Reverse-Engineering the Panasonic UJ-240 OPU Interface

[RobertE]

OVERVIEW

I am developing OpenSilicon, an open-source maskless laser photolithography platform built around consumer optical pickup units. The premise is simple: a Blu-ray OPU already contains a 405 nm laser, a high-NA objective, electromagnetic focus and tracking actuators, and a photodiode receiver with servo-grade feedback. That is most of the optical engine of a direct-write lithography tool, mass produced and available for scrap prices. Rather than building a UV projection system from scratch, the project characterizes an existing OPU and takes control of it at the electrical level.

The current hardware target is the Matsushita/Panasonic UJ-240, a slim internal 6x Blu-ray writer. It was selected for its writer-class 405 nm diode, wide surplus availability, and a sled assembly that separates cleanly from the loader. I have multiple donor units, one of which is sacrificial.

The mechanical and optical portions are reasonably tractable at this point. The electrical interface between the UJ-240 electronics and the OPU is now the primary bottleneck, and that is what this post is about.

-------------------------------------------------------------------

THE ARCHITECTURAL WALL

The drive was designed as a tightly integrated consumer appliance, and the internal interfaces are almost entirely undocumented. Electrically it is three boards:

    Host SATA
        |
    Main board: Panasonic MN103SE9KB controller + SPI firmware flash
        |
        |  CN1001, 65-pin FFC
        |
    Sled driver PCB: TI driver IC + discrete FET island
        |
        |  CN1003, 71-pin FPC (flex marked JDJE0246ZA-A)
        |
    OPU: 405 nm LD, monitor PD, PDIC, actuator coils

Every signal the lithography application needs crosses that 71-pin FPC, and none of it is public. No service manual with schematics has surfaced, the flex has no published pinout, and the MN103SE9KB is a custom optical-drive SoC with no documentation.

The largest single obstacle is the driver silicon. The IC on the sled driver board carries TI topmark 1386G4 in a 56-pin HTSSOP PowerPAD, which puts it in the TPIC1386 family. TI publishes no datasheet for any TPIC13xx part. These appear to be OEM devices built to Panasonic's spec, similar to what others have reported for the TPIC1356.TI and Panasonic have both denied my requests for documentation.

The native control path is also closed, so "just talk to the drive" is not an option. The MN103 answers legal MMC/SCSI commands but rejects diagnostic and vendor opcodes, watchdogs any attempt to modify the relevant mode pages, and will not arm the laser or servos without media. I dumped the 2 MiB firmware flash; it is genuine firmware with a compressed code payload, useful for analysis but not a control surface. The firmware, servo logic, driver IC, and OPU form a closed loop, and the only realistic way in is physical.

-------------------------------------------------------------------

WHAT THE APPLICATION ACTUALLY NEEDS

For lithography I do not need a working Blu-ray drive. The end state is external, low-level control of a short list of nets: laser emission and modulation, the focus actuator, the tracking actuator, and photodiode feedback for focus lock. The sled gets mechanically locked and the spindle stays off, so those interfaces only need to be identified well enough to be positively excluded.

The plan is to isolate those nets from the stock driver at the board level and attach an external controller on the OPU side: a hardware-current-limited laser driver behind a physical enable switch, small H-bridges for the coils, and buffered listen-only taps on the photodiode nets. Nothing gets driven while the stock IC can still source current into the same net.

-------------------------------------------------------------------

EMPIRICAL FINDINGS SO FAR

The approach has been measurement before assumption: unpowered continuity, resistance, and diode-mode mapping through an FPC breakout, trace tracing under magnification, and cross-comparison across donor boards. One practical note that cost me some time: continuity beep into the unpowered driver outputs is unreliable because of MOSFET body diodes. Everything is done in ohms and diode mode now.

High-level results, with full tables and confidence tags in the project repo:

Ground partitioning. Three FPC pins measure as hard chassis ground at about an ohm. A separate group of four pins reads a consistent few hundred ohms to chassis; those are not ground, and the signature suggests a bias or APC divider mesh. There is also a large plane on two further pins that I have not yet proven equal to chassis, so no takeover electronics reference it yet.

Actuator inventory. The textbook OPU model of one focus pair plus one tracking pair did not survive contact with the hardware. Resistance-matrix work found five matched pairs on the connector: three in the single-digit-ohm range, which is voice-coil territory, and two around 60 ohms that terminate in H-bridge-shaped pin groups on the TI part. The 60 ohm pairs are almost certainly a stepper for the spherical-aberration collimator, which a fixed-working-distance tool can simply park. Which of the low-ohm pairs is focus and which is tracking is not yet assigned; they behave as differential coil pairs driven by the IC, not as anything referenced to a fixed rail, so the assignment will come from an isolated low-current twitch test rather than from labels. One of the low-ohm pairs currently has no discoverable connection to either the driver IC or the 65-pin FFC, which is an open puzzle. Two further candidate bridge pairs on the connector have not had their resistance measured yet; that single measurement could reshuffle the whole assignment.

Laser path. The leading LD feed candidate is a pair of paralleled conductors that route through a discrete FET island next to the driver IC and emerge paralleled again on the 65-pin side. Doubled copper on both sides of a switching element is a current-rail signature. What I cannot yet distinguish is whether that island switches a supply rail feeding a laser driver mounted on the OPU itself, or whether the island is the drive stage and the diode is wired straight out. The two models have opposite isolation cut points, so nothing gets cut until a diode-mode sweep and a photographic survey of the OPU near the diode settle it. I am deliberately mapping the monitor photodiode and APC arrangement before connecting anything, because driving a writer-class 405 nm diode without understanding the existing power-control loop is how you get an uncontrolled current condition and a dead pickup.

Signal partition. Roughly half of the 71 pins bypass the driver IC entirely and run straight through to the MN103 board. That pool almost certainly holds the PDIC quadrants, RF, front-end rails, and thermistor. A couple of pins route to both the TI part and the MCU, which suggests shared or sensed nets. Powered rail mapping of this pool, with the laser disabled, is the next major session.

-------------------------------------------------------------------

WHAT I AM ASKING FROM THE COMMUNITY

If you have worked on Panasonic/Matsushita slim Blu-ray drives (the UJ-2xx family), TPIC13xx-era TI optical-drive drivers, or a comparable OPU takeover, I would like to compare notes. In rough priority order:

1. TPIC1386 or any TPIC13xx documentation in any form: datasheet, application schematic, service manual, board photo with traced nets, or a confirmed pinout from a different product using a 13xx-marked device. Knowing whether the part exposes a serial control port versus plain logic/PWM inputs would by itself decide whether the stock IC can be commanded directly or has to be cut out.

2. UJ-series OPU FPC pinouts, full or partial. Panasonic reuses OPU and flex designs across the UJ-230/232/235/240/260 line, so a mapping from any sibling drive is worth diffing against mine. The flex here is marked JDJE0246ZA-A.

3. Laser drive parameters for writer-class BD pickups: threshold and safe operating currents, typical forward voltage at operating point, monitor-PD sensitivity, and anything on the APC implementation.

4. Sled and spindle driver mapping on these boards, mainly so I can positively exclude those nets from the OPU work.

5. Anything on the MN103-series servo architecture, or prior work bypassing the drive MCU on Panasonic optical hardware.

6. Critique. If you see a flaw in the isolation sequencing, the ground-partition logic, or the laser-path reasoning above, I would much rather hear it now than after I have made cuts.

Empirical data beats labels. A report like "pin X idles at 0 V and shows 18 kHz PWM at 3.3 V during seek" is worth far more to this effort than "pin X is servo." I am happy to get into specific pin numbers, resistance tables, and scope captures in the thread; the full measurement log is versioned in the project repository, every assignment carries a confidence tag, and measured data overrides any inferred claim on conflict.

The goal is not to rebuild a Blu-ray drive. It is to establish the minimum electrical interface that turns a proprietary consumer pickup into a characterized, externally controlled 405 nm positioning and patterning subsystem, and to document the whole path openly so nobody has to do this twice.