My Wierd G3/266 desktop (Beige)

05/25/00 - Willy McLachlan

I purchased this machine from a local user via an Ebay auction. After getting it home and setting it up, I noticed it reports in various system reporting tools (Apple System Profiler, Norton System Info, XLR8 and PowerLogix cache software) the processor seems to want to run at 200mhz. Tasks seemed "bogged down" somehow compared to the basic snappiness of my G3/266 Wallstreet Powerbook, and the audio output was distorted - similar to the output one hears when a Digital Audio Workstation based on a Mac is having audio timing issues. Not to mention it locks up at very strange times...like after being ignored for many hours, left sitting doing nothing. It also just about always without fail crashes at the very last stage of shutdown, but that's probably software so I'll get to it later.

I decided at this point it wouldn't hurt to investigate. I inspected the motherboard, and removed the tape over the J16 jumperblock, where various Bus timings are controlled thru shorting various combinations of positions within the block (a.k.a. "overclocking").

Refer to this article at Mike Breeden's XLR8YourMac! site for the Beige G3 motherboard J16 jumper block timing settings.

The motherboard is Rev. A using Rev C roms, and uses a "multiblock" jumper - a long plastic block with slots for jumper contacts that covers all 9 positions of the J16 jumperblock on the motherboard. That multiblock contains the right amount of jumper contacts, all in the right positions to set this machine for a 33.33mhz PCI bus, a 66.66mhz System Bus, and a 4x processor speed, which should equal roughly 266mhz.

Top view of the multiblock:

multiblock, top view

Bottom view of the multiblock, with position numbers:

multiblock, bottom view

The Problem:

The System Bus reports running at 66.7mhz, but the processor reports running at 200mhz. The only way for that to be happening is for the J16 timing block to be set to a 3x multiple (66.7 x 3 = 200.1), and not the 4x multiple that the above jumper configuration as it shipped from the factory should yield. Therefore, somehow position 3 of this motherboard's J16 jumperblock is shorted. I removed the motherboard and inspected the underside, and the soldering is clean and the pads are totally isolated from each other. If there is a short it is in the internal layers of the motherboard itself, in the J16 header itself, or somewhere else that I probably won't be able to find on my own...that is unless someone can tell me where the 18 points at J16 come out elsewhere on the board :-)

The J16 block with individual mini-jumpers added in the same configuration as the stock multiblock:
(Pardon the bad focus...it was late)

J16 set for 33/66/266mhz

The speed reported by the PowerLogix control application:
(Don't mind the climbing temperature- I had the case open a tad too long)

PowerLogix report

To confirm this, I reset the jumpers to the only other configuration possible that would take into account the combination of the apparently permanently shorted position 3 and positions 2 and 4 containing jumpers - setting the 7-8-9 section of the J16 block to jump positions 8 and 9 instead of only position 7. The result should be slightly upping the PCI bus to 35mhz, which would in turn slightly up the system bus to 70mhz, and in end result up the processor to roughly 210mhz. Sho'nuff, it worked. Further proof that the motherboard is somehow seeing position 3 shorted along with 2 and 4 which are jumpered.

The reset J16 jumper block:

J16 set for 35/70/210?

The result as shown in the PowerLogix application:

PowerLogix report

Somehow I need to figure out how to unshort that position 3 jumper. However, I could turn this into a good thing in terms of a Zif Upgrade! If I really want to be clever, I could just work with it and install a G3 ZIF that will accomodate some of the other bus multiples available to me with a permanently shorted position 3:

Multiple Jumper G3 Speed - 33.33mhz PCI/66.66mhz Bus G3 Speed - 35.0mhz PCI/70.0mhz Bus
3x 2,3,4 200mhz 210mhz
5.5x 2,3 367mhz 385mhz
6x 3 400mhz 420mhz
6.5x 1,3 434mhz 455mhz
7.5x 1,2,3 500mhz 525mhz
8x 3,4 534mhz 560mhz

Suddenly that Ibm Copper G3/450 ZIF upgrade doesn't look so bad! The strange lockups lessened greatly after removing and reseating the RAM, but I'm hoping that the lockups aren't somehow related to this shorted jumper block going intermittent on me.

Email me with any further thoughts, suggestions, information, or best of all solutions.

-----
WM