26 Mar
2005
26 Mar
'05
11:54 a.m.
We've got a Sun 4/330 which has bit 6 stuck high in every character
output on the console. I don't suppose anyone has schematics for these
boards or know how the serial hardware works, do they?
The machine looks to be booting OK, so it appears address & data buses
are fine - but at the point where characters must be loaded into a
parallel register and converted to serial for output it's falling
over...
I suspect the machine's hard-wired to use only serial port A as console,
but either way I couldn't get hold of the right cable to see if the
other three serial ports are OK (the latter three being 9-pin male,
whilst port A is 25 pin male)
I can't think of any line settings, faulty voltage rails etc. that'd
cause bit 6 to be stuck high, so I'm assuming it's a hardware fault...
cheers
Jules
27 Mar
27 Mar
1:58 p.m.
Jules Richardson wrote:
We've got a Sun 4/330 which has bit 6 stuck high
in every character
output on the console. I don't suppose anyone has schematics for these
boards or know how the serial hardware works, do they?
There's a Zilog serial chip, quite close to the serial connector.
If it's just bit 6 that's faulty, it sounds like a fault in that
chip (in the parallel section). I think it's a Z8530.
--
John Honniball
coredump at gifford.co.uk
11:44 p.m.
On Sun, 27 Mar 2005 22:58:01 +0100, John Honniball
<coredump at gifford.co.uk> wrote:
There's a Zilog serial chip, quite close to the
serial connector.
If it's just bit 6 that's faulty, it sounds like a fault in that
chip (in the parallel section). I think it's a Z8530.
Suns (as well as old Macintoshes, etc.) do use Z8530s for serial
ports. Depending on the vintage of Sun, it's likely to be in a PLCC
package, though. Dunno much about the Sun 3 line, so I can't say. I
do have some experience with the Z8530... they were shipped in several
speed increments (we used to use the 4MHz version on our later
COMBOARDs), so make sure if you find/pull a replacement that the speed
matches. I _think_ 8MHz was the fastest version made.
-ethan
28 Mar
28 Mar
10:41 a.m.
On Mon, 2005-03-28 at 02:44 -0500, Ethan Dicks wrote:
On Sun, 27 Mar 2005 22:58:01 +0100, John Honniball
<coredump at gifford.co.uk> wrote:
Just to confirm - Zilog parts labelled Z0853006PSC (SCC) are the same as
AMD part AM8250H-6 ? (I would think so)
This Sun board has two AMD parts which run the four serial ports (plus a
third which looks to handle keyboard and mouse)
I've identified whch of the three on the board is responsible for the
console port (and faulty). However, I've got a Sun "Comm Processor II"
board in a Sun 3/150 which has eight of the Zilog ICs on board. We don't
need that many ports for anything, so I'm tempted to do a swap (if I
don't find a surplus AMD chip in my spares bin).
Earlier I cut the A/-B track (which selects which of the two ports on
the SCC chip to use) and diverted it via an inverter, so that when the
Sun 4/330 thought it was accessing the console port it was actually was
using the second serial line - hooked up the physical console to the
second port then, but no change in the garbled output. So basically that
confirms the the SCC chip is broken, and that the parallel - serial
conversion is common to both ports on the chip (which is
understandable). I was hoping for a quick win though :)
Time to get the desoldering equipment out I think...
cheers
Jules
There's a Zilog serial chip, quite close to
the serial connector.
If it's just bit 6 that's faulty, it sounds like a fault in that
chip (in the parallel section). I think it's a Z8530.
Suns (as well as old Macintoshes, etc.) do use Z8530s for serial
ports. Depending on the vintage of Sun, it's likely to be in a PLCC
package, though.
10:45 a.m.
On Mon, 28 Mar 2005 18:41:12 +0000, Jules Richardson
<julesrichardsonuk at yahoo.co.uk> wrote:
Just to confirm - Zilog parts labelled Z0853006PSC
(SCC) are the same as
AMD part AM8250H-6 ? (I would think so)
I do not know what the AMD equivalent of a the Zilog part is, sorry.
One thing I _did_ notice is that the Zilog part you listed should be
the 6MHz version (4, 6 and 8 are the ones I'm familiar with).
Obviously, one can use either a 6Mhz or 8Mhz where a 6MHz is called
for.
This Sun board has two AMD parts which run the four
serial ports (plus a
third which looks to handle keyboard and mouse)
That makes sense (the Z8530 for kb and mouse).
-ethan
12:11 p.m.
Jules wrote:
Just to confirm - Zilog parts labelled Z0853006PSC
(SCC) are the same as
AMD part AM8250H-6 ? (I would think so)
No. I wasn't aware that AMD had ever made the 8250, but if they did,
it's definitely not equivalent to a Z8530. The AMD equivalent is
the Am8530, though there are also improved variants like the Am9530H
and Am85230.
The 8250 is a single serial port chip which was used in the early
IBM PCs and compatibles. Later it was replaced by the 16450, 16550,
and 16550A.
Eric
12:56 p.m.
On Mon, 2005-03-28 at 12:11 -0800, Eric Smith wrote:
Jules wrote:
Argh! I meant the AM8530 - got a brain full of different chip numbers
at the moment :-(
Anyway, you told me what I wanted to know, even if I didn't ask
properly ;)
I've found I've got *one* Zilog 8530 in my box of spares... but it
doesn't have any kind of speed rating on it (which may mean it's only
4MHz).
Still, I'll take the dead AMD one off the Sun board, stick a socket on
there and give this found one a try. If not I'll have to desolder a good
6MHz Zilog chip from the Sun comms board and use that...
cheers
Jules
Just to confirm - Zilog parts labelled
Z0853006PSC (SCC) are the same as
AMD part AM8250H-6 ? (I would think so)
No. I wasn't aware that AMD had ever made the 8250, but if they did,
it's definitely not equivalent to a Z8530. The AMD equivalent is
the Am8530,
1:41 p.m.
Jules Richardson wrote:
Just to confirm - Zilog parts labelled Z0853006PSC
(SCC) are the same as
AMD part AM8250H-6 ? (I would think so)
No, the AMD part is an Am8530. The 8250 is something different.
I've identified whch of the three on the board is
responsible for the
console port (and faulty). However, I've got a Sun "Comm Processor II"
board in a Sun 3/150 which has eight of the Zilog ICs on board. We don't
need that many ports for anything, so I'm tempted to do a swap (if I
don't find a surplus AMD chip in my spares bin).
I have four Z8530APC chips here, if you want one.
Earlier I cut the A/-B track (which selects which of
the two ports on
the SCC chip to use) and diverted it via an inverter, so that when the
Sun 4/330 thought it was accessing the console port it was actually was
using the second serial line - hooked up the physical console to the
second port then, but no change in the garbled output.
No change? That's odd.
So basically that
confirms the the SCC chip is broken, and that the parallel - serial
conversion is common to both ports on the chip (which is
understandable).
Erm, no the serial/parallel conversion isn't common to both ports
on the chip. If it was, you wouldn't be able to set the two
ports to different baud rates, for example. The two channels
of the 8530 are completely separate.
This implies that some buffer chip that feeds to 8530 is
faulty. It may be a 74LS244, and 74LS245 or something similar.
Try to trace bit 6 of the 8530's data bus back to a buffer.
Time to get the desoldering equipment out I think...
Not so fast! Check that buffer first!
--
John Honniball
coredump at gifford.co.uk
29 Mar
29 Mar
3:25 a.m.
On Mon, 2005-03-28 at 22:41 +0100, John Honniball wrote:
Jules Richardson wrote:
Hmm... curious. I'd downloaded the datasheet for the chip and it wasn't
clear either way so figured I'd try the inverter trick.
Earlier I cut the A/-B track (which selects which
of the two ports on
the SCC chip to use) and diverted it via an inverter, so that when the
Sun 4/330 thought it was accessing the console port it was actually was
using the second serial line - hooked up the physical console to the
second port then, but no change in the garbled output.
No change? That's odd.
So basically that
confirms the the SCC chip is broken, and that the parallel - serial
conversion is common to both ports on the chip (which is
understandable).
Erm, no the serial/parallel conversion isn't common to both ports
on the chip. This implies that some buffer chip that feeds to 8530
is
faulty.
If they are totally seperate channels (no common internal buffering
first, say) then yep, it looks that way. There's no buffers immediately
by the serial logic, so I'll look further afield.
I've only got the board here with me, not the machine, so can't do any
work under load conditions - but then I'm pretty sure we don't have any
bus extension boards that'd fit this Sun board anyway (it's extremely
deep - much more than I've typically seen on machines with a cardcage)
cheers
Jules
8:46 a.m.
Jules Richardson wrote:
I've only got the board here with me, not the
machine, so can't do any
work under load conditions - but then I'm pretty sure we don't have any
bus extension boards that'd fit this Sun board anyway (it's extremely
deep - much more than I've typically seen on machines with a cardcage)
Sun's so-called VME boards are much bigger than real VME boards,
both in height and width. I have some card "frames" here that
allow you to plug ordinary VME cards into Suns, if you need any.
--
John Honniball
coredump at gifford.co.uk
1:16 p.m.
On Tue, 2005-03-29 at 17:46 +0100, John Honniball wrote:
Jules Richardson wrote:
I gather from poking around that the top two backplane connectors follow
the accepted VME standard pinout, but the bottom connector is the one
that carries the power on Sun machines and is proprietary...
unfortunately it'd be the bottom one I'd need to know in order to power
up the board away from the machine!
I've only got the board here with me, not the
machine, so can't do any
work under load conditions - but then I'm pretty sure we don't have any
bus extension boards that'd fit this Sun board anyway (it's extremely
deep - much more than I've typically seen on machines with a cardcage)
Sun's so-called VME boards are much bigger than real VME boards,
both in height and width. I have some card "frames" here that
allow you to plug ordinary VME cards into Suns, if you need any.
Hmm, I can't immediately think of a use for them to be honest. Our Sun 2
is multibus, and our 3/50 is just an all-in one thing. They'd fit our
3/150, but that's reasonably tricked out with goodies already.
The machine with this 4/330 board in it only has one slot free on it;
the rest are crammed full of boards with the number-crunching DSP
processors on (4 processors per board, and I believe 16MB of DRAM per
processor - it's a monster!).
cheers
Jules
11:35 p.m.
Sun's so-called VME boards are much bigger than
real VME boards,
I thought they *were* "real VME boards", just for a different and
somewhat less common VME spec (9U instead of 6U). Was I confused?
/~\ The ASCII der Mouse
\ / Ribbon Campaign
X Against HTML mouse at rodents.montreal.qc.ca
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B
12:31 p.m.
On Mon, 2005-03-28 at 22:41 +0100, John Honniball wrote:
Jules Richardson wrote:
This implies that some buffer chip that feeds to 8530 is
faulty. It may be a 74LS244, and 74LS245 or something similar.
Try to trace bit 6 of the 8530's data bus back to a buffer.
Right, no buffer unfortunately.
It'd appear that one of the big LSI chips handles the data bus to the
SCC 8530 chips, which is also common to the clock IC, the board's ID
PROM, and also an octal D-type flip-flop IC (as an input). the latter's
responsible for latching the diagnostic LEDs on the edge of the board I
think.
I'd be surprised if the system would boot (which it appears to be doing
fine) if the LSI chip was busted and the machine couldn't talk to the ID
PROM (and probably the RTC chip too), but I suppose it's possible. I
don't recall it having lamp #6 of the diag LED array jammed on either.
So, it's *probably* the SCC but I think I'll take the board back to the
museum next time I'm there and as Dave said, stick a 'scope on it just
to be certain.
One SCC drives serial ports A and B, and another drives C and D.
Unfortunately it's probably more trouble than it's worth to try and fool
the system into swapping those over via a few hack wires on the board to
see if it fixes things - as there's an interrupt chain it's not as easy
as just swapping chip select lines.
cheers
Jules
6989
days inactive
6993
days old
12 comments
5 participants
participants (5)
-
coredump＠gifford.co.uk -
eric＠brouhaha.com -
ethan.dicks＠gmail.com -
julesrichardsonuk＠yahoo.co.uk -
mouse＠Rodents.Montreal.QC.CA