AGA Toolkit '97/The AGA Toolkit '97.iso/miscellaneous/hardware/a3000-probs/roba3000problems.doc

Text File | 1996-09-07 | 11.0 KB | 266 lines

The following consists of a list of problems I've had with my Amiga
and how they have (or haven't) been fixed.  I haven't picked
over this with a fine-toothed comb as I have little time to spare
but I decided it would be better to release this in a rough form
than not release it at all.

I received help with some of these problem on csa.hardware.  I can't
find all of the old mail and I can't remember everyone who answered
my questions.  I do remember that Mr. Haynie was disproportionately
helpful.

/*############################################################################*/
Problems:

Amiga 3000-25:
1) Melting hard drive.
   File holes in Amiga case.  Install brushless fans.

2) Lack of an external SCSI terminator.
   220/330 Ohm resistor pack.

3) Incorrect (signal) termpower diode for A3000 SCSI bus.
   Installed correct (power) diode.

4) Flawed early revision daughter board.
   No solution.

Samgsung Synchmaster 7 Monitor:
5) 60 Hz limited horizontal flicker.
   Resolder PTC thermistor.

/*############################################################################*/
1) Melting hard drive.
   File holes in Amiga case.  Install brushless fans.

Old Quantum drives and most Seagates (old and new) run hot.  If you
put them in a location where they don't receive adequate ventalation,
you may get stiction or warped platten.

Definition time:
Stiction:
Prolonged hot running causes internal lubrication to vaporize and
increases wear rate for drive motor.  When the drive is stopped, the lubricant
deposits on the platten (the disk) and causes the head to stick to the drive.
The drive has a hard time starting up but runs fine once spinning until, one
day, it won't start at all.

Warped Platten:
Bent hard disks.

What to do if you have it:
Try rapidly rotating the whole drive body with power applied to break the bond
and get the drive spinning.  Copy the daty to another drive.

If you just can't get the thing going, open the case and (if you're lucky) there
will be a slot on the spindle that you can use to "jump start" the drive with
a screw driver.  Put the top back on immediately and copy the data to another
drive.  After letting dust in the drive, the error rate goes up.  You may be able
to keep using the drive, through.  Most have a dessicant pack that also acts
as an air filter and will keep most dust out of harms way.

Note: If you ever have your machine at a very low temperature, give it pleanty of
time to warm up before turning it on.  Condensation can form inside hard drives.

How to keep your drive cool in the fist place:
For low-profile drives there is sufficient space to fit a 12 volt brushless DC
fan between the case and the drive.  Drill mounting holes in the case and file
out holes for the fan air-paths.  Tap the fan into the floppy drive power supply.
You may want to use a 10 Ohm resistor as a voltage divider to quiet down the fan
and reduce power consumption.

My machine has two such fans in it.  My Baracuda runs hot and needs the cooling.
My Maxtor is fine w.o. any fans.  These little 12V fans are one of the few high
quality things that Radio Shack provides at a good price.

/*############################################################################*/
2) Lack of an external SCSI terminator.
   220/330 Ohm resistor pack.

The Amiga 3000 has a design flaw.  If you want to connect a drive to the
external SCSI port (which is the 25 pin hack anyway) you have to disconnect a
terminating resistor pack deep in the guts of the machine.  If you want to
unplug the external device, you have to put the internal pack back in
place.  There are three ways of dealing with this problem:
   1) Leave any external devices plugged in all the time.
   2) Ignore the SCSI specs and leave the terminator in all the time
      (and wonder why you can't reliably transfer data).
   3) Get an external terminating pack.

"3" is the best solution but finding a 25 pin DB terminator is not an easy
thing to do.  One can, however, find a couple of 220/330 ohm terminating
resistor packs [1] that can be soldered to a DB connector.  Check out the SCSI
FAQ [2] and page B-11 of your Amiga 3000 manual for everything you need to
know about the proper connections to make for this terminator.

If you ever have problems with your SCSI bus, I recommend that you check out the
SCSI FAQ [2] before posting to csa.hardware.  It's not that requests for help
are a bad thing.  The big problem is that "helpful" people who don't know what
they're talking about give advice, causing no end of trouble to the victims who
take this advice.

/*############################################################################*/
3) Incorrect (signal) termpower diode for A3000 SCSI bus.
   Installed correct (power) diode.

In a number of A3000s, the termpower diode (D800) for the SCSI bus is installed
in the wrong direction.  It typically shorts so termpower is connected
(unreliably) to +5.  If your SCSI bus is in conformance with the SCSI standard
but is acting flakey, this is a good thing to check.  The diode should allow
current to flow from +5 to termpower, not the other way around.

The diode used in a significant number of A3000s was incorrect, as well.  On
the A3000s I know about, a 1N914 is used.  This is a signal diode, not a power
diode.  It also has a rather large Vt for use as a termpower diode when the
input signal is only +5.

If you have determined that your diode is cooked, I suggest that you replace
it with a 1N5817.

/*############################################################################*/
4) Flawed early revision daughter board.
   No solution.

This is the only story with a sad ending in this list.  I wanted to be able
to expand my memory with 60ns SIMMs that will be useful in a 68060 accelerator
(if I ever get one).  I bought a DKB3128, which didn't work quite right with
my system.

Mail from Mr. Haynie about this problem follows:

<Rob>
> I recently installed a DKB3128 (Z3) memory expansion board in my
> Amiga 3000.  My system "freezes" instantly if burst mode is enabled
> with "cpu burst".

<Dave>
That's fairly odd, especially considering that instruction burst is
enabled by default (actually, in SetPatch). So all you're doing here
is additionally enabling data burst.

<Rob's note: Instruction burst is enabled by default on my system
but data burst is not and data burst is what resulted in crashes.>

> With burst mode turned off, memory reads and writes that happen at the
> same time as hard drive accesses are unreliable.

It sounds like you may have a marginal DMAC, though it's impossible to
be sure that's the problem without additional testing. The DMAC should
talk reliably to anything the CPU can.

> The memory itself appears to be fine as I have repeatedly tested it
> w.o. problem with a number of memory test programs including C='s
> SysTest.

Most memory tests are all but useless, they tend to identify gross
errors, but can actually hide more subtile errors. Though I think the
SysTest, if recent enough, has memory tests that can identity some
real problems.

> though I think I have my DMA mask set to 0x7ffffffe.  I have a Rev
> 11 Buster in the machine.

<Rob's note: The problem still occurred with a mask of 0x7ffffffe.>

If the DMAC is at fault (with access to your Zorro-based Fast RAM),
setting you mask to 0x0ffffffe will keep the DMAC out of that Fast
RAM, limiting it to motherboard memory. If this clears up the problems
with the Zorro bus memory in the system, the "might have a DMAC
problem" column gets another checkmark.

> These numbers are printed on the motherboard:  FAB 313312, Part Number
> 313313.  Does this specify revision as well?

Nope. The revision is specified, but it's fairly clear, like "Rev 7.0"
or whatever.

> This is an old A3000, '91 I think.

It's possible your Zorro III problems are due to trouble with the
system's backplane card. I recall the original layout had some
problems, which were reworked twice (the first time, the rework was
wrong). Kind of stupid problems, but unfortunately I didn't have the
time to personally supervise every aspect of the backplane card
design. My vague recollection is that there was missing termination on
one of the data lines, possibly also one of the new data strobes
(there are four data strobes for Zorro III cycles, two are the same
lines used in Zorro II cycles) and/or one of the burst lines. Even if
the burst line itself isn't wrong, enabling burst out to the expansion
bus (which really happens with the Rev 11 Buster) makes timing more
critical, so a problem with either of these other signals will be
worse.

Unfortunately, I can't find any record of what was screwed up on the
early backplanes. I think the troubles were cured by Rev 8, definitely
by Rev 9 (again, why so many revisions? I dunno).

> 81 Super DMAC
> 41 Fat Gary
> 51 Ramsey

Those number aren't the chip revision number. The Rev number will be
printed as a letter, or as a -xx number after the Commodore part
number.

                    -Dave Haynie
........

So, after a few more tests, I tried to find a later revision daughter
board.  After calling all of the Amiga dealers I know of, I was unable to
find anything more recent than a revision 7.0.  Anyway, I sent back the
DKB 3128 and now I'm back to 6 megs.

/*############################################################################*/
5) 60 Hz limited horizontal flicker.
   Resolder PTC thermistor.

About every fourth row of my screen was shifting to the left by a pixel
occasionally.  This problem was most severe immediately after power-up.

It turned out that the engineers had used a PTC thermistor in series
with the degaussing coil to handle power-on degaussing.

Definition time:
Degaussing - An alternating magnetic field of gradually decreasing intensity is
generated to eliminate residual magnetism in the monitor that may be alterning
the paths of the electrons in the CRT.  This is done on power-up for most
monitors.  Some also have a degaussing button.

Thermistor - An electronic component, the resistance of which changes as a
function of its temperature.

PTC - Positive Temperature Coefficient.  The resistance goes up as the
temperature goes up.  NTC means negative temperature coefficient.

Back to the story:
This means that the thermistor is hot whenever the monitor is on.  It had
desoldered itself from the PCB.  After resoldering the thermistor, the problem
went away.

/*############################################################################*/
[1] A few mail order electronics supply houses in the U.S.
Digi-Key 1-800-344-4539
Mouser 1-800-346-6873


[2] I quote from the SCSI FAQ:
Where to get the latest copy of this FAQ:
  The comp.periphs.scsi FAQ is posted to Usenet during the first week
  of each month. In addition, a recent version can be obtained
  via anonymous ftp from:
  ftp.wang.com:
      garyf/scsi/scsi-faq.part*
  OR
  rtfm.mit.edu:
      pub/usenet-by-group/comp.periphs.scsi/comp.periphs.scsi_FAQ_part_*
  OR
  via World Wide Web (WWW):
  http://www.cis.ohio-state.edu/hypertext/faq/usenet/scsi-faq/
  OR
  (an old version but in HTML format)
  http://alpha.med.pitt.edu:9000/LINK/SCSI_FAQ.html


/*############################################################################*/
-Robert Dick ([e-mail])-