Bringing together everything from the last couple of days of digging and bench testing, since it's spread across a few posts now.
Myths busted
- "Only the 02-02 does 16MHz" is false. Multiple 00-02s here run clean at 16MHz.
- "WD stopped making the 02-02" is false. I have factory 02-02 stock date coded 1994, one 1995.
- "Rated at 10MHz" isn't invented, it comes straight out of WD's own clock duty timing table, but neither chip is specified anywhere near 16MHz, on either revision, in either datasheet.
- Nobody who ever made the original claim published how many chips they tested. I now have a real sample, with a control chip re-verified between every swap and every 16MHz failure re-checked at 8MHz to prove it's a good part.
Code: Select all
Marking Date Result at 16MHz
------- ---- -----------------------------
02-02 9427 OK, x2 units
02-02 9522 OK
00-02 8614 OK
00-02 8618 FAIL x3, all load fine at 8MHz
00-02 8622 OK x2, FAIL x1, the fail loads fine at 8MHz
00-02 8627 OK x3
The bit that actually matters, and the disappointing part
I went as far as matching a working 8622 against a failing 8622 down to the lot code stamped on the underside. Identical, top and bottom. Same lot, opposite result. Two 8627s from the same production week also gave opposite results earlier.
So date code doesn't predict it, lot code doesn't predict it, and country of assembly doesn't predict it either, since that's confounded with revision anyway. Every single thing you can read off the outside of the package has now been checked and none of it tells you anything. The honest conclusion is suck it and see, chip by chip. Not the satisfying answer I was hoping to end up with, but it's where the evidence actually goes.
The one real pattern
8618 is the exception. Three out of three failed, reproducibly, and all three share the same lot code. That's still a genuine result, just a narrow one: it says something about that specific lot, not about the 00-02 revision in general.
What this confirms
This lines up with what I found years ago and already had on the site: 16MHz is an overclock on any 1772, 00 or 02, and whether a given chip takes it is unpredictable per chip. What's new is having an actual sample behind that statement instead of just my word for it, and having finally killed off the idea that the 00 flatly won't do it.
Still open
- Whether 00 and 02 differ in reliability at all. Can't say yet. My 02 history is years of real 16MHz exposure through the kits, but only ever in short access-only bursts, never sustained. Today's 00 testing is hours of continuous 16MHz. So it's a duty cycle mismatch between the two datasets, not that the 02s have never seen 16MHz.
- Sustained 24/7 operation. Nobody has tested that, including me. 4-5 hours is the longest clean run so far.
- All of this used an external oscillator, not a Shifter derived clock. I found the Shifter clock intermittent on an original STFM about 30 years ago and never got to the bottom of why. None of today's results say anything about a chip fed a worse clock.