@stepleton A bit of nosing around on Quest Comp or Ebay turns up supplies of TI 74S225s, often with the suss 1983/Malaysia date codes :-( but they are out there for $2/ea on up. The 74F are quite a bit more rare and expensive, but they do exist. On the CPU board, though, it looks like they run them at the system clock rate, so stock 10Mhz parts *should* be fine. Hmm. Must've been a bad batch from TI?

Seems Three Creeks *loved* the '225, while the smaller '224 (4 x 16, rather than 5 x 16) would have sufficed in nearly every case, saving a fair bit of board space. But the '224s seem more rare for some reason. Huh.

On the MEM board where they do push the 10MHz parts to around 12MHz, the ideal replacement IC would be the AMD 74S225A -- rated for 20MHz. There are only a couple of 'em there, in the cursor video data path. I haven't been able to locate any '225As yet, though.

@stepleton Argh!! I suppose I'd better go stock up on a tube of those before getting back to reviving my hardware, knowing that they're so prone to failure. :-(

It might be worthwhile looking for faster alternatives; the '225 is only rated to 10Mhz and IIRC Three Creeks is running them closer to 12Mhz in parts of the video pipeline (using the ~80ns clock)? The 74F225 is rated for 25Mhz but faster edge rates might give grief... I was just reading about the 74FAST series from National (I think Philips bought them?) and their design guidelines -- if 3RCC paid any attention to keeping trace lengths short enough for 74S, we *might* get away with dropping in a 74F part without turning it into a microwave. :-)

@stepleton On the chance that this is PERQ related, I might suggest you look at a "thick to thin" transceiver instead. I've had very little luck on my PERQs with AUI-to-10baseT converters, but thin coax or direct connection to a DELNI seem to work much better.

I keep an old HP or 3Com hub with an AUI on it around to then bridge the home network with the 10Mbit hosts (PERQs, NeXTs, older Suns) and provide a little segregation. I've found situations where modern 10/100/1000 switches behave strangely or cause hard to diagnose issues with very old clients. Autoneg? Fuggitaboutit. And imagine your 10GbE NAS seeing NFS v2 over UDP requests come in and shrieking "are you *kidding* me? Get this stuff outta here." Virtual ketchup thrown.

@stepleton I can't believe NAG would care about code that's almost 40 years old and implemented in a language nobody beside a handful of us can actually read or use... but hey, lawyers and suits run the world. Even if we had to redact any NAG sources and just distribute the binaries, it'd still be a neat thing to put out there.

Did you find a PDF copy of the paper? My son still has amazing access through school (at least for another term or so) and I was going to nag him about searching for it this weekend...

I was kind of kidding about the ',' key. :-) None of the standard US/English layouts have it but the PERQ-2 keyboard was basically a VT100 with "OOPS" and "HELP" added? Now we can remap keys so that Flex gets all the right keycodes regardless of the vagaries of our host hardware.

@spaf The "many worlds interpretation" equivalent to this sentiment might be "I'm the version of me that consistently makes the bad decisions so that all the other versions of me are fabulously well off and cloyingly happy." Taking one for the team, as they say.

You can experience this yourself in the millisecond just after your pinky hits the Return key in that root shell.

Woo! Lookit me! I'm creatin' universes!