Dove MacSnap 5.0 RAM Upgrade Upgrade

[demik]

Whoever made this connection clearly understood this; it is more than probable that the R4 and C4 values are not providing the correct RC time constant for the SCSI interface to work reliably on your specific board.

Demik is the right person to test whether this bodge connection improves the reliability of the MacSnap SCSI replica!View attachment 96539

Yes that's what's finicky about this board. That RC circuit. I mean it's Voodoo level: in a couple of boards I had to "pair" the ATFs as well (like you do on audio amps), in order to have the correct sink/source capabilities so the RC circuit would do the correct timing.

The orignal EP320 were likely running in turbo mode to avoid that.

This is actually really interesting! It’s the first time I’ve seen a connection like this.

It looks like it’s injecting the LDS signal directly from the TSM PAL (Pin 9) into the SCSI board at J1, pin 2. I’m not 100% sure on the exact logic yet, but if I had to guess, it’s probably a failsafe to tighten up the data latching for the SCSI chip.

I also never seen this jumper as well. That's very interesting, @pezter22 thank you for posting this.
Reproducing the jumper on the RE board is worth a try. The issue is I don't have any others boards that are instable around anymore.

@pezter22 also noted that the SCSI termination circuit was removed from his board, which is likely related. If the board consumes less power and the voltage are more of what's intended, the RC circuit timing is more correct.

It is no wonder the original and replica MacSnap SCSI boards are so finicky.

The RC time constant from R4 and C4 is clearly designed to mimic a full CPU access cycle, which always takes 4 clock cycles on these early Macs. The problem is that any shift in that timing, whether from aging components, voltage drops on a tired analog board, or extra capacitance from crusty pin headers, will easily push that constant out of spec. Once that happens, your SCSI data integrity goes right out the window.

The primary goal for DOVE designers was creating a product that a regular user could simply press into a socket without having to fire up a soldering iron. It was a brilliant marketing and usability decision for a product that was only expected to be relevant for a few years.

This board remains the absolute best way to open up a closed system like the 512K to the modern storage world. Getting a SCSI emulator running on a 512K completely transforms the machine from a neat display piece into a highly usable system, especially when paired with a fully working 4MB RAM expansion board.

It really looks like adding that bodge wire is almost mandatory if you are having SCSI issues on a Mac 128K or 512K. It is the only way to swap over to solid, CPU-synced timing. Fortunately, the PLD logic is already tailored to use the /LDS signal. All you need to do is tap /LDS from the motherboard and connect it to pin 2 of J1.

Since J1 was removed from the replica board made by Demik, the owner of a replica board will instead need to connect the bodge wire to the junction of R3 and pin i9 of the DOV1ß PLD. It should work, though it will definitely need to be tested first.

All of this is thanks to pezter22, who published that ultra-rare photo of this SCSI board showing the bodge wire!

As @Builder68 suggested, you can tap on the R3 side that's connected to the GA

These are great boards and Demik's recreation is fantastic.

You are too kind it was a team effort. @Builder68 found what was wrong with the prototype, I was just staring at it without ideas
As far as other advices, everything @Builder68 said is correct.

I would try to find one combinaison that work first, then try to fix the rest. Am I wrong assuming the OG board doesn't work as well or does it work ? Do you have real SCSI drives from back then just to double check ?