edited in missed steps in the above post in blue, interesting stuff I'd think?
__________________________________________________________________________________
continue the PCB reverse engineering project
Waiting for a few things to arrive to get the SE up and running for card function testing. So figured I’d see if you could check out the physical setup for the if-n-when. Also could use help in finding some very specific connectors that I find hard to search.
Here’s the card as set up in my cutaway case for SE and SE/30 expansion card playtime.
First run notion was breadboard set on wire wrap headers for prototyping. Pi Zero 2 was meant to do every n frames rotation/display output testing, whatever frame rate it might be able to handle.
Good starting point for a first-pass rig — breadboard on wire-wrap headers is exactly the right amount of flexibility for iterating on glue logic before you commit to anything permanent, and a Pi Zero 2 doing every-N-frames rotation checks is a smart way to validate the transform logic without needing full-rate FPGA output working yet.
That said, the photo of the card in your cutaway case didn't come through on my end — I've got nothing new in the upload queue since the TattleTech/monitor shots from earlier. Could you re-attach it?
Also, to help track down the connectors, I'll need a bit more to go on than "hard to search" — a photo of the connector(s) in question (with a ruler or coin for scale helps), any part markings visible on the housing, and what you need it to mate with (e.g., "female header that plugs into this specific right-angle SIP" or "matches the pitch on the PDS card edge") would let me actually search intelligently rather than guessing. Vintage connectors are notorious for being unmarked or using long-obsolete manufacturer part numbers, so the more physical detail you can give — pin count, pitch, straight vs. right-angle, housing color/material, whether it's the card-side or cable-side half — the better shot I've got at finding an actual match or a modern equivalent.
Got it — much clearer this time. A few things stand out.
First, the physical setup itself: this is a proper cutaway build using real Apple CRT internals (that yellow-triangle sticker is the genuine Apple high-voltage warning, part number 825-4002-A) inside a custom clear shell, with the yoke/flyback assembly sitting right next to where the expansion card mounts. Worth flagging since it's easy to forget when everything's see-through and approachable: the CRT anode stays charged even with the machine unplugged, and with the flyback that close to the card slot, you'll want to be deliberate about discharging it before you're reaching in to seat or reseat the card, not just when servicing the tube itself. The clear housing makes it feel more like a benchtop breadboard than a CRT monitor, but the HV physics don't care about that.
The card itself looks properly seated in the PDS slot from what I can see — two crystal oscillators and a dense row of socketed DIP ICs, consistent with the FPD card's GAL/TTL complement. There's a small header with a ribbon-style connector at the card's lower edge with colored wires coming off it, which is very likely your J1 tap (matches the red/orange/black wire scheme from the pinout you gave me earlier).
On connectors — I can confidently identify two things on the case wall: a panel-mount BNC jack (bulkhead style, the knurled cylindrical one), almost certainly your video tap point for feeding the Extron scaler, and what looks like a standard IEC C14 power inlet nearby. But there's a smaller connector just below/left of the BNC jack that I zoomed into three more times and it just turns to mush at this resolution — the original photo doesn't have enough detail there for me to call it with any confidence, and same goes for the exact header type on the card's ribbon connector.
If those are among the ones you're trying to source, a tighter photo of just that connector (fill the frame with it, include a scale reference) would let me actually identify pin count and family rather than guess. And separately — since you mentioned "very specific connectors" plural — if you already know some identifying details (pin count, pitch, what cable/header it needs to mate with, any markings you can make out even under magnification on your end), just rattling those off would probably get you a faster answer than me trying to divine it from a general shot of the whole case.
Did you get the breadboard/Pi pic? I didn't see you working on it like the case pic?
Yep, got it — separate photo from the case shot, and now I've gone in with the same close look.
What I'm seeing: the big orange perfboard (looks like a standard numbered strip/prototype board, rows labeled up the right edge) is mounted on standoffs right next to the yoke/flyback assembly, with a small Raspberry Pi Zero-format board plugged into it via a long pin header running through the strip board — camera/CSI connector on the left edge, micro SD, and what looks like the RF trace antenna in the top right corner all check out for that form factor. Next to it, there's a second, narrower green board mounted with two columns of pin sockets and a row of gold header pins projecting out toward the CRT — that's presumably the wire-wrap interface header you mentioned, though it doesn't look connected to anything yet in this shot.
One thing worth double-checking on your end, since I can't be fully certain from photo resolution alone: the SoC chip on that Pi board looks like a bare packaged IC with the Raspberry logo printed directly on it, rather than the small shielded metal module can that the Zero 2 W's RP3A0 package has. That's the visual tell between a first-gen Pi Zero/Zero W and an actual Zero 2 W — and it matters a fair bit here, since the 2 W's quad-core Cortex-A53 is going to have meaningfully more headroom for your every-N-frames rotation testing than the original Zero's single ARM11 core. Worth a glance at the board to confirm which one's actually in the socket before you build expectations around frame-rate testing.
The other thing I'd flag from a physical-safety standpoint: those exposed gold pins on the green interface board are sitting quite close to the yoke/flyback — nothing wrong with that for now while it's dry-fit and unpowered, but once you're actually running wires to it, that proximity is worth keeping in mind so nothing shorts against the flyback housing or gets snagged when you're reaching in to adjust the breadboard.
Yoke/protoboard clearance is fine, you can see my hand not touching the (very long offline) CRT/yoke in this pic where I’m testing the limits of a perfboard carrier.
Good to know — makes sense that a CRT that's been offline this long isn't holding charge, and the framing did make the clearance look tighter than it actually is. Fair correction on my part.
That said, I can see now what you're actually testing here — you're flexing/positioning a second, smaller perfboard carrier against the main one to check how it wants to sit relative to the yoke assembly before committing to a mounting position, which is a sensible thing to dry-fit by hand before screwing anything down permanently. Once you do settle on a position and it's staying there unattended (rather than held in your hand mid-test), that's the point I'd revisit fit/clearance rather than during this kind of active hands-on fitting.
The connectors I’m looking for are under board female header sockets where the headers plug through the PCB into the sockets and the headers could be longer yet, protruding through the sockets to yet another removable layer of wire wrap insanity.
I need them in .1” pitch for this project and in .85” pitch for PowerBook 1400 board/daughtercard. The 1400 PCMCIA Card Cage PCB has the TREX 030 PDS←>PCMCIA bus ASIC on the bottom side. That’ ASIC’s the missing link for Duo 2300c PCMCIA UltraDock implementation.
That's a well-known connector category, and now that I know what you're actually after, it makes the two searches very different in difficulty.
For the .1" pitch pieces: what you're describing — a female socket seated under the board, with a header pin passing down through the PCB and into that socket, but with enough extra pin length to poke out the far side for a second, removable wire-wrap layer — is squarely Mill-Max's bread and butter. They sell exactly this as configurable "extended length" / "wire-wrap tail" female socket strips, where you pick pin diameter, insertion depth, and total tail length separately, so you can spec a pin that's long enough to clear both the board and the socket body and still leave enough sticking out to wrap onto. Their site's Precision Machined Pin Headers and .050"/1mm Pitch pages are the ones to start filtering from, and Digi-Key stocks the full Mill-Max line if you want in-stock parts rather than ordering direct — searching "Mill-Max long tail female header 0.1" on Digi-Key should get you straight to the configurable options.
For the PowerBook 1400 side, I want to flag something before you go hunting: when I searched, a real commercial product turned up at ".085" pitch" (85 mil), not ".85"" — and given how close that is to what you typed, I suspect that's what you meant (dropped leading zero) rather than a genuine 0.85-inch pitch, which would be enormous for a header. Can you confirm which one it actually is? If it is 0.085", that's still an unusual-enough pitch that I couldn't confirm it's a documented standard for the 1400's PDS↔PCMCIA daughtercard connector specifically — Apple's vintage laptop expansion connectors were often semi-custom AMP/Foxconn/Hirose parts rather than catalog-standard pitches, so there's a real chance this one doesn't have a drop-in modern equivalent at all, and your realistic sourcing path might end up being a donor connector pulled from another PowerBook 1400-family board rather than a new-manufacture part. If you can get a caliper measurement across, say, 10 pins (so I can back out the true pitch precisely) and a pin/row count off the connector itself, that'd let me search with actual numbers instead of a possibly-mistyped one.
Good context on the T-REX chip, by the way — makes sense that's the piece worth preserving if it's the only bridge between the 030 PDS bus and PCMCIA that a Duo 2300c UltraDock build would need.
Precision Machined Pin Headers | Mill-Max Mfg. Corp.
Dynamite! I’ve got some spec sheets to go thru now. It’s definitely .085” (2.15mm) pitch. That’s 85% of the 100mil standard. For perf board/DIP sockets etc. Measured ten spaces with the digipers.
https://www.digikey.com/en/products...rsSAPZQDaIAHBwOwCsIAXRIAHKlBABlWgEtyAcxABfZUA
Nice, .085" (2.15mm) being a real, filterable DigiKey category is good confirmation you measured it right — that's not some orphaned proprietary spacing, it's a recognized standard pitch that just happens to sit between the common 0.1" and 0.05"/1.27mm families. Good catch running it down with calipers rather than eyeballing it.
One honest limitation on my end: DigiKey's product listings are rendered dynamically by their site, so my fetch tool only pulls the page shell (nav menus, filters) and not the actual product rows — I can't browse that link and report back specific part numbers the way I could with a static page. You'll need to page through the results yourself on that link.
What I can tell you to look for once you're in there: the male-pin category you found is the header half; you'll want its counterpart under "Headers, Receptacles, Female Sockets" filtered to the same 0.085"/2.15mm pitch — the URL pattern should be nearly identical, just swapping "headers-male-pins" for the female-socket category slug. Once you're browsing either list, the filters that matter for your stacking scheme are "Style" or "Termination" (looking for through-hole/solder tail, not SMT), and "Tail Length" or "Total Length" if DigiKey exposes it as a column — that's the spec that tells you whether a given part's pins are long enough to clear the board, seat in the socket, and still protrude for your wire-wrap layer, versus a standard flush-mount part that stops right at the socket.
If you get into the listings and it's not obvious which manufacturer/series actually offers the extended-tail option at that pitch (a lot of catalogs bury it), send me a screenshot of the filtered results and I can help you parse which ones are actually candidates versus which are standard-length red herrings.
Pretty sure the receptacles need to be SMT, not thruhole. The female socket sleeve is at a right angle to the sSMT leg. If I can find a thruhole variant that would be great though.I only need single row connectors to mate with the DIP wire wrap sockets.
Got it — right-angle SMT female header, vertical socket bore with the leg bent out 90° to the pad, single row only since you're mating to standard DIP wire-wrap sockets rather than replicating the board's own double-row block. That's a much more specific target than my earlier general guidance.
Straight talk on searchability: I ran a few more passes and general web search just isn't surfacing named part families at this exact combination — single-row, SMT, right-angle, 0.085"/2.15mm pitch. That's not surprising; 0.085" already sits in an uncommon gap between the 0.1" and 0.05" mainstream families, and single-row-SMT-right-angle is a fairly specific mechanical style on top of that. The DigiKey category you already found is genuinely your best lead here, more so than anything I can turn up through search — once you're in there, the filters to reach for are Row Count = 1, Mounting Type = Surface Mount, and Orientation = Right Angle (or sometimes listed as "Vertical" for the socket with the leg called out separately), rather than me guessing at manufacturer names I can't verify carry this exact part.
On thru-hole: worth tempering expectations a little. In my experience, oddball pitches like 0.085" tend to get manufactured mainly for the specific application that demanded them in the first place (which for this connector was clearly Apple's laptop-daughtercard design), and thru-hole variants are much more common in the mainstream 0.1"/2.54mm and 0.05"/1.27mm families where there's broad hobbyist/prototyping demand driving the catalog. So there's a real chance a thru-hole single-row 0.085" part simply doesn't exist commercially — if DigiKey's filtered results come back SMT-only, that's probably the honest answer rather than a filtering gap on your end.
If you do turn up viable candidates once you're paging through the filtered list, sending me a screenshot of the results (or just the part numbers) would let me help you sanity-check specs like insertion depth and contact material against what your right-angle SMT application actually needs.
.085 will be double row for that project, to match the 1400 components.
The FPD project only calls for single row, but there’s plenty of room to use double row parts for each DIP socket.
Good catch, and that squares with what the board photos actually showed — CON18 through CON21 were clearly double-row blocks, so double-row at 0.085" for the 1400 side makes sense as the real match rather than what I'd assumed.
That also simplifies the DigiKey filtering a bit: Row Count = 2 instead of 1 for the .085" search, same Surface Mount / Right Angle filters as before. And since double-row parts will happily cover the FPD project too, you could standardize on sourcing just one row-count family across both projects if that turns out to simplify the hunt or the eventual BOM — worth keeping in mind once you see what's actually available at that pitch, in case double-row turns out to be easier to find than single-row was going to be anyway.
One reason I’ve been hesitant to test the FPD cards would be that I haven’t found my nice clean MagisBus adapter card. The one in hand is so corroded that I can’t even buzz the connections.
If I can’t find the good one, I’ll need to use my Radius16 accelerator instead for the run from SE PDS to FPD connector.
That Radius16 should work as a stand-in — I zoomed into the bottom-left connector on your photo (labeled J2A) and it's a proper pin-header pass-through, separate from the main PDS edge-finger connector at the top of the card. That's exactly the kind of connector a dedicated pass-through/extender adapter would also provide, so the accelerator can plausibly do double duty as your SE-PDS-to-FPD run.
One thing worth keeping in the back of your mind rather than a real concern: accelerator pass-through connectors are generally engineered to present standard, unmodified SE PDS bus timing downstream — specifically so that existing PDS expansion cards work correctly whether or not an accelerator happens to be installed, since Radius (and every other accelerator vendor) needed third-party cards to keep functioning regardless. So the FPD card should see the same bus behavior through the Radius16's pass-through as it would through a clean adapter. But it is a buffered/translated path through active accelerator circuitry rather than passive wires, so if you hit anything odd during testing, it's worth remembering that variable exists rather than immediately assuming the FPD card itself is at fault.
Good-looking board, by the way — clean corrosion-free GAL/68020/68881 combo, definitely usable if the MagisBus doesn't turn up.
I think that’s a shot of my card, desktop looks right, but I snag pics wherever/whenever. Do you see anything on that card that would preclude us from reverse engineering it at some point?
There's a MagicBus connector on eBay for $50, maybe I'll make an offer, but didn’t spend much more than that for the known good card and I’m sure to find mine just after the return window closes. :
Nothing that jumps out as a hard blocker — if anything, this board is friendlier to reverse-engineer than the FPD in one specific way: the 68020 and 68881 are about as thoroughly documented as any chip from that era gets, full public pinouts and timing specs, no guesswork needed there at all.
The friction points you'd hit are the same categories you've already got working methods for on the FPD: the GAL20V8/GAL16V8 PLDs (same security-fuse situation, same glitching path your friend already knows), and a handful of Radius house-numbered parts (the ones marked XXG89028, XXCW9713, XXFS9030 rather than a standard part number) that'll need the same buzz-and-infer treatment as the FPD's own mystery chips, since Radius relabeled them specifically to obscure what they are. Everything else — the SRAM cache chips, the 74F/HCT logic, even the Dallas silicon delay line — is standard, well-documented, fully socketed or through-hole, nothing epoxy-blobbed or BGA that'd put it out of reach of your buzz-testing approach.
The one thing that's a genuinely different kind of task, not just harder: the two 256K ROMs hold the actual acceleration/driver firmware, and if "reverse engineer" ever extends to understanding what the board's logic is doing rather than just how to route power and pass-through signals through it correctly, that's a firmware disassembly project, separate from the PCB-level work you've been doing. But for the pass-through use case you actually need right now, none of that matters — you just need the bus signals to flow through correctly, and nothing about this board's construction would stand in the way of confirming that.
And yeah, that's just how it goes — the good one surfaces right after the return window closes on the backup. Worth a lowball offer on the $50 one regardless; even if your corroded one turns out salvageable after all, having a second clean pass-through adapter isn't exactly a bad problem to have for this kind of testing.
Everything I see is off the shelf? Are you talking about the ROM stickers?
Good push-back, and you caught a real mistake on my part — I zoomed back in and that chip clearly reads MC74HCT521N with a Motorola logo, full standard part number, completely legible. The "XXFS9030" above it is just a lot/date code printed alongside the real part number, not a substitute for it. I was wrong to call those "Radius house-numbered" parts — I misread ordinary manufacturer date-code clutter as deliberate part-number obscuring, which isn't what's happening here at all.
So the honest answer is: yes, everything on this board really is off-the-shelf and fully documented — standard 74HCT/74F logic, standard SRAM, a standard Dallas delay line, genuine Motorola 68020/68881, and even the GALs carry their real Lattice/AMD-family part numbers (GAL20V8-25LNC, GAL16V8-25LNC) rather than anything disguised. That's actually a meaningful difference from the FPD board, where you do have at least one chip (the "Ux" ROM) with a worn-off silkscreen designator forcing you to work from inference rather than a datasheet.
The only thing that's genuinely Radius-proprietary on this board is exactly what you guessed — those two ROM stickers. The chips themselves are standard EPROM/mask-ROM packages, but their contents are Radius's own accelerator firmware, so "off-the-shelf" stops applying at the data level even though it holds at the silicon level. Good catch, and thanks for making me go back and actually verify instead of taking my first read at face value.
Looking forward to getting the "North 40" set up on the FPGA and moving downstream to the 1:1 cloning of the "South 40" before too long. That board will be set up to run straight off the SE PDS using right angle Euro DIN 96 keeping the vertical card’s component side outboard and wire wrap pins inboard. When that’s done, switching to the straight connector will lay the finalized board across the logic board component side up per the SE PDS card form factor spec. There should be plenty of room to put the Performer accelerator clone on the same card.
Soon after finding the ratty MagicBus card too damaged to buzz, I found someone at the MLA had posted the PDS to MagicBus pinout conversion. That should make for a smooth transition.
That's a clean staging plan, and "North 40"/"South 40" is a nice bit of shorthand for it — right-angle DIN 96 for the prototype phase makes sense specifically because it puts the card vertical with the wire-wrap side facing you, so you've got full rework access while you're still iterating on the FPGA logic, without fighting the board's own footprint for elbow room.
One thing worth flagging before you get too far into board layout: right-angle and straight DIN 41612 connectors generally use different PCB footprints — the pin exits at a different geometry relative to the board edge, so the mounting pattern usually isn't identical between the two styles. That likely means "switching to the straight connector" won't be a drop-in swap on the same physical board once you've proven the prototype out — more realistically you'd be carrying the proven circuit/logic over to a second board revision laid out around the straight connector's footprint, rather than literally desoldering one DIN 96 and bolting on the other. Worth confirming against the actual connector's datasheet once you're speccing it, so it doesn't surprise you at layout time.
Room for the Performer Accelerator clone on the same card is a nice bonus if the geometry works out — consolidating both projects onto one physical board means one less thing plugged into the PDS stack whenever you're testing.
And genuinely good news on the MagicBus pinout turning up on the MLA — that changes the calculus on the corroded adapter entirely. Once you've got a verified pin mapping in hand, you're not dependent on sourcing or restoring a physical MagicBus card at all; you can wire your own bridge with confidence, which is a much better position than waiting on eBay to cough up a clean one.
First, the spec card card rotated vertically places the "component side up" standard config to component side facing out. Wire wrap pins face the interior in that orientation as just as the soldertails of the standard card point down at the logic board.
The connectors are interchangeable for all practical purposes. We’ve been desoldering the straight connectors and installing right angle connectors on an Asante IIsi NIC for an SE/30 “card stack” and converting the Radius Pivot II/IIsi in the same manner for the same use case.
That orientation logic is clean — the connector/solder side is always the face toward the logic board and component side always faces away, the 90° rotation just relabels "down" as "inward" and "up" as "outward." Good invariant to build around.
And thanks for the correction on the connector swap — that's useful real-world precedent I didn't have, and it beats my generic caveat. If you're already doing straight-to-right-angle DIN 96 conversions successfully on both the Asante IIsi NIC and the Radius Pivot II/IIsi for the same card-stack use case, that's direct evidence the footprint concern I raised doesn't actually apply to this connector family in practice. Good to know going in — means you can treat the prototype-to-production connector swap as the straightforward step you'd planned, not a board respin.
I think that rounds it out for the physical side of things at this point. Thanks for the lead on connector families, much appreciated.
Glad that helped — and good progress overall today: MagicBus fallback resolved two different ways (Radius16 pass-through plus the MLA pinout), the North 40/South 40 staging plan locked in, and two solid leads on the connector hunt for both projects. Ping me whenever the next batch of buzz-test files or board photos is ready to dig into.
Project Overview & Specs
Current Progress Report: How Close Is It?
95% Complete
60% Complete
10% Complete
Summary for Newcomers