Ack! Dead 3b1.

Thaddeus P. Floryan thad at btr.BTR.COM
Fri Jun 21 18:56:18 AEST 1991


In article <PMC.91Jun20211936 at arrakis.engin.umich.edu> pmc at engin.umich.edu (Paul S McClay ) writes:
| [...]
| I believe the video RAM is bad. On power up the self test sequence
| hangs with the LEDs indicating a fault in the video RAM and there are
| glitches in the <16bits on/16bits off> pattern on the screen.
| [...]
| The bit rot always occurs in the upper third or less of the screen. It
| is roughly, but not perfectly, confined to one or two bands a few
| pixels high and horizontally periodic. I suppose the video RAMs are
| 64kbit chips, in which case one would cover a bit more than 1/4 of the
| screen. I can't say for sure weather the bit rot is confined to the
| first chip or not.
| 
| Is there any possibility that this might *not* be fixed by replacing
| the #1 video RAM? (except that possibility that it might be the first
| two chips)
| 
| Where can I get replacements?

An almost-similar thing happened to me at the beginning of the UNIX Users'
Group meeting two months ago ... how embarassing: the ONE time I did have all
the examples, source code, etc. on the machine, and it wouldn't boot!  :-)
The greed LED (test state 2) stayed on solid; would be "neat" if the BOOT ROMs
could bypass this (and ONLY this (the video RAM)) test if, say, a certain key
or mouse button were depressed on boot.

In my case, the problem was a bad RAM chip (the 4416 at 17A to be precise).
What I noticed was that the first 4 pixels every 256 addresses in the "spider
web" pattern were blank.  Noting the RAM chips are organized 64Kx4, it was
easy to track down the bad chip from the schematics. I bought four 4416 as
replacements but decided to trust my sleuthing and only replaced 17A, and the
system booted just fine afterwards.

As for your problem, note the RAM usage is "interleaved"; the first 4 bits of
each 16, BD0-BD3, are from the 4416 at 17A, the 2nd 4 (BD4-BD7) at 16A, the 3rd
4 (BD8-BD11) at 15A, and the last 4 (BD12-BD15) at 14A.

The screen RAM is "free-running", so if the problem isn't a RAM chip than it'd
have to be one the two shift registers (74S299) at 14B or 15B; the two
74F245 at 13A and 17B serve only to gate the data bus into the video RAM.

If you get the pattern that looks like:

----____----____----____----____----____----____----____----____----.....
----____----____----____----____----____----____----____----____----.....
----____----____----____----____----____----____----____----____----.....

                                                            ^^^^
see if you can "count" the missing pixels.  Each one of the |||| is a 16-bit
word in video RAM displayed BD0-BD15 (left to right) at a monotonically
increasing address.  If you can determine a regular pattern, then it'd be
quite straightforward to determine precisely which RAM chip(s) is (are) bad
in the 4-chip video RAM array.  The chips are common parts, generic number 4416
(aka ``MB 81416-12'') at 120nS, and shouldn't cost more than a few US$ each in
small quantity; I got mine at Fry's Electronics but note that since they are
common parts you shouldn't have difficulty locating them.

The 3B1 video display is 720H x 348V, so, from the upper left corner, the RAM
address locations are:

	 0 thru 44	for the first video row
	45 thru 89	for the second video row
	90 thru 134	for the third video row
	   etc.

and at each video RAM address the 16 bits, left to right, (again) are:

| BD0 | ... | BD3 | BD4 | ... | BD7 | BD8 | ... | BD11 | BD12 | ... | BD15 |
|                 |                 |                  |                   |
|   RAM at 17A    |    RAM at 16A   |    RAM at 15A    |      RAM at 14A   |

The above info "should" be enough to help you isolate the problem to a given
chip.

As for Paul's other problem and question:

| Procedural problem: I *can not* disconnect the floppy drive power
| cable.  I've pulled *very* hard, with no success. I can not see any
| sort of locking mechanism on the connector. Is there a lock?

The newer 3B1 systems used an IDC connector for the floppy drive's power and
it is a real pain to remove those; I use a long-nosed needle-nosed plier and
lever it against the floppy frame to pry the connector out perpendicularly.

| (This is a 3b1. Someone said some 3b1s were built with 7300 boards, which
| power the floppy from the board. It was implied, then, that in "real" 3b1s,
| the floppy power comes direct from the PS and the top plate can be trivially
| removed from the rest of the frame. Is this the case? Do I then have a
| wimpy PS for sure, or might it be a 3b1 PS?)

It's the HARD DRIVE power cable on the 3B1 that goes straight to the drive;
the floppy power for both the 7300 and 3B1 comes from the motherboard.

Thad Floryan [ thad at btr.com (OR) {decwrl, mips, fernwood}!btr!thad ]



More information about the Comp.sys.3b1 mailing list