cri-one
ScopePhi Interposer Blueprint

ScopePhi Interposer Blueprint

$247,000
$247,000
SKU SCOPEPHI-INTERPOSER

Go to cart

Why you'd want it

The 28-layer board that connects the ScopePhi die's fine ball grid to the much wider grid of the card.

  • Who it's for: board and packaging teams that need the ScopePhi interposer design ready for an HDI fab.
  • The problem it solves: moves 0.8 mm die pitch to 2.0 mm board pitch; 162 of 165 signals routed.
  • What you get: KiCad file, Gerbers, drill, test netlist and fab notes on encrypted Blu-ray; fab fees not included.
Complete design package for the ScopePhi Interposer: 28-layer, 94.0 x 62.0 mm, 1,536 die-side pads at 0.8 mm pitch, 1,536 board-side pads at 2.0 mm pitch, Megtron 6 class HDI.
Payment is by bank transfer: through Stripe up to $999,999.99, and by wire invoice above that. Questions: christopher@cri-one.com.

ScopePhi Interposer Blueprint

The 1,536-ball interposer that bonds the ScopePhi die to the ScopePhi Card. Its job is to translate between two very different ball pitches — the die's 0.8 mm pitch on one side, the board's 2.0 mm pitch on the other — without losing a signal in the middle. 28-layer HDI with 22 signal layers, blind + buried + stacked microvia, routed to 162 of 165 real signal nets.

Spec.

Routing state. 162 of 165 real signal escapes closed; the last three outer-ring signals on rings 0–1 close with one more signal layer bump (24→28 handled; 28→32 available) or a manual GUI pass. Zero shorts. Zero mask bridges. All production artifacts generated and sitting in production-pack-28layer-v3/.

Deliverable (ships now).

Blind + buried + stacked microvias confirmed in Advanced Circuits capability as of 2026-10-24. Patent pending: USPTO 19/835,210 (umbrella).

Non-exclusive on purpose. Multiple patrons may pledge; each receives their own copy; the IP, package program and routing pipeline stay with the inventor.

Inventor. Christopher Gabriel Brown (cri-one.com), sole inventor.


Exhaustive deliverable checklist

What could go wrong, and what we have done about it

Tests and measurements (internal, in the deliverable)

References

Also patent-pending under Quantum-Reactor-Scale. The compute-core architecture of this product is additionally covered by U.S. patent-pending applications 19/865,419 and 19/865,443 (filed 2026-11-06), with the chip-scale instantiation available as QRS-116 Compute Core Blueprint, and the reactor-scale embodiments available as QRS-117 Quantum Electric Reactor Blueprint (electricity-primary) and QRS-118 Compute Reactor Blueprint (compute-primary).

For the Curious

In between the die and the board sits a third piece that most readers have never looked at closely: the interposer. It is a short, thick slab of printed-circuit material whose only job is translation. The ScopePhi die wants to talk to the outside world on a grid of 1,536 pads at a pitch of 0.8 millimetres — because that is the pitch its 858 square millimetres can afford to spend. The ScopePhi Card wants to receive the same 1,536 signals, but at a pitch of 2 millimetres — because that is the pitch a datacenter-grade PCIe board can be built and tested at without the escape routing becoming impossible to debug.

The interposer is what makes those two numbers meet. On the top face it is 37.6 by 24.8 millimetres and lays out a 48-by-32 grid of pads at 0.8 millimetres apart — the die's dialect. On the bottom face it is 94 by 62 millimetres and lays out the same 48-by-32 grid at 2 millimetres apart — the board's dialect. In between, inside the slab, 28 copper layers (22 of them for signal) carry every real net from one grid to the other, with microvias small enough to pass through 0.1-millimetre openings and stacks of microvias large enough that the signal does not meander through 300 millimetres of trace to go 10 millimetres across the chip.

There are 165 real signal nets that need to make it from the die to the card: 73 for the DDR3 memory, 64 for the PCIe host interface, 64 for the QBeam optical pairs that carry bytes between layers at light speed, and 36 for management (SMBus, JTAG, SPI, UART, charge, power). Four hundred and forty-one of the die's remaining pins carry high-speed I/O that we deliberately parked inside the chip — tied to ground with fresh microvia stitches — because they are spare capacity for a future respin and routing them out would just waste layers. Two hundred and eighty-two more management pins got the same treatment, for the same reason. That decision — what to leave on the chip — is what unblocked the routing of the real signals.

At the end of the session on 2026-10-24, 162 of the 165 real signal escapes were closed, zero shorts, zero mask bridges, with the last 3 outer-ring signals on rings 0 and 1 waiting on either one more signal-layer bump or a short manual GUI pass. Every production artifact is already generated: Gerbers, Excellon drill, IPC-2581, STEP, IPC-D-356 netlist, PDF layer book, top-and-bottom renders, and the fab-notes README written for Advanced Circuits — who have blind + buried + stacked microvia confirmed in their current capability list.

A patron pledging on the interposer is pledging on a piece that already exists in file form. The pledge covers the finishing pass on the last three signals, the fabrication of the HDI panels at Advanced Circuits, the test coupons that come with any new HDI build, and the handoff. When first interposers come back, every patron gets their AES-256-encrypted Blu-ray with the full build on it — the KiCad file, every artifact above, and the fab-notes exactly as submitted. Non-exclusive on purpose.