s description, was a box which we used to use on older-style Macintoshes which had a SCSI port on the rear panel. This Cabletron box has a DB25 on one side and an RJ45 on the other side and a socket for a wall wart to provide power. I have chased a few leads today; the only positive response was from Cabletron Systems so far. They have 11 in stock right now in their "Good as New" department. They want US$175 each. Order info: (per Mike Orr) Cabletron Systems Good as New Dept PO Box 5005 Rochester, NH 03866-5005 U.S.A. Mike can be reached at: 1.603.337.2108 He said they trickle in occasionally. I'm still searching for other sources. From: paz <paz@world.std.com> To: pc532@bungi.com Date: Wed, 12 Feb 1997 11:59:28 -0500 (EST) The network "card" I'm using is a Cabletron EA419. It's probably the same device you've seen discussed on the mail list. It's a box with an RJ45 on one side and a DB25 on the other. It was originally intended to provide a network interface to Macintoshes with a SCSI interface which came out of the Mac at a DB25, I think. Cabletron has a "good as new" department where refurbs of this unit can be gotten for ~US$150, if memory serves correctly. I'm uncertain how many pc532s are so equipped right now; I just hacked mine onto a Vector ISA prototyping card so it would plug into the pc532 SCSI expansion bus. I got it running over the past weekend. Seems to be working fine.
This report from Julian Stacey <jhs@freebsd.org>
No doubt the class (2) guy might prefer an ICU & FPU too, but nobody's doing a PC532 PCB run so that option is not available Anyway, maybe class (2) folk might increase total board quantity & reduce prices for us class (1) folk ? Is this daft ? or Should we offer it to the Net & Free BSD lists ? Comments Please.
cfg: ether.cfg schema.cfg ser.cfg doc: copeth.doc dramc.doc ether.bom ser.bom dec32.doc dramen.doc ether.txt wait.doc gerber: ether1.gbr ether4.gbr etherdrl.gbr ser1.gbr serslk.gbr ether2.gbr ether5.gbr etherslk.gbr ser2.gbr sersmk.gbr ether3.gbr ether6.gbr ethersmk.gbr serdrl.gbr lbr: local.lbr pals: copeth.jed dec32.jed dramc.jed dramen.jed wait.jed copeth.tdl dec32.tdl dramc.tdl dramen.tdl wait.tdl postscript: ether.s01 ether.s03 ether.s05 ether.s07 ser.s01 ether.s02 ether.s04 ether.s06 ether.s08 ser.s02 schema: ether.p01 ether.p03 ether.p05 ether.p07 ser.p01 ether.p02 ether.p04 ether.p06 ether.p08 ser.p02
The et532 is a 6 layer PC/AT sized (but with 62 pin SCSI) board. The piggyback board (after a bit of a fight with the CAD software) is a 2 layer board and is around 5.5" x 3.5". It plugs into the middle area of the et532 on two connectors that carry all the serial (TTL) signals up from the 2 OCTARTs. There are also 3 sets of holes on both boards that line up and enable nylon standoffs to support the piggyback board.Michael Hermann & Steve Liggett asked stats on PCBS, to get us prices, well I havent counted number of holes on pcb (freq Q. that), but:
No surface mount devices as per PC532 (I assume George ?) I presume we call this `6 layer' as far as PCB manufacturers are concerned ?
> > - Do we also need to make piggy back octart PCBs ? > Only if you want to have more serial ports. Looking at George's ET532 schematics: The ET532 can take 2 * SCC2698, each 2698 supports EIGHT serial ports ! The optional `piggy back' board has 16 * 145406, 16 * DIL Headers, Caps, & 2 Long DILs to connect to ET532 main board. For the ET532 to function just as an ethernet card, it should not be necessary to populate the 2698 sockets, nor to have a piggy back card. I do however foresee need for a few of us to have one serial port, as I guess software development'll probably go in 3 stages: 1 Write/debug eprom for ET532 to talk to a serial port (& prove hardware design works) 2 Extend ET532 monitor to talk over SCSI bus to some kind of software interface on the PC532, so PC532 can download code to ET532 & set breakpoints etc ?. 3 Develop ET532 ethernet code & ET532/PC532 higher level protocols etc. Main thing is to get the ET532 PCBs fabricated :-) . Daniel Berglund <db@etek.chalmers.se> wrote 3-94: > > > It's probably not hard to temporarily wire in a 2681 > > > (or whatever they are called, > > SCC 2698 > Nope. I mean a 2681, they are cheaper. That's my point: if you > want only one serial port for debugging until the firmware is done, > then you probably want to avoid expensive octuarts (2698).
> from the Bay Area.... > Hamilton/Avnet (a distributor for National) > 1-24 25-99 100+ > NS32GX32-ANU20 $110.00 $95.00 $86.50 > NS32532-25 $825.00 $685.00 $595.00Mark_Geisert@l66a.ladc.bull.com said
> Bell Industries, a NatSemi distributor in California, > QTY1-24 QTY25-99 QTY100-999 > NS32GX32U-20 90.00 78.50 71.00 > NS32GX32U-25 100.00 91.00 83.00 > NS32GX32U-30 120.00 105.00 95.00johnc@manatee.UUCP (john connin) Wed, 7 Oct 1992 19:16:47 GMT said:
> it might be easier and cheaper to pull the ethernet > related parts from a no-name 8-bit ethernet card. Are the parts the same as I posted in George's build list ?Could someone please ask at a US West Coast IC shop, what an IC kit price would cost, as per build list (presumably a) excluding GX32, b) 1 off & ~25 off pricing ). ( I believe jkh@whisker.hubbard.ie can provide names of California IC shops for someone in the USA to phone. )
The ether-bom (Bill Of Materials) is available via ftp to enable pricing exercises. Don't ask me, I'm in the wrong country for cheap or easily available prices, VOLUNTEERS PLEASE !
"I have the software for the Ethernet already done, including an Ethernet-based monitor, and DOS client. I'll do the low-level Ethernet driver (tricky, 'cause you have to handle the special cases)."(so I assume you want 3 PCBs bdale ? ) Rennolet (Kumar)" <rennolet@cs.umn.edu> said Wed Oct 7
"I will help write the drivers."21 May 1996: Mark Mason <mason@ease.com> interested in helping.
More can be learnt by looking at the Gerber diagrams. George long ago wrote:
IE "GO"/NO GO" but no WIBNIs (Wouldnt It Be Nice If's) please, Thanks :-)
Please hand forward this notice to dormant pc532 friends who may have fallen off the pc532 list, or changed their mail address.
A = Number of ET532 PCBs you will pay for, ie B = Number of optional Serial piggy back pcbs you will pay for C = Number of glue chip IC kits you will later want. D = Number of GX32 CPUs you will want E = Email F = Real Name G = Date of last communication H = Comment @ = Date person last confirmed PCB order info. # PCBs ; EMAIL shown by "From:" ; .signature EMAIL NAME ; HUMAN NAME ; COMMENT SERIOUS INTEREST in 96 A B C D E F G H 1 1 1 1 jhs@freebsd.org Julian Stacey @May96 $ Dependent 1 1 1 1 eyal@eyal.emu.id.au Eyal Lebedinsky @May96 $ Dependent 1 1 1 1 admin@maxrnd.com Mark Geisert @May96 1 ? ? 0 leo@marco.de Mathias Pfaller @May96 will swap 32532 chip for pcb 1 ? ? ? phil@cs.wwu.edu Phil Nelson @May96 PCB <$100 1 1 1 1 mason@ease.com Mark Mason @May96 PCB <$100 1 1 1 1 n1epo4tl@ibmmail.com Stuart Arnold @May96 cond. on $ 1 ? 1 1 sverre_froyen@nrel.gov Sverre Froyen @Apr96 cond. on $ ? ? ? ? jdl@jdl.com Jon Loeliger @May96 1 1 0 0 antsu@sandra.pp.fi Antti-Pekka @May96 1 0 1 1 Ian.Dall@dsto.defence.gov.au Ian Dall @May96 1 0 1 1 ricker@erols.com Rick Rodman @May96 1 1 1 1 kls@chicago.com Karl Swartz @May96 ------------ From paz@world.std.com (Philip A Zimmermann) Name email quant date comment Dave Rand dlr@daver.bungi.com 1 20May96 can src files Robert Vincent bert@willowpond.com 1? 20May96 maybe. $;parts avail? P. Zimmermann paz@world.std.com 1 20May96 do fab? Lee Damon nomad@castle.org 1 20May96 two hands, tho'. Julian Stacey jhs@freebsd.org ? 21May96 ET proponent Sverre Froyen sverre@sst.nrel.gov 1? 21May96 if < $?, parts avail? Raymond Wiker etorwi@eto.ericsson.se 1? 21May96 if bare pcb <$100 Mark Mason mason@ease.com 1 21May96 write drivers for... Eyal Lebedinsky eyal@eyal.emu.id.au 1 22May96 want serial also ------------------ YES - DEFINITE PCB ORDER in 94 1 1 1 1 tbr@tfic.bc.ca Tom Rushworth @Feb94 1 1 1 1 etorwi@hisoy.etn.ericsson.se Raymond Wiker @Feb94 Bd<$300 ex RAM 1 ? ? ? pete@pelican.com Pete Carah @Feb94 Bd<$300 NoMain532 FreeBSD 1 1 1 1 fellows@unb.ca David Fellows @Feb94 1 ? ? 0 via leo@marco.de Tjeerd Mulder @Feb94 1 0 1 1 neil@skatter.usask.ca Neil Johnson @Feb94 1 1 1 ? marcb@werj.com.au Marc Boschma @Mar94 1 1 1 0 dlr@daver.bungi.com Dave Rand @Mar94 1 1 1 1 rjohnson@mpr.ca Robbin Johnson @Mar94 cond. on $ 1 1 1 0 nomad@CASTLE.ORG Lee Damon @Mar94 < $100/pcb 1 1 1 1 khe@pcs.dec.com KarlHeinz Eckmeier @Apr94 1 1 1 1 jonb@metronet.com Jon Buller @May96 BD<$100 ( Jon Buller, 2301 Pebble Vale #2121, Plano, TX 75075, (214) 422-3051 ) ( email goes away between semesters, but really wants one! ) ( email address from a 11-94 posting) 2 ? ? ? bdavis@netcom.com Bryan Davis @Dec93 PEOPLE WHO'VE NOT MAILED YES/NO IN 1994 - PLEASE MAIL ME YES/NO 0 ? ? ? gj@pcs.dec.com Gary Jennejohn 1 ? ? ? jkh@whisker.hubbard.ie Jordan K. Hubbard 0 0 0 0 gs@chips.com George Scolaro 1 0 ? ? mch@regent.e-technik.tu-muenchen.de Michael Hermann 3 ? ? ? bdale@gag.com + Fred & John Bdale Garbee If drvr 1 ? ? ? rennolet@cs.umn.edu Kumar Rennolet 1 ? ? ? john.connin@OAU.ORG John Connin 1 ? ? ? ackley@banana.cis.ohio-state.edu john@softspace.com John James Ackley 0 0 0 0 sef@kithrup.COM Sean Eric Fagan 0 0 0 0 glowell@allegro.com Gary Lowell 0 0 0 0 frank@ginger.sri.com Victor Frank 0 0 0 0 mem@zwac.mv.com Mark Mallett 0 0 0 0 rhyde@ucrengr.ucr.edu Randy Hyde 0 0 0 0 montana!asgard@convex J.R. Stoner 0 0 0 0 david@marvin.jpl.oz.au David Magnay 0 0 0 0 bolo@cs.wisc.edu Joe Burger NO DEFINITELY NOT 0 0 0 0 jvh@microsoft.com Johannes Helander @ 21 May 1996 0 0 0 0 tom@lajax.pp.fi Tom Javen @Mar94 0 0 0 0 mattias@lysator.liu.se Mattias Olofsson @Mar94 0 0 0 0 gtw@img.com.au Grant Waldram @Apr94 (info from marcb -- --From: paz@world.std.com (Philip A Zimmermann)
I started compiling a list of interested parties; I have to append Julian's list from when things went dormant and contact those people to see if they're still interested. Here's what I've got so far:
[ SEE MERGED LIST ABOVE ]
I'm currently working on the BOM, trying to source the difficult parts. I'll get pricing when the number of participants is firmed up. I am willing to help source the difficult parts (doing a quantity buy on the 'GX, etc. will probably help costs), but would rather not source the popcorn & glue. Offhand, I'd say that it would be worthwhile to have ET532 go to fab if the lot size is ~25.
I've had a good deal of experience dealing with pcb vendors over the years, and especially lately. I appreciate the concern for keeping the pcb fab cost down. My experiences lately have been with some vendors who've done quick-turn jobs for us. Some of the consequences besides higher costs have been poor quality, e.g. over-etching (where traces disappear and the reappear an inch or so away), under-etching (causing shorts between traces) and warpage (a problem if the product is a card to be inserted into a cage). I know of companies which do high-quality work and am hopeful that if we accept a longer delivery time, an acceptable, affordable and reliable product will result. No one likes finding out AFTER paying for and constructing a six-layer pcb that a few traces on some buried (inner) layer are open-circuited or shorted out due to poor fab quality. (Gets off soap-box.)
Some things are sketchy on the BOM; I'm hoping to submit my questions to Dave Rand off-list for clarification, then add sourcing info to BOM.
I'm only interested in doing smaller step at a time, i.e., ET532 first, if it's still practical. Then, if that works out and there's enough interest, I may be prepared to help out with the serial daughtercard. I just don't want to be overwhelmed with too much all at one time.
|
|