Need opinions, dead 9600/350

[Outlander]

Hello again. I am still searching for a reason as to why my Kansas 9600 logicboard bit the dust a few months ago. I ran this machine since new, almost 12 years(or more) of service with no issues at all. Installed a 36GB Seagate SCA drive a few years back with an SCA to 50 pin SCSI adapter installed on it. No problems after that. Then I had 10.0, than 10.1, than 10.2.8 running without issue. The last thing I did was install a Sonnet Crescendo 1Ghz G4 upgrade card in the machine and load 10.3 a few months back(maybe more like a year). Everything seemed okay, I ran my 9600 pretty hard, going from 8.6, 9.2, and 10.3 all the time, benchmarks, Strata 3D work, Photoshop, etc. Than one day it started acting funny. The machine took longer to boot and when switching from OSX to classic, the system folder kept on becoming unblessed for no reason. A few more weeks passed and it got worse, it got to the point where in OpenFirmware it refused to see the onboard SCSI chip. Now, the machine is even hanging with a 2930cu SCSI card installed and nothing on the onboard controller. I also have noticed, not long after it was put into production, the Sonnet Crescendo G4 card is off the market. :?:

So I'm wondering which of these things could have killed my 9600 board. The cheap SCSI adapter? The 1Ghz G4 card? Or maybe Xpostfacto and running 10.x? Any opinions on what could of happened? Thanks.

 

[LCGuy]

Have you tried different RAM?

 

[porter]

Battery?

El Nino?

 

[Outlander]

Ram yes, changing it out did not work, one of the first things I did. Was using 4x 128mb dimms, interleaved. Battery is full 3.6v. El Nino? lol

 

[equill]

Maintenance? HDD: file defragmentation/file-directory repair/archiving files (to maintain less than 85% fullth [a technical term]). Break-and-make of all connector-socket junctions. Cable-integrity check, especially the HDD cables where they exit from the fold-up part of the chassis if you have drives on the floor or SCSI HBAs. Does the 80/50 adapter have high-byte termination and provide termination power? Why not consider installation of a Wide SCSI HBA and 68/80 adapter so as not to strangle the Seagate's performance? Are you using either EDO or FPM RAM, but not both, inserted in A6/B6 and A5/B5 slots?

MLB inspection? Nothing burnt, missing or vaporized/leaking?

Full reset of the MLB? All power removed after measuring the PRAM battery's voltage in circuit as 3.3V or more. Leave unpowered for at least 10min. Use the CUDA for not less than 40sec after reassembly but before restoration of mains voltage, closure and powering-up.

de

 

[coius]

I have a PM 9500 Motherboard that I am pretty sure works, but if you want it, it's Free+shipping to you from Omaha, NE 68132 USA

Let me know

 

[beachycove]

Another poster has suggested a 9500 logic board as a replacement. Having recently tried this, blindly following a thread in these fora, I can report that transplanting a 9500 logic board into a 9600 in fact will not work.

9500 logic boards will indeed physically fit (more or less) in an 8600/ 9600 case. However, the power supply connections are different, so that it is impossible to supply power to the board. Furthermore, a 9500 psu is physically very different than the psu in a 9600, and will not go in the 9600 case. So a functional 9500 logic board is no real use to someone wanting to fix a 9600, unless you are into serious hacking.

There is also the "Kansas" factor to remember: the 604ev originally in the late 8600s and 9600s used a different logic board than the earlier 8600s and 9600s, which were very like the 9500's architecture.

The failure of that well-used 9600 over time sounds to me rather like what happened to my 840av some 5-6 years ago when the capacitors first went bad. One simple diagnostic, after the pram purge, would be to give the logic board a careful wash and dry. If it then revives, you need new caps. It is a machine from 1996-97, after all. Its time has come.

 

[Outlander]

Another poster has suggested a 9500 logic board as a replacement. Having recently tried this, blindly following a thread in these fora, I can report that transplanting a 9500 logic board into a 9600 in fact will not work.
9500 logic boards will indeed physically fit (more or less) in an 8600/ 9600 case. However, the power supply connections are different, so that it is impossible to supply power to the board. Furthermore, a 9500 psu is physically very different than the psu in a 9600, and will not go in the 9600 case. So a functional 9500 logic board is no real use to someone wanting to fix a 9600, unless you are into serious hacking.

There is also the "Kansas" factor to remember: the 604ev originally in the late 8600s and 9600s used a different logic board than the earlier 8600s and 9600s, which were very like the 9500's architecture.

The failure of that well-used 9600 over time sounds to me rather like what happened to my 840av some 5-6 years ago when the capacitors first went bad. One simple diagnostic, after the pram purge, would be to give the logic board a careful wash and dry. If it then revives, you need new caps. It is a machine from 1996-97, after all. Its time has come.

I'm gonna try the hack job route for the moment, 9500 motherboard. Not worried about the PSU issues or CPU(since I'm going to be running my 1ghz G4 sonnet card). The 9600 board is clean, no caps have leaked anywhere. The on board SCSI chip is simply fried. and unfortunately it is integrated into a chip on the board. The board probably has other issues too. I guess I could try the cap replacement route even thou they all look okay, could be dried up or something I guess. Still wouldn't fix the onboard SCSI issue thou :-/ Guess I'll order a bunch from mouser, I think the 20 or so capacitors I used to fix my friends G5 iMac was only like 5 dollars if that. Yeah why not, I guess I'll try both routes, see which hack works out the best. }

 

[trag]

The on board SCSI chip is simply fried. and unfortunately it is integrated into a chip on the board. The board probably has other issues too.

It is rare to have more than one cause of failure at a time. It happens, it's just not that common.

So has SCSI Bus 0 or Bus 1 failed? The one that resides in the MESH chip, or the one that resides on CURIO?

If it is on CURIO, that is not too hard to fix, as the CURIO chip can be stolen from any other machine from the x100 family forward through the entire Power Surge (X500/X600) family. The soldering looks intimidating, but it can be done with a bottle of flux, desoldering braid, some modeling clay, a heat gun, and 40 and 15 watt soldering pencils from Radio Shack. Oh, and some dental picks and a stop watch don't hurt either.

If you don't have the heat gun and such, the tools quickly outstrip the price of the board though. Heck, just the shipping cost to send it round trip to someone who can do the soldering is a pain. Still, you might check with Mike Richardson, present on this board, and see if he'd take a hack at replacing that CURIO chip, if you're pretty certain that's the problem, and assuming the washing doesn't fix it up. I'd offer to do it, but I just don't have time. I still owe a nice fellow in the UK a ROM DIMM assembly as it is...