Yes—combining several smaller raw NAND chips behind one SSD controller is how many SSDs are built. For example, four 512 GB NAND packages could provide roughly 2 TB of raw capacity, with some reserved for reliability and management.
The architecture would be Mac → ASMedia USB4 bridge → one NVMe SSD controller → multiple NAND chips. The Mac would see one drive, so this approach does not inherently prevent Apple silicon Mac compatibility.
The difficulty is getting an SSD controller with firmware that explicitly supports the exact NAND parts, plus usable design documentation. NAND chips are not interchangeable. This is possible, but considerably harder than using one complete solder-down SSD—and combining multiple complete SSD chips is a different architecture that needs additional switching or bridging.
Agreed. We’ll favor one complete solder-down NVMe SSD for the simpler design, and treat capacities beyond the largest qualified single device as a separate, multi-SSD architecture with its own power, firmware, and Mac compatibility validation. No board changes yet.
Understood—the Mac must see one drive, not multiple drives combined through macOS software. A single solder-down SSD already meets that requirement; a larger, multi-SSD version would need hardware aggregation that exposes one logical storage device over USB4.
A PCIe switch alone does not combine SSDs into one drive. That larger version would need a suitable hardware RAID/storage controller and verified Mac-compatible firmware. Also, combining drives for capacity does not automatically provide redundancy—failure of one SSD could lose the entire volume.
Understood—we’ll keep this as a concept and leave the board unchanged. The preferred direction is a single solder-down NVMe SSD, with a separate higher-capacity architecture only if needed; either way, the Mac must see one drive.
This section overrides stale historical statements later in the file.
USB-C/PD ownership: U10 TPS65987D is the sole CC/PD policy owner. J4 CC1/CC2 connect to U10, not U1. Hardware straps select the dead-battery BP_NoWait / Configuration 2 fallback path, and U11 provides the external SPI configuration store. Production remains blocked until the SPI image is validated to allow only 5 V fallback and a fixed 9 V / 3 A operating contract; it must not request 15 V or 20 V.
Power path:VBUS_PORT -> U10 PP_HV1 -> VBUS_RAW -> U3 TPS259474 -> VBUS_PROT_9V. U10's VBUS and PP_HV pins are rated for 4/4.5 V to 22 V recommended operation and 24 V absolute maximum. U3 is rated for 2.7 V to 23 V recommended input operation and 28 V absolute maximum.
Strict 9 V gating: U3 OVLO remains approximately 10.27-10.62 V rising, so unintended 15 V/20 V contracts cannot energize downstream regulators. D12 SMAJ12A is retained only under the strict 9 V image requirement: its 12 V standoff is not compatible with continuous 15 V/20 V operation.
U3 EN/UVLO: R34 = 470 kΩ and R35 = 200 kΩ produce nominal EN voltages of 1.418 V at 4.75 V, 1.493 V at 5 V, and 2.687 V at 9 V. With 1% resistors and datasheet threshold limits, calculated rising turn-on is 3.908-4.155 V and falling turn-off is 3.555-3.792 V. D13 clamps abnormal high-VBUS pin voltage with only about 23 µA estimated diode current at 23 V. Canonical connectivity is correct (U3:EN/UVLO belongs only to EFUSE_EN_UVLO_SAFE), but stale legacy wire geometry still visually joins the pin to VBUS_RAW; remove that geometry before release/export.
9 V budget: At the 8.55 V PD minimum, the prior 18.903 W peak converter-input estimate requires approximately 2.25 A, leaving about 0.75 A / 25% margin to the 3 A contract. The full-power mode is not approved on 5 V fallback.
SSD switch: U7 is now TPS22965DSGR, not TPS22995. It is rated for 6 A continuous and 8 A for Current-state notice (2026-08-27): The external-PD implementation later in this file supersedes the earlier fixed-5 V/current-state descriptions below. Those paragraphs are retained as the pre-change analysis that motivated the new architecture.
The current design is not safe/complete as an unrestricted USB-C 5 V-only sink.
A realistic typical workload requires about 2.08 A at 5.0 V or 2.20 A at 4.75 V. It can operate only when the source advertises 3 A and the SSD remains near the assumed typical workload.
The full planning peak requires about 3.92 A at 5.0 V or 4.14 A at 4.75 V. This exceeds USB-C 5 V / 3 A and therefore requires either a negotiated higher-power PD contract or enforced load/current limiting.
The current schematic contains no separate PD controller. ASM2464PDX is the only active Type-C/PD sink controller connected to CC1/CC2; its datasheet identifies integrated UFP CC logic, built-in 5.1 kΩ Rd resistors, and fixed-5 V sink/BMC operation.
The current schematic does not expose a verified mechanism that converts CC advertised-current detection into dynamic eFuse/SSD-load enforcement. Until ASMedia documents that policy or a verified controller/current-limit architecture is added, operation from Default USB / 900 mA / 1.5 A sources is a hard blocker.
A speculative 9 V PD architecture must not be enabled as drawn: U1 VBUS is a 5 V-tolerant input directly tied to VBUS_RAW. Raising raw VBUS above 5 V without a vendor-approved VBUS sense/power-path arrangement would overstress U1.
The controller power tree is not functionally complete: 3V3_LOGIC has no verified regulator source, while the 1.9 V and 1.05 V converters are enabled directly from protected VBUS. This does not implement the required VCCH-first sequence.
Live schematic changes made
Connected the actual USB-C receptacle J4:
J4 CC1 to USB_C_CC1.
J4 CC2 to USB_C_CC2.
All four J4 VBUS contacts to VBUS_RAW.
All four J4 signal-ground contacts to GND.
Preserved the direct J4 CC1/CC2 to ASM2464PDX architecture. No external 5.1 kΩ Rd resistors were added because U1 specifies built-in 5.1 kΩ Rd on both CC pins.
Changed TPS259474 ILM resistor R6 from 464 Ω to 1.11 kΩ. The datasheet equation RILM = 3334 / ILIM gives approximately 3.00 A. The previous 464 Ω value was below the recommended 549 Ω minimum and requested more than the device's 6 A setting range.
Updated U3/R6 role notes to state that the 3 A eFuse ceiling is only an absolute cap; it does not enforce lower CC advertisements.
No PCB components were moved or rotated, and no routing was performed.
Rail-by-rail load assumptions
Table
Rail
Load
Typical planning current
Peak planning current
Status
3V3_LOGIC
ASM2464PDX VCCH/VCCA33
7 mA
10 mA
Provisional; PDX datasheet current is TBD
3V3_LOGIC
W25Q64 SPI flash
12 mA
25 mA
Datasheet-supported operating range
3V3_LOGIC
TPS3839 supervisor
0.00015 mA
0.0005 mA
Datasheet-supported
3V3_LOGIC
Load-switch bias and small logic allowance
1 mA
2 mA
Engineering allowance
1V9
ASM2464PDX VCCL
320 mA
455.7 mA
Typical estimate; peak inherited from earlier ASM2464PD data
1V05
ASM2464PDX VDD
1.27 A
1.8123 A
Typical estimate; peak inherited from earlier ASM2464PD data
SSD_3V3
M.2 2280 NVMe SSD
2.0 A
4.0 A transient
Realistic enclosure planning allowance: 6.6 W typical, 13.2 W transient
Margins applied:
3V3_LOGIC: 20%.
ASM 1.9 V and 1.05 V rails: 10%.
SSD peak is already a planning allowance, so no additional margin was added.
Rail totals
Table
Rail
Voltage
Typical current
Peak current
Typical load power
Peak load power
3V3_LOGIC
3.3 V
24 mA
44 mA
0.079 W
0.147 W
SSD_3V3
3.3 V
2.000 A
4.000 A
6.600 W
13.200 W
Combined 3.3 V buck
3.3 V
2.024 A
4.044 A
6.679 W
13.347 W
1V9
1.9 V
0.352 A
0.501 A
0.669 W
0.952 W
1V05
1.05 V
1.397 A
1.994 A
1.467 W
2.093 W
Reflected VBUS budget
Conservative low-input efficiencies used:
3.3 V external buck: 88%.
1.9 V buck: 85%.
1.05 V buck: 80%.
TPS259474 loss modeled separately at its 45 mΩ maximum on-resistance.
Table
Load state
Converter input power after eFuse
VBUS source current at 5.0 V
VBUS source current at 4.75 V
Typical
10.210 W
2.081 A
2.195 A
Peak
18.903 W
3.919 A
4.142 A
Available power by Type-C advertisement
Available converter-side power includes the eFuse drop.
Table
Source case
Available after eFuse at 5.0 V
Available after eFuse at 4.75 V
Verdict
0.9 A
4.464 W
4.239 W
Cannot run the assumed system
1.5 A
7.399 W
7.024 W
Cannot run the assumed typical system
3.0 A
14.595 W
13.845 W
Typical operation possible; peak not supported
5 V-only verdict: only a verified 3 A advertisement can support the assumed typical workload, and even that cannot support the 4 A SSD transient allowance. The design must either negotiate more power through a vendor-approved PD architecture or actively cap/gate SSD and controller operation to the detected source capability.
CC and VBUS architecture review
Historical pre-change review. See “2026-08-27 implemented external-PD architecture” for the live state.
Verified
U1 CC1 and CC2 include built-in 5.1 kΩ Rd resistors.
J4 CC1/CC2 now connect directly to U1 through the CC protection network.
There are no separate CC pull-down resistors to create a conflicting parallel Rd.
There is no second active Type-C/PD controller in the current BOM. U1 is the only CC/PD-capable controller connected to the port.
U1 VBUS is described as a 5 V-tolerant input and is connected to raw connector VBUS.
Unresolved / blocking
The public PDX datasheet does not document how detected Default/1.5 A/3 A source current is exported to system power-control logic.
U3 is a fixed analog current limiter; its ILM pin is not dynamically controlled by U1 in the present schematic.
Therefore the schematic cannot be shown to honor all CC advertisements.
Any higher-voltage PD contract would put the higher VBUS directly on U1's 5 V-tolerant VBUS input. A vendor-approved sense isolation/divider/clamp or different rail topology is required before PD can be implemented.
Component and thermal/headroom checks
J4 USB-C receptacle — pass as a component
Rated current: 5 A.
This exceeds the calculated 4.14 A peak source-current planning case.
Source, cable, CC advertisement, and USB-C contract remain the actual limiting factors.
U3 TPS259474 eFuse — conditional pass at the new 3 A cap
Operating input range and voltage rating cover 5 V and potential 9 V power paths, but U1 VBUS does not.
New R6 = 1.11 kΩ gives approximately 3.00 A.
At 3 A and 45 mΩ, dissipation is about 0.405 W.
Estimated rise is about 17 °C with the datasheet custom-board thermal-via condition or 30 °C using the poorer no-via metric.
It intentionally trips/limits before the calculated full-system peak. This is safe behavior but means the peak workload is unavailable.
U4 LM5143 3.3 V external buck stage — power components mostly adequate, control architecture incomplete
At 5 V in, 3.3 V out, 400 kHz, and 3.3 µH:
Inductor ripple: 0.850 A p-p.
Peak inductor current at a 4 A SSD load: 4.425 A.
10 mΩ shunt gives cycle-by-cycle peak limit of approximately 6.6 A minimum / 7.3 A typical, before ripple interpretation.
Power-stage checks:
Q1/Q2 CSD17578Q5A: 9.3 mΩ maximum at 4.5 V gate drive; adequate current and conduction-loss headroom.
L1 IHLP2525CZER3R3M01: 6 A heat-rating current and 13.5 A saturation current; passes the 4.425 A peak estimate. DCR loss is about 0.48 W at 4 A.
R1 10 mΩ shunt: about 0.16 W at 4 A. Electrical stress is acceptable for a 0.25 W-or-higher part, but the live library metadata is inconsistent: its displayed part name implies a 2512/3 W part while the MPN/package data identify a 0603 low-ohm part. BOM identity must be reconciled before release.
C3 100 µF polymer: ESR up to 55 mΩ and 1.5 A ripple rating at 45 °C; useful bulk storage, but it does not replace the required effective ceramic output capacitance.
C1/C2 10 µF 25 V 0805 input ceramics: voltage rating passes; effective capacitance under DC bias was not available from the component sheet and must be confirmed.
Functional blockers:
U4's 54.9 kΩ / 20.0 kΩ EN divider sets approximately 7.5 V turn-on, so U4 does not start from 5 V.
3V3_LOGIC is not sourced by U4 or another verified regulator.
The compensation network remains first-pass/placeholder and was not validated by a vendor design tool.
U7 TPS22995 SSD load switch — fails the 4 A peak allowance
Maximum continuous current: 3.8 A.
Maximum on-resistance near 3.3 V: 25 mΩ.
At 3.8 A: about 0.361 W and about 51 °C rise using the datasheet 141.6 °C/W metric.
At 4.0 A: the device exceeds its continuous-current rating and would dissipate about 0.400 W with about 57 °C estimated rise.
Required decision: either constrain the supported SSD to a verified peak below 3.8 A, or replace U7 with a load switch/eFuse rated for at least the adopted SSD transient allowance. No replacement was selected because that would lock a power architecture before the source contract is resolved.
U5 LMR33630 1.05 V rail — IC current rating passes; implementation incomplete
IC rating: 3 A, above the 1.994 A planning peak.
The 2.2 µH generic inductor has no current/saturation rating in the schematic. Estimated peak is about 2.13 A at 5 V input, but the inductor must also survive converter current-limit conditions.
BOOT is left on a placeholder single-pin net rather than connected through the required bootstrap capacitor to SW.
FB is connected directly to 1V05; the exact output/tolerance implementation is not demonstrated with a validated divider.
Output effective capacitance and transient performance have not been validated for the nearly 2 A load step.
This rail is not production-ready despite the regulator IC's current rating.
U6 TPS62901-Q1 1.9 V rail — IC rating passes; configuration incomplete
IC rating: 1 A, above the 0.501 A planning peak.
L3 = 2.2 µH is within the datasheet 1–4.7 µH range, but the generic part has no current/saturation rating.
C12 = 22 µF nominal matches the recommended nominal output capacitance, subject to DC-bias effective-capacitance confirmation.
1.9 V VSET requires a 26.1 kΩ, 1% programming resistor and a valid S-CONF mode selection. The current schematic uses placeholder 10 kΩ networks and split duplicate symbol pins; the configuration is not validly demonstrated.
This rail is not production-ready until the exact symbol-pin mapping and VSET/S-CONF wiring are corrected.
Capacitor derating and inrush
U3 dV/dt capacitor C4 = 220 pF gives approximately 9.1 V/ms from the TPS259474 equation. Inrush depends on all downstream capacitance and enabled converters; a complete startup capacitance inventory is still required.
U7 CT capacitor C14 = 10 nF implies a very slow SSD ramp (roughly 13 ms at 3.3 V from the datasheet equation), reducing inrush. The resulting SSD power-up timing must be checked against HDDPC and M.2 requirements.
All ceramic bulk values must be checked as effective capacitance at bias, not nominal capacitance.
Power sequencing blocker
Historical pre-change finding. R33 now sources 3V3_LOGIC from the 3.3 V buck and U5/U6 EN now follow 3V3_LOGIC; the remaining sequencing blockers are the unfinished U5/U6 regulator implementations.
ASM2464PDX requires:
VCCH after VBUS/5 V.
VCCL after VCCH.
VDD after VCCH.
Current schematic behavior:
U5 and U6 EN pins are tied to protected VBUS, so 1.05 V and 1.9 V attempt to start immediately.
3V3_LOGIC/VCCH has no verified source.
U4 is held off at 5 V by its approximately 7.5 V UVLO threshold.
The required power sequence is therefore not implemented. Correcting it requires an ASMedia-approved sequencing and power-mode policy, especially if a higher-voltage PD contract is introduced.
Hard blockers before release
ASM2464PDX current data: official PDX typical/maximum rail currents are TBD; obtain production power data or measured characterization.
Source capability enforcement: demonstrate how Default/900 mA, 1.5 A, and 3 A advertisements limit system/SSD loading.
PD architecture: peak operation needs more than 5 V / 3 A, but no PD controller is present and U1 VBUS cannot be exposed to negotiated 9 V.
3.3 V logic source and sequencing: add a verified VCCH-first rail and sequencing implementation.
U5 implementation: complete BOOT, feedback, inductor rating, and output-capacitance validation.
SSD switch: TPS22995 is limited to 3.8 A; either lower the supported SSD peak or select a higher-rated switch after the source architecture is fixed.
LM5143 validation: finalize compensation, current-limit target, capacitor effective values, startup, and thermal performance with TI design tools or bench validation.
Shunt identity: reconcile R1 part name, MPN, package, power rating, and PCB footprint.
Final status by requested item
2026-08-27 implemented external-PD architecture
The live schematic now uses TPS65987DDKRSHR (U10) as the single CC/USB-PD policy owner. ASM2464PDX CC1/CC2 are explicitly no-connect, while U10 CC1/CC2 connect to J4 through the existing CC ESD network. U10 I2C2 SCL/SDA and IRQ connect to ASM2464PDX I2C_CLK_P, I2C_DATA_P, and GPIO1. A dedicated W25X05CLSNIG TR (U11) stores the TI configuration bundle; production firmware/configuration remains a release blocker.
Supported contract: fixed 9 V / 3 A only for full operation. U10's hardware fallback strap is Configuration 2 (5 V sink) using R19 = 75 kΩ and R20 = 24 kΩ, DIV = 0.242. If the configuration image is missing or 9 V is not negotiated, the LM5143 EN divider keeps the main 3.3 V converter off, preventing an unsupported high load on 5 V.
At the USB-PD minimum 9 V tolerance of 8.55 V, the existing 18.903 W peak converter-input budget requires 2.252 A after including 25 mΩ U10 and 45 mΩ U3 series resistance. This leaves 0.748 A (24.9%) margin to the 3 A contract. Typical load requires 1.206 A. The U3 OVLO divider (768 kΩ / 100 kΩ) trips at 10.268-10.616 V rising, so 15 V and 20 V contracts cannot energize the downstream regulators even under a configuration fault.
VBUS path is now: J4 VBUS_PORT -> U10 internal 20 V/5 A sink path -> VBUS_RAW -> TPS259474 U3 (3.00 A limit, OVLO near 10.4 V) -> VBUS_PROT_9V -> regulators. SMAJ12A-13-F (D12) clamps connector-side VBUS with 12 V standoff and 19.9 V maximum specified clamp. D11 USBLC6-2SC6's VBUS reference was moved from 9 V VBUS to 3V3_LOGIC, because it is a 5 V USB2 protection array.
ASM VBUS sensing is isolated from negotiated VBUS by Q3/Q4 (MMBT3904LT1G) and R29-R32. U1 now sees ASM_VBUS_SENSE_3V3: approximately 0 V with no protected VBUS/VCCH and approximately 3.3 V when both are valid. The U1 pin therefore never sees the 9 V rail. R29 dissipates approximately 0.69 mW at 9 V and less than 1 mW at the U3 OVLO maximum.
Power sequencing was improved by R33 = 0 Ω from BUCK_3V3_PRE_SWITCH to 3V3_LOGIC. U5 and U6 EN now derive from 3V3_LOGIC, so VCCH/3.3 V rises before the 1.05 V and 1.9 V regulators are enabled. Exact U5 BOOT/feedback and U6 VSET/S-CONF completion remain pre-existing regulator blockers.
The datasheet extraction service returned the governing values without immutable inline citation IDs in this run; source verification was against the user-supplied ASM2464PDX PDFs and TI TPS65987D/TPS259474 datasheets. Do not change the 9 V-only configuration or strap values without re-verifying those sources.
Focused ERC / power-check result after edits
J4 CC1, CC2, all four VBUS contacts, and all four signal-ground contacts are now members of the intended nets; their previous floating-pin findings were cleared.
No CC/VBUS driver conflict or new power-pin conflict was reported in the focused review.
Remaining J4 floating pins are the USB2, SBU, USB4 high-speed, and mechanical-mount terminals outside this power-input repair scope.
ERC confirms U5_BOOT_PLACEHOLDER is still a one-pin net, supporting the documented conclusion that the 1.05 V regulator implementation is incomplete.
Additional existing floating status/control pins remain on U3/U4/U5/U6 and require resolution as part of the blocked regulator/sequencing completion, not by adding speculative connections.
Table
Requested item
Status
Enumerate loads and reflect power to VBUS
Completed with documented provisional ASM assumptions
Determine 5 V-only vs PD/current limiting
Completed: 5 V-only cannot support peak; PD or enforced limiting required
Verify CC1/CC2 and no parallel controllers
Completed; J4 repaired, U1 is sole active controller, built-in Rd retained
Verify VBUS sensing limits
Completed: U1 is 5 V-tolerant; higher-voltage raw VBUS is blocked
Verify power-path components
Completed; conditional passes and specific failures/blockers listed
Authoritative schematic repairs
Completed for J4 CC/VBUS/GND and U3 ILM
Full regulator/sequencing repair
Unattempted because ASMedia policy and PD architecture are unresolved
Focused ERC/power checks
Completed after edits; results recorded in task summary
2026-09-28 authoritative design state
Current release blockers
Scope and verification status
Executive verdict
Live schematic changes made
Rail-by-rail load assumptions
Rail totals
Reflected VBUS budget
Available power by Type-C advertisement
CC and VBUS architecture review
Verified
Unresolved / blocking
Component and thermal/headroom checks
J4 USB-C receptacle — pass as a component
U3 TPS259474 eFuse — conditional pass at the new 3 A cap
U4 LM5143 3.3 V external buck stage — power components mostly adequate, control architecture incomplete
U7 TPS22995 SSD load switch — fails the 4 A peak allowance
U5 LMR33630 1.05 V rail — IC current rating passes; implementation incomplete
U6 TPS62901-Q1 1.9 V rail — IC rating passes; configuration incomplete