After Porchy from Jammaarcade got us across the finish line with reverse engineering the original HAL16L8, i thought the functions might be possible to replicate using discrete logic.
This would of course be to benefit those that are making an SE Reloaded board, who don't want to, or can't salvage their GLU chip, and don't have the tools to program an ATF16V8 with the .jed that was reverse engineered.
Of course, this may require some very, very tiny fiddly parts...
Diving deeper into the 'Inside Macintosh' tomes, I actually came across a selection of diagrams that detail exactly what the GLU does, in terms of functionality.
Now, this looks pretty simple to me - it's literally just a few tri-state buffers and some inverters. Basically, the HAL that apple had made, essentially smashes a 74LS04 and two 74LS125's together, to save on chip count and extra signal routing.
So I fired up KiCAD and spent about an hour hammering this out:
And then routed it into a 4-layer PCB, specifying VQFN parts, that sit inside the footprint of the original HAL16L8!
Now, this is using VQFN-14 & SOT-353 parts that are approximately 3mm square at their largest.
U1 is a 74AHCT04
U2 is a 74AHCT125
U3 is a 74LVC1G125 (a single gate of an LS125 in it's own package)
No voltage regulation needed as the parts are rated for 5v operation. I've yet to do a run of these just yet. I'm wondering if i should add additional decoupling capacitors to the design, or if the single decoupling cap designed for the original GLU will suffice.
Also - i think i might need someone to double check my schematic before i'm happy to send a few of these to manufacturing.
When the Macintosh Classic came out, because it dropped the 2nd floppy drive along with the widespread adoption of SMD components, the functionality could be replicated by discrete logic rather than using a HAL/PAL.
Instead, the Classic used a single 74LS368 and the 74LS04 was omitted for the clock signal entirely, with the oscillator feeding directly into the BBU without first being buffered - cost savings at work there!
This would of course be to benefit those that are making an SE Reloaded board, who don't want to, or can't salvage their GLU chip, and don't have the tools to program an ATF16V8 with the .jed that was reverse engineered.
Of course, this may require some very, very tiny fiddly parts...
Diving deeper into the 'Inside Macintosh' tomes, I actually came across a selection of diagrams that detail exactly what the GLU does, in terms of functionality.
Now, this looks pretty simple to me - it's literally just a few tri-state buffers and some inverters. Basically, the HAL that apple had made, essentially smashes a 74LS04 and two 74LS125's together, to save on chip count and extra signal routing.
So I fired up KiCAD and spent about an hour hammering this out:
And then routed it into a 4-layer PCB, specifying VQFN parts, that sit inside the footprint of the original HAL16L8!
Now, this is using VQFN-14 & SOT-353 parts that are approximately 3mm square at their largest.
U1 is a 74AHCT04
U2 is a 74AHCT125
U3 is a 74LVC1G125 (a single gate of an LS125 in it's own package)
No voltage regulation needed as the parts are rated for 5v operation. I've yet to do a run of these just yet. I'm wondering if i should add additional decoupling capacitors to the design, or if the single decoupling cap designed for the original GLU will suffice.
Also - i think i might need someone to double check my schematic before i'm happy to send a few of these to manufacturing.
When the Macintosh Classic came out, because it dropped the 2nd floppy drive along with the widespread adoption of SMD components, the functionality could be replicated by discrete logic rather than using a HAL/PAL.
Instead, the Classic used a single 74LS368 and the 74LS04 was omitted for the clock signal entirely, with the oscillator feeding directly into the BBU without first being buffered - cost savings at work there!
