USB-C Thumb Drive Design

flux

These are real architecture options. My ranking:
  1. Best product path: Option 4 — SM2320 + matched NAND
    • Native USB-C / USB 3.2 Gen2x2-class.
    • Public specs suggest up to ~4 TB and ~2 GB/s class performance.
    • Best fit for “own PCB + fast.”
    • Main blocker: likely needs Silicon Motion / NAND partner support for firmware tools, NAND AVL, and full datasheet.
  2. Best DIY/prototype path: Option 1 — SM3281AB + NAND + HD3SS3220 mux
    • More accessible tooling appears to exist.
    • USB 3.0-class, but lower performance and more trial-and-error.
    • Needs mux because the controller exposes one fixed SuperSpeed lane.
  3. Good silicon but support-dependent: Option 3 — Phison PS2251-17 / U17
    • Strong USB 3.2 native storage controller.
    • But public production tooling for U17 looks much less clear than older Phison parts.
  4. Not for this product: Option 2 — FC1179AB USB 2.0
    • Useful learning build, but too slow.
  5. Future/overkill: Option 5 — SM2322/SM2324
    • Very fast, but newer and likely harder to access/support.
I documented the full review here: USB Flash Controller Options Review.
My recommendation: pursue SM2320 first if we can get partner/vendor support; otherwise prototype with SM3281AB + known-supported NAND.
D

denisagarkov

Yes, pursue SM2320 first and list exactly what I need to request from Silicon Motion.

flux

I’ll make SM2320 the primary architecture and create a concise vendor-request checklist you can send to Silicon Motion/NAND partners.

flux

Done — I made SM2320 the primary architecture and created the exact Silicon Motion request checklist here: SM2320 Vendor Request Checklist.
In short, request:
  • Full SM2320 datasheet, ball map, footprint, and reference schematic.
  • USB-C implementation guidance, especially whether an external orientation mux is required.
  • NAND approved-vendor list with exact Kioxia/Micron part numbers and supported capacities.
  • Firmware / mass-production tool access and VID/PID/product-string configuration flow.
  • Power, thermal, high-speed layout, reflow, and production-test guidance.
  • NDA, MOQ, lead time, lifecycle, and approved partner information.
Do not design the final SM2320 schematic from the public product brief alone; we need the full engineering package or partner support first.
Draft email and plan
Draft SM2320 email
Update project spec
Create backup plan
Explain NDA basics
Sign up to send

Files

Schematic

Layout

3D View

USB Flash Controller Options Review
Context
User provided five candidate architectures for an own-PCB USB-C flash drive. The goal is to identify which path is realistic for a fast product and what dependencies remain.
Option 1 — SM3281AB + raw NAND + USB-C mux
Verdict: plausible DIY/prototype path, but not lowest risk.
  • Controller: Silicon Motion SM3281AB / SM3281 family.
  • Interface: USB 3.2 Gen1 / USB 3.0-class, single-channel flash-drive controller.
  • NAND: raw NAND such as Kioxia BiCS stacked TLC.
  • USB-C: needs a SuperSpeed orientation mux because the controller exposes one fixed SuperSpeed lane.
  • Tooling: Dyna Mass Storage Production Tool / SMI MPTool-type flow appears publicly available from community/archive sources.
  • Risk: full engineering datasheet/pinout and approved NAND list may not be public. MPTools from unofficial sources are operational/security risks.
Option 2 — FC1179AB + raw NAND, USB 2.0 over Type-C
Verdict: easiest learning path, not suitable for the fast product.
  • Controller: FirstChip FC1179AB.
  • Interface: USB 2.0 only.
  • Expected real speed: roughly 25–40 MB/s.
  • Capacity: could potentially support high capacities if NAND/tool support aligns, but performance is the blocker.
  • USB-C: simple D+/D− wiring plus CC pull-downs; no SuperSpeed mux required.
  • Tooling: FirstChip FC1179 MPTool V1.0.7.2 / 2024-02-21 tooling is visible publicly.
  • Use only for a learning board or low-speed utility drive.
Option 3 — Phison PS2251-17 / U17 + raw NAND + USB-C mux
Verdict: technically strong, but likely requires Phison ecosystem support.
  • Controller: Phison PS2251-17 / U17.
  • Interface: USB 3.2 Gen2-class native USB storage controller.
  • Performance class: public Phison material references roughly 1000 MB/s-class behavior.
  • Capacity: public product material references roughly 250 GB to multi-TB-class use.
  • USB-C: likely requires mux/orientation handling unless controller/reference design includes Type-C orientation logic.
  • Tooling: public MPALL references for this exact modern controller are weak; old MPALL tools mostly target older Phison controllers.
  • Strong silicon, but not recommended unless Phison or a module/NAND partner will provide datasheet, firmware, NAND AVL, and production flow.
Option 4 — Silicon Motion SM2320 + NAND, native Type-C
Verdict: best real-product architecture if vendor access is available.
  • Controller: SM2320 portable SSD controller.
  • Interface: USB 3.2 Gen2x2, native Type-C support.
  • Public specs: 4 NAND channels, up to 4 TB, up to about 2100/2000 MB/s read/write, 9 × 9 mm TFBGA.
  • NAND: 3D TLC/QLC Toggle/ONFI NAND.
  • USB-C: native Type-C CC logic support reduces/eliminates external orientation mux complexity.
  • Risk: full datasheet, layout guidance, NAND validation list, firmware/tooling are likely NDA/customer-channel only.
  • Best for product-grade fast storage, but needs Silicon Motion or partner support.
Option 5 — Silicon Motion SM2322 / SM2324 + NAND, native Type-C/USB4
Verdict: excellent future direction, overkill for first board.
  • SM2322: USB 3.2 Gen2x2 20 Gbps, up to 8 TB, about 2100/2000 MB/s class.
  • SM2324: USB4 40 Gbps, integrated PD 3.1, up to 32 TB, about 4000/4000 MB/s class.
  • Risk: newer, more complex, likely even more dependent on vendor NDA/support and high-speed PCB design expertise.
  • Not recommended for Rev A unless we already have vendor access.
Recommendation
For a serious fast own-PCB product, pursue Option 4: SM2320 + matched NAND as the primary architecture. It best matches the product goal: native USB-C, high speed, high capacity, no donor thumb drive, and single-controller architecture.
For a lower-risk experimental board, pursue Option 1: SM3281AB + supported NAND + HD3SS3220 mux because tooling appears more accessible, but expect more trial-and-error and lower performance.
Avoid Option 2 for the fast product. Treat Option 3 as strong only if Phison support is available. Defer Option 5 until after a successful SM2320/SM3281-class prototype.
What must be obtained before schematic capture
For the chosen controller:
  1. Full pinout and package drawing.
  2. Reference schematic.
  3. PCB layout guide, including USB SuperSpeed and NAND routing rules.
  4. Supported NAND list / AVL with exact NAND part numbers and capacities.
  5. Firmware or mass-production tool access.
  6. USB VID/PID/product-string configuration process.
  7. Power tree requirements and decoupling values.
  8. Crystal/clock requirements, if any.
  9. Boot/config strapping pins.
  10. Production test/programming workflow.
Current design implication
The existing USB-C front-end can support the SM3281/Phison style if a mux is added. For SM2320/SM2322/SM2324, the schematic should pivot to the controller vendor reference design and likely remove the external USB-C SuperSpeed mux unless the datasheet explicitly requires one.

  • Context

  • Option 1 — SM3281AB + raw NAND + USB-C mux

  • Option 2 — FC1179AB + raw NAND, USB 2.0 over Type-C

  • Option 3 — Phison PS2251-17 / U17 + raw NAND + USB-C mux

  • Option 4 — Silicon Motion SM2320 + NAND, native Type-C

  • Option 5 — Silicon Motion SM2322 / SM2324 + NAND, native Type-C/USB4

  • Recommendation

  • What must be obtained before schematic capture

  • Current design implication