B&W G3: 820-0987-A board . . . with no FireWire ports??

[Trash80toHP_Mini]

Got this board to build a bridge Mack oh so many years ago, log before I discovered just how iffy the info on LEM can be. Figured it to be Rev A from the part number and set this one aside in hopes of scoring a Rev.2 board. Pulled it out of the drawer while consolidating today and looked it up before stacking it atop the pile in the Mac MoBo box. EveryMac says nothing at all about there being two revs. LEM says to look for "640" on the CMD ATA controller to confirm it's a Rev.2/B/Whatever.

At any rate, both sites say it should have a pair of FW400 ports? No gots, are they both wrong?

The Linux gang seems to support the CMD PCI646U2 ATA controller on my board with only a passing nod to (dead link) the 640 variant LEM claims to be the cat's meow. 640 would be an earlier rev, no?

Dazed and confused. :blink:

 

[MOS8_030]

The FW ports were in a separate module that connected to the board.

 

[Trash80toHP_Mini]

AHA! Thanks, there it is, didn't know about it, don't need it and I hope I can do without it and the front panel control board?



Is there anything else I do need? Is there anything to the LEMrumor about two different revisions of the board? I figure even a 300MHz rev,1 B&W board is better than anything in Beige?

 

[EvilCapitalist]

If you've got a rev 1 B&W board and are using the onboard IDE controller, yeah there can be problems.  I noticed it with most when I tried to install larger, more recent drives but from what I've read it also happens if you have more than one HD installed.  The problem manifested for me as random bombs in OS 9.  I never tried OS X on that machine so I don't know how it reacted.

Personally I prefer the Beige G3s over the B&Ws.  In spite of the higher bus speed (100 vs 66 in the beige) and faster memory in the B&W it didn't "feel" all that much faster than the Beige.  The one advantage I will give it is the case.  No question the B&Ws are much easier to work with when it comes to part swapping.

 

[Trash80toHP_Mini]

I f the G3 board will POST w/o the FireWire and front controller board I'm happy. The thing I like about is that 66MHz slot for a VidCard, I almost never use onboard video, even Beige unless you count screwing the temporary Plexi backplane plate to it.

Are the ATA problem any worse than those of the slower Beige with large HDDs? I'm sure I've got a fast G4 with more Cache for that socket and the board's a better fit for the target box:

Notice the tray for one of the two dedicated UltraSCSI bays. Those controllers I've got.

edit: Beige in the pic of the Dell Box.above.

 

[Cory5412]

Are the ATA problem any worse than those of the slower Beige with large HDDs

It should be fine if you use the stock drive with it. It's a different problem entirely than what the Beige G3 had. It's unstable on any drive with UDMA, which is most drives after the blue-and-white was launched. (It also has the LBA48 limitation, but you're going to hit the UDMA limitation before you hit LBA48 capacities.)

The Rev-1 B&W can also only have one disk on its main hard disk channel.

The Beige's main limitation is the whole 8-gig thing. It doesn't become unstable with UDMA disks, to the best of my knowledged.

Given the nature of the IDE problems on this system with "newer" drives, a Rev-1 B&W would be a good system to put a SATA/IDE/SCSI controller in and skip the onboard entirely, if you can.

I figure even a 300MHz rev,1 B&W board is better than anything in Beige?

I'd say it depends.

Of course, the Beige didn't have the same kinds of IDE problems the RevA/1 Blue-And-White did, so there's that in its favor. If you need something to put a G4 upgrade in, the blue-and-white will be better because of its faster bus. If you need OS X, same deal, except for mostly that you're not wasting all the hardware in a beige that OS X can't use. (a/v, the not fully accelerated onboard Rage, serial/localtalk.) The Blue-and-white also has USB, which is nice and 10/100 ethernet, which is nice, even if it's mostly meaningless under classic mac os. So if you need a MacOS 8/9 box specifically I'd say it's kind of a wash because just around that point is where OS 9 starts to fall off in terms of getting tangible day-to-day benefits out of more performance.

So, like, it really depends on what the machine is for. Both systems are upgradeable, so it's not like the IDE is going to cause you trouble, unless you already have all of the slots spoken for and none of them is something you can boot from.

Another option if you need more out and out performance is to just advance to a G4 motherboard. You'll get better everything as a base for whatever it is you're doing.

CMD PCI646U2 ATA controller on my board with only a passing nod to (dead link) the 640 variant LEM claims to be the cat's meow. 640 would be an earlier rev, no?

Mmmm. Here's what wikipedia says:

The CMD chip on Rev. 1 logic boards is PCI646U2 and on Rev. 2 logic boards is 646U2-402

So, both chips are the PCI646, it's just that the revisions are U2 and 402, with 402 being newer.

Wikipedia also has a more detailed description of the exact failure/problem mode the RevA B&W exhibits.

LEM's page doesn't mention a 640 at all, although it may have been updated at some point.

 

[Trash80toHP_Mini]

Thanks for clearing that up, confuzzlement a/o typo on my part there, I meant 402, which was LEM's marker for the Rev.2 board.

It's too bad, the the B&W board looks so much better in the case than the dinky little BG3DT board in the picture above. But I pulled out the replacement/right angled drive cage I've been setting up for this this box and spotted the Black Mac FDD in the 3.5"-5.25" adapter and the PC FDD front bezel that's a perfect fit to hot glue to the cage bezel  .  .  .  for any logic board with a Floppy connector! :

So Beige is back, the rev.1 G3 goes into the storage box and the Dell case back up on the TelCo rack entertainment system. Its main job title is "counterweight" as it stands.

 

[Trash80toHP_Mini]

Why does the "Front Panel" interface need 20 pins? How much could be going on there to need anywhere near that many? The connector is even the same! Too bad they weren't using n.c. and ground lines on the FDD interface for front panel functions with a stealth FDD port implementation.  :mellow: If wishes were horses  .  .  .