In trying to add an Ethernet card to my 575, I encountered a documented hardware bug in a small production range which the community should be aware of.
Two known good LC PDS Ethernet cards exhibited the same behavior: they would power on with the machine after a hard reboot (link LED on with a cable connected), then fade out and die as soon as the boot process began, before SCSI was even polled.
But I was able to get the card to run if I held it down with my finger... so, loose contact or cracked solder joint, right?
Nope. Turns out, the magic was just having my finger — or a screwdriver — *touching* pin C7 on the card or the underside of the slot connector.
So I traced it out against the schematics. PDS slot pin C7 (RST_EXP) is supposed to run to a resistor — R74, which it does — and from there to a 74F00 quad NAND gate labeled U33.
Which... was not there. There's a blob of adhesive there, but no component has ever been soldered on where U33 is supposed to be.
This finally led me to search the Tech Info Library on my own site for anything relevant to the 575. And... yeah. Should have done that a couple of weeks ago when the card failed to work.
Per TIL 17047: "Some Macintosh LC and Performa 575, 577, and 578 computers manufactured from October to November 1994 do not properly recognize option cards installed in the PDS or communication slots. This is caused by the absence of a component at location U33 on the bottom side of the logic board near the slots..."
So they just never installed the damn thing.
Thanks, Spindler.
And the solution per Apple? A logic board swap, but only if the customer has reported the problem. Which the original owner of my benighted machine apparently did not.
Not only was U33 missing, but so was U32 — supposed to be a 74F74 dual flip-flop — and the 0.1µF caps at C89 and C90.
The schematic appears to read "Do NOT stuff U32, U33, C91, C89 when using PrimeTime2". I would guess that these machines were meant to get an upgraded PrimeTime ASIC but didn't, hence the unpopulated U32 and U33. (I don't know what U33 is meant to do, and the machine is working fine without it. @Phipli suggested that U33 should not be populated and so far it doesn't seem necessary.)
I acquired the missing U33 IC (fortunately, easily available from parts houses) and capacitor, and fitted them, and I now have working Ethernet!
(I bought a few extra ICs in case anybody else runs into this issue, though I suspect I may have the only surviving 57x that was built in the narrow production window where this problem occurred. Let me know if you want one and I'll send to you, along with the tiny capacitor at C89.)
Two known good LC PDS Ethernet cards exhibited the same behavior: they would power on with the machine after a hard reboot (link LED on with a cable connected), then fade out and die as soon as the boot process began, before SCSI was even polled.
But I was able to get the card to run if I held it down with my finger... so, loose contact or cracked solder joint, right?
Nope. Turns out, the magic was just having my finger — or a screwdriver — *touching* pin C7 on the card or the underside of the slot connector.
So I traced it out against the schematics. PDS slot pin C7 (RST_EXP) is supposed to run to a resistor — R74, which it does — and from there to a 74F00 quad NAND gate labeled U33.
Which... was not there. There's a blob of adhesive there, but no component has ever been soldered on where U33 is supposed to be.
This finally led me to search the Tech Info Library on my own site for anything relevant to the 575. And... yeah. Should have done that a couple of weeks ago when the card failed to work.
Per TIL 17047: "Some Macintosh LC and Performa 575, 577, and 578 computers manufactured from October to November 1994 do not properly recognize option cards installed in the PDS or communication slots. This is caused by the absence of a component at location U33 on the bottom side of the logic board near the slots..."
So they just never installed the damn thing.
Thanks, Spindler.And the solution per Apple? A logic board swap, but only if the customer has reported the problem. Which the original owner of my benighted machine apparently did not.
Not only was U33 missing, but so was U32 — supposed to be a 74F74 dual flip-flop — and the 0.1µF caps at C89 and C90.
The schematic appears to read "Do NOT stuff U32, U33, C91, C89 when using PrimeTime2". I would guess that these machines were meant to get an upgraded PrimeTime ASIC but didn't, hence the unpopulated U32 and U33. (I don't know what U33 is meant to do, and the machine is working fine without it. @Phipli suggested that U33 should not be populated and so far it doesn't seem necessary.)
I acquired the missing U33 IC (fortunately, easily available from parts houses) and capacitor, and fitted them, and I now have working Ethernet!
(I bought a few extra ICs in case anybody else runs into this issue, though I suspect I may have the only surviving 57x that was built in the narrow production window where this problem occurred. Let me know if you want one and I'll send to you, along with the tiny capacitor at C89.)
