Sony PlayStation 2 Repair Reference

This page covers the Sony PlayStation 2: every Fat from the launch SCPH-10000 to the SCPH-55000 (boards GH-001 to GH-029), and every Slim from the SCPH-70000 to the SCPH-90000 (GH-032 to GH-072), with the PSX DVR as a delta at the end.

Two things before you read further, because both change how you should weight what follows.

First: I have not bench-tested a PS2 console. What I have built and measured is a USB-C power supply for the Slim, run into a resistive dummy load on my bench. Everything about the console itself on this page comes from Sony’s service manuals, Sony’s instruction manuals, the community wikis and published board scans. Where I would normally say “I see this constantly,” I have said “the sources report” instead, and you should read those as the community’s frequency claims rather than mine. The only data entry tagged bench is my supply’s dummy-load measurement, and it is not a measurement of a console.

Second: the documentation is unusually good for part of this console and does not exist at all for the rest. Sony’s service manuals are real schematics, with net names and IC pin numbers, and the public set covers GH-004 through GH-026 and then GH-035. Then it stops. Nothing exists for the Japanese launch boards (GH-001, GH-003), for GH-008, GH-016, GH-018 or GH-029, or for any Slim after GH-035. For every SCPH-75000, 77000, 79000 and 90000 the evidence base is community board scans and psdevwiki. And GH-035 has exactly one manual, so every Slim pin table on this page is tagged single-source even though the document is Sony’s own.

The organizing idea for this machine: identify the board by its GH code, not by the SCPH number. The board code predicts the pickup, the power supply, the mod wiring, the softmod ceiling and the fault set. One SCPH number routinely ships more than one board. SCPH-700xx alone ships GH-032 (separate EE and GS chips) and GH-035 (combined), and those two need different shields and heat sinks. A pin table is valid for the board it was drawn from and nothing else, and this console breaks that assumption repeatedly.

Safety first

The Fat power supply is a live mains switcher. Probing it plugged in is a lethal-shock risk, and the unplug, replace, re-plug loop is the correct default. The SCPH-90000 went back to an internal supply too, so a dead 90000 carries the same risk as a Fat.

Never put a 12 V supply into an 8.5 V Fat. The Japanese launch boards (GH-001 and GH-003: the SCPH-10000, SCPH-15000 and the GH-003 build of the SCPH-18000) take 8.5 V from their internal supply. Every Fat from GH-004 on takes 12 V. That boundary is settled on three kinds of source: psdevwiki, the GH-004 manual’s own block diagram, and an electronic-load test of real supply boards. The trap is the swap. psdevwiki warns in bold that the AB-chassis (GH-008) 12 V supply is mechanically identical to the 8.5 V unit, with no keying. That claim has only one source, but it is cheap to respect and expensive to test, so check the board code before you fit any A-shaped supply. The later B/B’/C/C’ supplies have an offset connector and are not drop-ins. Note that the SCPH-18000 ships on both GH-003 (8.5 V) and GH-008 (12 V), so the model number does not tell you which one you have.

The Slim’s barrel is shared with consoles at other voltages, and all of them are center-positive. The PS2 Slim takes 8.5 V on a 4.8 x 1.7 mm EIAJ-03 barrel. The PSone uses the same barrel at 7.5 V. The Genesis 2, 32X, Nomad, CDX and US Game Gear use it at 10 V. Nothing mechanical or electrical stops any of them seating in any other:

The printed voltage is the only guard that exists. Label anything on this barrel by its voltage.

Power the console fully off before you insert or remove a Fat Network Adapter. The expansion connector is fine-pitch and shorts easily, and the documented outcome is blown internal SMD fuses.

Know your board first

From the outside

The last digit of the SCPH number is the region: 0 Japan, 1 US and Canada, 2 Australia and New Zealand, 3 and 4 UK and Europe, 5 to 7 Asia, 8 Russia. An R suffix on a Fat (SCPH-3000xR) means DVD Player 2.10 in ROM. GT, SS, AQ and similar suffixes are bundle and color SKUs, not regions.

With the lid off

• Read the silkscreen. The GH code and sub-revision are printed on the board. On a Slim the sub-revision decides whether Sony’s MechaCon-crash fix is fitted (below), and no software tool can read it. PS2Ident identifies the silicon, not the sub-revision.
• Count the large BGAs under the shield. Two means separate EE and GS (every Fat, plus GH-032, GH-037, GH-041, GH-061, GH-062). One means combined (GH-035, GH-036, GH-040, GH-051, GH-052, GH-070, GH-071, GH-072). This alternates across the Slim run; later Slims did not simply combine them.
• A separate daughterboard with regulators, SysCon and the coin cell means B, B’, C or C’ (GH-004 to GH-007 with GD-011, GD-012 or GD-013). The daughterboards are not interchangeable: GD-011 gives 1.5 V to both EE and GS, GD-012 gives 1.5 V to the EE and 2.5 V to the GS.
• Battery: a holder on the board means GH-015 to GH-029 or GH-051 on. A 2-pin battery cable means GH-032 to GH-041.
• Pickup ribbon: 30-pin on every Fat with a manual, 24-pin on the Slim. The one reported 24-pin Fat is GH-029.
• Lid safety switches: two on the K, L and M chassis and GH-061, one (on the power and LED board) on GH-062 and every P/R board. Count them before you defeat one for a bench test.

The boundaries that change a repair

The Slim DC input

This is the part of the console my own work plugs into, and the spec is settled against multiple Sony documents.

• Voltage: 8.5 V. Sony prints DC IN 8.5V on the console and in four separate documents. It is not 12 V and not 7.5 V.
• Polarity: center positive, sleeve negative. Read off the IEC polarity symbol beside DC IN 8.5V in three Sony instruction manuals (the Japanese SCPH-70000 and the English SCPH-70001 and SCPH-79001).
• Barrel: 4.8 x 1.7 mm, EIAJ-03. No Sony document states the geometry. It comes from a customer measurement, confirmed against a physical adaptor, and it is the plug size printed on my own supply’s case. The widely repeated 5.5 x 2.1 mm is wrong.
• Current, and it depends on the model. Sony’s spec table gives an SCPH-700xx console draw of about 5.3 A. The SCPH-70100 brick (SCPH-700xx, 750xx and 770xx) is rated 8.5 V at 5.65 A, about 48 W. The SCPH-79100 brick for the SCPH-790xx is 8.5 V at 4.5 A, about 38 W, and that console’s own draw is 4.5 A. A supply rated 5.65 A covers every Slim with a DC jack. One rated 4.5 A does not cover the 700xx to 770xx.
• The SCPH-900xx has no DC input at all. It went back to an internal supply with a mains inlet, and no Slim DC supply of any kind covers it.

What is behind the jack

On GH-035 (the one Slim with a manual), the jack is J4000, Sony part 9-886-100-44, and the schematic draws it as a three-terminal switched barrel jack: one supply conductor, and a switch pair tied together to ground. That is why “fit a generic 5.5 x 2.1 mm jack” was always the wrong framing for a replacement. The three-terminal construction is single-source, since there is only the one manual. The supply pin runs through two jumper links, JL4003 and JL4004, into F0001, a 6.3 A fuse, and onto the 8.5 V rail.

Two consequences. The fuse is 6.3 A against a 5.65 A brick, so the board’s own fuse will not protect the console from a 6 A supply: the supply’s current limit is the real protection. And Sony’s part number for the jack is a console-specific service assembly number with no distributor cross-reference, so a replacement means a donor or a reflow of the original.

The 5 V rail on GH-035 comes from IC8207, a JCP78M05CDTRKG, which is a 78M05-family linear regulator dropping 8.5 V to 5 V. A linear regulator burns the difference as heat, so a little under 8.5 V at the input is kinder to a twenty-year-old console than a little over. The 3.5 V and 1.8 V MechaCon standby rails are live the instant the barrel seats: the switch-board lead carries an always-on EVER_+8.5V net, and EVER_ is Sony’s prefix for rails that stay up with the console off.

A dead Slim, in order

1. Meter the brick under load, not open circuit. The sources report the brick as a more common failure than the jack.
2. Check whether it is a recalled brick. Slim supplies with date codes 2004.08 through 2004.12 and serial numbers beginning F3 were recalled for overheating and melting. psdevwiki names the Hipro HP-AT048H03 and explicitly not the HP-AT048H03B.
3. Reflow J4000. It is a through-hole jack taking the cable’s mechanical load.
4. Check the two jumper links and F0001. If F0001 is open, find the short before you replace it.
5. Check standby at the switch-board connector. CN4561 pin 1 carries EVER_+8.5V with the console off. If it is there, the brick, the jack, the fuse and the standby path are all good and the fault is downstream.

A console that is dead at the button with its standby rails present is not necessarily a power fault. MechaCon comes up on the standby rails, watches the power switch, and then enables the main regulators. If the standby rails are up and the main rails never follow, look at MechaCon, its EEPROM and reset, and the PWR_SW and OPEN_DETECT lines from the switch board.

One more on GH-035 that is single-source and worth knowing: a Slim that boots to the browser but whose drive is completely inert (no spin, no seek, no eject attempt) should have PS0013, a 400 mA micro-fuse, and the output of IC7001 (a second 78M05 feeding the drive) checked before the drive is condemned. The designators will be different on GH-051 and later, for which no manual exists.

psdevwiki observes that as of 2025 Slim supplies, external and internal, are usually still fine on their electrolytics. On a Slim, do not lead with a recap.

Running a Slim from USB-C

I build a USB-C power supply for the Slim, the CA-DPS2U1. It takes a fixed 20 V USB-C PD contract and makes a regulated 8.4 V at up to 6 A, on the 4.8 x 1.7 mm center-positive plug, for the SCPH-70000, 75000, 77000 and 79000 series. It does not fit the SCPH-90000 or any Fat, neither of which has a DC jack.

These numbers are the supply measured on its own, into a resistive dummy load on my bench, not in a console: at 5.7 A it holds 8.42 V at the converter and 8.10 V at the plug, at 93.8% efficiency sealed in its case, and it held that for a 49-minute soak with the hottest part inside at 81 C. The 8.4 V setpoint is deliberate, for the linear-regulator reason above. The factory load line quoted on the product page (8.45 V open circuit, 8.2 V at 5 A) comes from a load test of Sony’s 8.5 V Fat supply boards (the SCPH-10000 / 15000 unit), not from a Slim brick.

Common problems and fixes

Ordered by how often the sources report each one. That is a working order, not measured data: nobody has published PS2 failure rates. The one hard datum nearby is that Sony settled a US class action over PS2 disc read errors on 2005-11-10, which establishes that the drive fault family was big enough to be legally material.

Disc read error, browser reachable

The sources report this as the dominant fault family, and the same “disc could not be read” screen is the terminal display for at least six different causes. The console’s own error text distinguishes none of them. Treat it as one diagnostic path and run it cheapest first.

Step 0: find out what “cannot read” means. Four discs, in this order: a pressed PS2 DVD, a pressed PS2 CD-based title, a PS1 disc, and a DVD movie. The pattern of what reads and what fails is the diagnosis (see the CD/DVD split below). If nothing spins at all, it is not a read fault.

Step 1, on a Fat: suspect the power supply. psdevwiki states that failing Fat PSU electrolytics can cause read errors in an otherwise intact drive. That link to read faults is single-source (the no-power and reset symptoms are independently corroborated, the read-error half is not), so treat the supply as a cheap co-suspect to exclude, not an established mechanism. The tells are intermittency that tracks warm-up, resets, or periodic clicking from the supply.

Step 2: watch a full read attempt with the lid switch defeated. Count the lid switches first. A disc that spins up and stops with no sled motion points at the sled drive or a pickup that never focuses. A sled that steps outward in jerks or stalls at one radius points at the rails or the rack. Repeated spin-up hunting points at focus and tracking, which means the lens or the diode. And listen: psdevwiki records that the drive emits a 2 kHz beep during disc initialization while the servos set their gains, twice on a dual-layer DVD. No beep means the console never got as far as servo calibration.

Step 3: clean the right things. The lens with a swab and isopropyl, no pressure. Then the rails and the sled track. The sources report dried or contaminated grease as a far more common cause of a mid-radius stall than a dead diode. Sony’s own D-chassis drawings show factory grease (Froil SG474C, Sony 7-651-000-22) at the sub-chassis and sled motor. Re-grease lightly with a plastic-safe synthetic.

Step 4: mechanics. Rack and worm teeth, tray belt, spindle grip, and the flex cable. The pickup flex flexes on every seek. Sony’s own drawing instructs you to lock the connector stopper so the ribbon cannot pull out, and a published bench report records a repair that regressed because the ribbon loosened afterwards.

Step 5: tilt and skew, only after the mechanism is clean and moving freely, because a sticky sled produces exactly the radius-dependent failure that tempts you to re-tilt. The adjusters are named Sony parts (the skew screw is 3-060-399-01 on the D, G, H and I chassis), and H/I is a two-axis adjustment where D, F and G are not. There is no OEM tilt specification in any manual, so this is empirical against a known-good disc. Mark every adjuster before you move it, and move in the smallest step that does anything.

Step 6: replace the pickup, with the ordering gates in the next section.

Step 7: the potentiometer, last, and see the next heading before you touch it.

The laser pots do not set laser power

There are two 2 k potentiometers on the pickup, one for CD and one for DVD, and they do not adjust laser output. Output power is regulated in real time by the drive’s automatic power control. The pots set the level of the monitor photodiode that the APC loop regulates against. So turning one overdrives the diode past its specified output. It buys tolerance to dirt in the optics or to mis-set mechanics, and it pays for that with the rest of the diode’s life. It does not restore a worn diode, because the wear was already being compensated. That mechanism is standard optical-pickup practice and is corroborated outside psdevwiki, and Sony’s own drive drawing prints “Don’t touch the adjustment resistor.” next to the pickup.

The circulating recipe (“Sony sets the DVD pot to 1100 ohms, never go under 800”) has no primary support, and its own figures do not agree with each other: the 2003 write-up it traces back to gives CD 1220 and DVD 850 and says values vary by pickup model. A console that reads only because its pot was turned will come back. If a pot has been turned, price the console as needing a pickup.

The wipers are brought back to the mainboard, so you can see where a pot sits without probing the pickup: CN801 pins 23 (CD) and 22 (DVD) on a Fat, CN7301 pins 12 (CD) and 13 (DVD) on GH-035. Tables below.

Reads CD-based discs but not DVDs, or the reverse

This split is diagnostic because the pickup carries two laser diodes on two separate drive pins: CD-LD on pin 11 and DVD-LD on pin 19 of the Fat’s CN801, CD_LD on pin 8 and DVD_LD on pin 10 of the Slim’s CN7301.

• PS1 and PS2 CD titles read, DVDs do not: the DVD emitter or its drive path. The CD reads show the rails, sled, spindle and tilt working, so those are largely exonerated.
• DVDs read, CD titles do not: the CD side. It often goes unnoticed until a PS1 disc is tried, because CD-based PS2 titles are a minority.
• Single-layer DVDs read, dual-layer fail: focus and servo margin, not the diode. The layer change is where a marginal focus loop fails first.

Ring out the select line before you order a pickup. Both connectors carry a CD/DVD select signal: CN801 pin 9 on the Fat and CN7301 pin 21 on GH-035. A select line stuck at one state, or open in a cracked ribbon, produces exactly this split at zero cost in diode life. On a Fat the emitter and detector are one package, so “the DVD diode is dead” and “replace the pickup” are the same sentence.

On an SCPH-10000 or 15000, “DVD Player is not set up” is missing software: those consoles load the DVD player from a memory card.

Worked, then stopped reading anything at all: the MechaCon crash

A console that read perfectly and is now dead to every disc, with a reset button that does nothing until you pull power, has a distinctive cause. The CXD3098Q drive DSP crashes MechaCon, on a badly readable disc or just on bad luck, and the drive keeps driving the focus and tracking coils until they burn. It often takes the driver IC with them, and ConsoleMods records it reaching the laser diode as well. Measure the coil resistances against a known-good pickup: no source publishes a coil specification, so this is a comparison, not a spec check.

Where it bites, per the sources. The CXD3098Q arrived at GH-017 and runs through GH-061. psdevwiki reports the G-chassis MechaCon as less prone to crash outright, so the problem becomes prevalent from the H-chassis on. ConsoleMods dates first susceptibility to the SCPH-390xx and says the rate rose considerably on the SCPH-500xx and again on the SCPH-700xx. It also calls the SCPH-390xx’s “most reliable Fat” reputation a myth, attributed to modchip installers liking it because it is the easiest to chip. SCPH-750xx and later are fixed from the factory. The PSX DVR does not crash.

Sony’s fix on the SCPH-700xx is per sub-revision, and it is readable in the silkscreen:

In software, MechaCon version 6.04 or higher has the fix; 6.00 and 6.02 do not. That is buy-or-avoid information readable before you open anything. What the factory fix actually does is unresolved across three sources: psdevwiki says it mutes the motor driver through its MUTE pin (pin 42 on the Slim’s RS2004FS-E2, OEM-confirmed), ConsoleMods’ model page says added shutdown circuitry, and ConsoleMods’ crash page says new DSP firmware.

The sources’ refurb policy follows: on any GH-023 or GH-026, and on any GH-032 or GH-035 sub-revision without the fix, the pickup you just fitted is at risk from day one. Fit the PIC fix (in the mods section) or price the console accordingly. The other passive fixes (the “Romeo” or “LA” fix and summ0ne’s fix) are reported not to reliably prevent the damage, with side effects including reading issues and a slow tray.

Slim only: discs come out with a new concentric scratch

The orange drive ribbon flexes up into the spinning disc. The SCPH-790xx is the only Slim with factory plastic holders that stop it, and they were removed again on the SCPH-900xx. On every other Slim, open the lid and look at where the ribbon rests relative to the disc plane before you spin anything you care about. The fix is retaining or re-routing the ribbon, and a creased one is replaced (GH-035: 1-864-648-11 / -21).

Fat: dead, ticking, resetting

psdevwiki: most, if not all, Fat PS2 power supply models have known issues with leaking or aging electrolytics, and every manufacturer is affected. The presentations it lists: completely dead with no standby LED; standby LED on until you press power; a periodic ticking from the supply as it shuts down and restarts; random resets mid-game.

Where to look. On the Delta boards, the small capacitor next to the heat sink, with no visible bulge. Three independent bench accounts land on C3 (33 µF 35 V): psdevwiki’s worked case where a leaking C3 caused instant shutdown and clicking, a badcaps.net Delta that read fine on capacitance and bad on ESR, and a nicau.ro teardown that measured C3 at 35 ohms ESR. So pull C3 first and test ESR out of circuit. ConsoleMods also names C20 (22 µF 50 V), but no published bench account confirms it, and the nicau.ro teardown measured C20 good. ConsoleMods’ sentence never says “Delta” either: the C3 / C20 scheme matches only the Delta boards, and the Nichicon and Panasonic supplies use different three-digit designators with no published “these two go first.” Read the sticker on the PSU board, not the console model number. One SCPH number ships several supply makers.

ConsoleMods’ order on a dead Fat: blackened parts, then the mains cord, then the switch, then the fuse, then visible cap damage, and if all of that is clean, the small caps above. A blown fuse means something else failed first. The mains inlet and power rocker are one assembly (on GH-023, 1-786-345-51), so a damaged inlet is a switch-assembly swap. On a Fat there is also no fuse in the 12 V path on the mainboard, which is why a downstream short shows up as a supply that ticks or shuts down.

Runs, then shuts down with a blinking red standby LED

Thermal shutdown: fan dead, disconnected or clogged, or dried thermal interface. Check in order: does the fan spin at power-up, is the connector seated, are the intake and exhaust clear, is the interface intact.

Sony ships die-cut thermal sheets, not paste, with separate part numbers for the EE, the GS and the driver IC on the same board. Measure the original thickness before you peel it and replace with a gap pad of that thickness. On GH-071 the heatsink is gone and the EMI shield contacts the SoC directly, so its interface material is not optional.

Know how your fan is driven before you scope it. On a Fat, CN403 sits at a smoothly variable DC voltage (roughly 4.5 V at nominal speed on the G-chassis sheet), because the fan driver is linear. No square wave there is normal. On the Slim, CN4562 pin 3 really is PWM.

Powers on, fan spins, no picture

The sources report a bad AV cable or port far more often than a board fault. Try a known cable and a second display first. The AV Multi Out carries composite, S-Video, RGB and YPbPr on one connector, and RGB and component on this console are cable purchases, so “no picture” is often a cable or TV-input mismatch.

Two software causes look like hardware: component or progressive output left on with a standard-definition display attached (ConsoleMods keeps a page just for undoing it), and bad GSM settings on a softmodded console, which can freeze at the PlayStation logo.

At board level on GH-035, each video output leaves the encoder through a 75 ohm resistor and an electrolytic coupling cap. One missing color with the other two correct is that resistor or that cap, not the GS.

Tray will not open, opens slowly, or grinds (Fat)

A perished or glazed tray belt, which degrades in storage rather than in use, so a low-hours console is not exempt. Or a stripped driving gear, or a dry cam. Pull the gear cover and look before you order a gear, because a slipping belt sounds identical. On H and I chassis the belt (3-065-505-02) and driving gear (3-084-526-01) are orderable Sony parts. On the D and G chassis neither is sold separately; the answer there is the tray assembly or a donor drive.

One controller port or card slot dead

A single dead slot with the other working is a connector or local supply fault, not the IOP, which serves both. Check continuity from the port pins to the board, then look for an open SMD fuse on that port’s rail, and find what shorted before you replace it.

On a Fat the front terminal board is an orderable assembly, and it lands on a 36-pin CN302 that is wired identically on GH-022 and GH-026, so a front board carries across G and I. On GH-035 the ports come off a 34-pin CN3001, and it splits the symptom: a jack that is completely dead points at that port’s 3.5 V pin, while a controller that works but will not rumble points at the 8 V pin (7 or 24). On a Fat, a controller that browns out under rumble is the 8 V rumble rail, not the SIO2 interface.

Clock resets every cold boot

The CR2032. Sony’s own Level-1 manual gives it 5 to 8 years, so on any PS2 today a dead one is expected, not a fault. On Dragon boards the region and configuration live in EEPROM, so a dead cell costs you the date and time and nothing else.

Faults that are software

These all arrive described as drive faults:

• The red screen with cubes, “Please insert a PlayStation or PlayStation 2 format disc”: a wrong-region or unsupported disc, not a read failure.
• “DVD Player is not set up”: FMCB 1.95 specifically, an unpatched burned disc, or an SCPH-10000 / 15000 that needs its DVD player installed to a card.
• Reads originals and refuses burned DVD-Rs: media or burn quality. On a marginal drive that is also the exact condition that triggers the MechaCon crash.
• PS1 games behave differently after a Slim swap: GH-036 (SCPH-750xx) and later have no hardware PS1 IOP.
• A softmod that will not autoboot after a board swap: the FMCB entry point is patched out on GH-071-42 and GH-072.

Stuck disc

Sony’s procedure: if power and the loading mechanism are alive, power it on and press the front button while the red LED blinks, then eject. If not, the Level-1 manual’s disassembly sequence reaches the four MD-block screws, and the disc comes out once the pulley bracket assembly is lifted. Mind the screw count: 10 bottom screws and bottom-first on the SCPH-10000 / 15000 / 18000, 8 screws and top-first on everything later.

Ordering a pickup: the board-revision gates

Ordering the wrong pickup is a board-revision error, not a drive error. Two independent gates decide whether a swap works, and a pickup has to pass both: it must physically fit the chassis and mate with its cable, and it must match the calibration data MechaCon holds.

“KHS-400B” and “KHS-400C” are not single part numbers. Sony prints several under each name, and which board a number is listed for matters as much as the number:

• KHS-400B: 8-820-141-01 in the B/C-design manual, -10 for the D/D’ chassis, -29 for GH-013 / GH-014. The -01 has a short base tab that cannot reach the relocated home switch on a D/D’ drive. For D/D’, order -10 or later, which is also back-compatible into the earlier drives.
• KHS-400C: 8-820-151-01 on the F-chassis only, -15 / -16 / -17 / -18 on G, -18 on H, and -19 through -33 on I. A -18 ordered for a GH-026 is the wrong sub-model.
• Sanyo SF-HD7: 1-758-717-xx on F and G, 1-758-939-xx on I. It is the same form factor as the KHS-400C and fits the same chassis. Sony’s own GH-023 list names no Sanyo pickup at all, which is a gap in Sony’s list, not proof that one never shipped. The community name “KHS-400R” is just the SF-HD7.

The mechanical gate: KHS-400B versus KHS-400C. psdevwiki says they are electrically compatible but differ in base thickness, and that a 400C fitted to an AB, B, B’, C or C’ drive “will seem to work at first” and then bends the flex badly once the pickup reaches the outer area. That claim is single-source and actively contested: two parts sellers say a 400C replaces a 400B in all cases, and the two independent forum reports put the failing direction the other way round. The only part two independent sources agree on is that the 400B is the taller pickup. Nobody has published a measurement or a bent-flex failure. Work to the conservative rule, a 400B for those drives, and do not present a working swap as proof.

The calibration gate: lens type. The T487 (white ring) and T609K (yellow ring) lenses need different MechaCon data. From the G-chassis on, only T609K is supported, and that half is corroborated in PMAP’s own code. Up to early C only T487 is supported, and D and F support both once configured (psdevwiki only). A yellow ring means T609K. A white ring does not reliably mean T487, because some late T609K units ship white; a violet mark at the skew screw, or “609” hand-written on the base, also means T609K, and marks in other colors mean nothing. Treat white as indeterminate and read the type out of MechaCon.

The clean example of a calibration-only gate is the SF-HD7: mechanically identical to the KHS-400C, and it will not read correctly until MechaCon is told what it is holding. Sony says so in its own GH-015 parts list: “EEPROM DATA have to rewrite BASED ON replacing OPTICAL DEVICE.” A replacement that reads worse than the old part is usually this, not a counterfeit.

Two tools, two different rigs. PMAP runs on a PC, a clone of Sony’s own adjustment tool, over RS232C to the Fat mainboard test points (TXD JL610, RXD JL611, 3.3 V JL613, ground JL609, and JL617 to ground for test mode). LensChanger runs on the console itself from a memory card, writes the console’s own EEPROM from the EE side, and needs no PC and no serial link. It does not support the SCPH-90000.

Remove the shorting solder. A new Fat pickup ships with a static-protection solder short across the laser terminals, and Sony’s drawing says to remove it after fitting. A drive that will not work after a pickup swap may just still have it.

On a Slim, order the traverse, not the pickup. Sony sells the Slim drive as a traverse assembly matched to the drive vendor, 1-788-193-11 Mitsumi or 1-788-194-11 Sankyo on GH-035, and names no pickup part number anywhere. The two are not interchangeable: the Sankyo mechanism takes only an SPU3170 and has a laser arm nothing else uses, and even the disc-guard screw count differs (two for Sankyo, one for Mitsumi). KHM-430, PVR-802W, KHS-430 and TDP-182W are four names for the non-Sankyo pickup. Replacement Slim pickups usually arrive with the wrong arm length, so transfer the original arm, and the drive cable comes in four lengths by chassis.

The J-chassis GH-029 is reported to use a Mitsumi PVR-702W on a 24-pin cable, and no KHS-400 fits it. That rests on psdevwiki alone, since no manual covers the board, but a GH-029 in stock is a parts-ordering trap either way.

Inside the family: what differs

Dragon and Deckard are two different chips in two different generations. Dragon is the ARM-based MechaCon from GH-023, and it absorbs SysCon, the EEPROM and the RTC. Sony’s own parts lists prove it: GH-022 carries a separate TMP87C807U SysCon at IC402, and GH-023 carries the Dragon alone. On those boards a “SysCon fault” is a MechaCon fault. Deckard is the PowerPC 440 IOP from GH-036 (SCPH-750xx), which emulates the old MIPS IOP with on-die help and absorbs the network and HDD controller. Deckard is the end of hardware PS1 compatibility and of the internal HDD path.

Connectors keep their designator and change underneath it. Every one of these is from Sony’s own sheets:

• The front-panel CN406 is 7-pin on GH-017 / GH-022 and 12-pin on GH-026, with the power-switch pin moved, the tray-LED polarity flipped, and pin 4 changed from ground to a driven LED line. A 50000-series switch board is not a donor for a 39000-series mainboard. Some GH-015 sub-revisions also need a switch board with a reversed connector; twisting the cable does not fix it.
• The sled connector is wired in exactly opposite order on the Fat (CN803) and the Slim (CN7302). Carry one onto the other and both stepper phases invert. Sony’s parts list calls the sled part a DC motor while the schematic drives it as a two-phase stepper, so a two-wire bench test tells you nothing.
• The Fat’s 13-pin CN804 carries the whole spindle motor and the inner limit switch. The Slim moves the limit switch onto the pickup ribbon. So a homing fault with no read attempt points at the pickup ribbon on a Slim and at CN804 (with spindle trouble alongside) on a Fat.
• Switch-board ribbons are numbered in opposite directions at their two ends on both form factors, so a pin-1-to-pin-1 continuity check reads a good cable as bad.
• Heatsinks and shields are revision-specific: GH-032 and GH-035 do not share one, GH-036 / GH-040 differ from GH-037 / GH-041, and GH-071 and GH-072 take different shields.

Rails change on the boards with no manual. A community trace of an SCPH-79003 (GH-061) shows a 1.25 V core rail and no 3.1 V, 1.7 V or 1.2 V rail where GH-035 has them. Do not carry GH-035 rail voltages onto an N-chassis board. That trace is one person’s measurements on one board.

The PSX DVR (DESR, XPD-001 and XPD-005 boards) shares the EE, GS and IOP lineage and the Dragon MechaCon, and has twice the EE RAM, four times the IOP RAM, and two optical front ends. Every disc is identified first by the DVR side pickup (KWS-200A or SF-DB11), and the PS2 pickup only becomes active after that, so a dead DVR pickup stops PS2 games too. Do not read a CD/DVD split off a PSX. It does not suffer the MechaCon crash and needs no PIC fix, and its HDD past the first 40 GB is encrypted behind the DVRP chip.

Mods worth knowing

This is an orientation to what is possible on which board, plus the traps that cost money on a resale build. The install guides are linked at the bottom.

The compatibility gate

• FreeMcBoot autoboot dies at boot ROM 2.30 (late SCPH-900xx, R-chassis GH-071-42 and GH-072-42, datecode 8C and later) and 2.50 (the Bravia KDL-22PX300 PS2 TV). Sony removed both update paths from that ROM, so there will never be a software fix. Datecode 8C is ambiguous, so dump ROMVER with PS2Ident before you quote a softmod on any SCPH-900xx. A patched unit can still be softmodded through FreeDVDBoot plus FunTuna, which is a longer job.
• MechaPwn needs the Dragon MechaCon, so SCPH-50000 and later only. Everything earlier needs a modchip for import and backup discs.
• Native HDD needs the expansion bay: every bay-equipped Fat, and the SCPH-700xx only among Slims (restorable, no connector fitted). FreeHDBoot is officially supported through the SCPH-700xx.
• No hardware ODE exists for the PS2 as of 2026-09. Anything sold as a “PS2 ODE” is mislabeled. The functional equivalent is OPL or Neutrino over the network (UDPFS) or an internal HDD, and the console still needs a working laser to read an exploit disc unless it is softmodded from a memory card.
• FunTuna and Fortuna cards are Slim-only. Using one on a Fat can damage the savedata on that card. Delete the folder before the card goes near a Fat.
• A modchip is the leading cause of FMCB failing on an otherwise compatible console, per the FMCB maintainer’s FAQ. It shows up as a freeze at the orb animation or a black screen on every app. Try an older FMCB build or PS2BBL.

The Matrix PIC fix

A protective mod for the MechaCon crash. A PIC12-series microcontroller in SOIC plus one resistor watches for the crash and powers the console off before the coils burn. Published solder points exist only for GH-023, GH-026 and GH-032 / GH-035; on anything earlier it is a reverse-engineering job, not a procedure.

• The resistor is per revision: 2.2 k on GH-023, 1.5 k on GH-026 and on GH-032 / GH-035, read off the wiring diagrams. The Matrix notes give the bounds: below 1 k the console sits in permanent reset, above 3.3 k the PIC cannot power it off.
• Buy the flash PIC12F508, not the one-time-programmable 12C508 of the 2004 original. A mis-programmed OTP chip is scrap.
• Bench-test it before the final install: wire pin 4 to ground instead of its point, and the console should power on at reset and turn itself off after three seconds.
• The “diode fix” is insufficient, not useless. The drivers still saturate if a crashed console is left on, so never rely on it alone, but the Matrix Team themselves recommend it alongside the PIC fix for lens life.

A MechaPwn’d Fat or SCPH-700xx is at higher risk of the coil burn, because the point of MechaPwn is to boot burned discs and bad data on a burned disc is the trigger. Pair the two or do not do it. And back up each console’s EEPROM before MechaPwn: a botched restore can land another factory’s laser calibration on the console and present as a drive fault.

HDMI

The only digital HDMI kits found are Hispeedido’s commercial boards. The Fat kit lists GH-019, GH-022, GH-023 and GH-026, which excludes every earlier board and GH-029. The Slim kit lists every SCPH-700xx to 790xx board, with five install manuals chosen by where the video encoder sits, and excludes the SCPH-900xx. No open-source PS2 digital-video board has been found.

What changes value

A softmod is a file, not a modification: FMCB and its relatives live on a card or HDD and leave the console electrically stock. The PIC fix is a real reliability upgrade on an H/I Fat or a pre-6.04 SCPH-700xx. A modchip is permanent solder work, visible on inspection, and it can break FMCB and suppress the DVD player, so disclose it and do not assume a chipped unit is worth more. MechaPwn can leave a console reporting a changed region, and between MechaPwn versions a wrong model name, which is visible in the system menu. Name the SCPH-790xx’s ribbon holders in a listing, and on an SCPH-900xx state the datecode.

Recap and parts

The headline: the Fat’s internal supply is the part that fails from bad capacitors, the mainboards mostly do not, and the Slim rarely needs one at all.

Things to carry into an order:

• Read the board before you buy parts. The 3 mm (ψ3) 10 µF positions are four on GH-004 through GH-015 (C261, C262, C263, C266), five on GH-015 (adding C993), and only two from GH-017 on (C261, C263). Recap an early board off a late-Fat list and you leave two or three 3 mm parts in it. And GH-022 differs from GH-023 at shared designators: C297 is 220 µF on GH-022 and 470 µF on the later boards, and C487 is 220 µF on GH-022 and 180 µF later.
• The 3 mm parts are effectively obsolete, and Console5’s kits ship tantalum for them. On a tantalum the marking bar is POSITIVE, the opposite of the aluminium part it replaces. That is the easiest way to destroy a board during a PS2 recap.
• The constraint is height and diameter, not capacitance. Sony’s parts sit under shields and heatsinks. Nichicon UWT matches Sony’s case list, Panasonic FK has the 5 mm parts UWT lacks at 0.4 mm more height, and the 470 µF 4 V ψ8 has no exact-case replacement at all (both options are about 10 mm tall). The substitutes are in the tables below.
• Do not put a general-purpose electrolytic in a polymer position. Those are low-ESR parts on a tightly compensated buck loop. If one has genuinely failed, replace it with polymer of the same value, voltage and case.
• Sony did not print capacitor designators on most boards, per Console5 (single-source). Photograph the board before you lift anything and work from the parts list.
• Order Fat PSU caps against the board’s own maker. The bulk cap alone is 330 µF 315 V on a Delta, 270 to 330 µF 300 V on most Nichicon boards, 82 to 90 µF 400 V on two Nichicon models and 270 to 470 µF 220 V on Panasonic. Measure the can: some Nichicon boards use short, wide caps that Console5 reports as a custom PS2-only part.

For a refurb pipeline, a per-board cap kit is usually the right call, since the 3 mm substitution and the counts are already decided. For a one-off repair driven by one measured bad cap, buy the single part.

Capacitor lists

Per board, straight from Sony’s parts lists and cross-checked against Console5’s maps where Console5 covers the board. Boards with no OEM list carry no table. The Delta supply and the Network Adapter are listed separately because they do not track the mainboard at all.

Replacement parts

The non-capacitor parts a PS2 job actually consumes, most of them board-specific.

Chip & connector pinouts

The component library: each connector defined once, with a pin table and, where the package geometry supports it, an interactive diagram. Read the revision badge on each card before you probe, because on this console a designator does not guarantee the same pinout across boards.

Schematic facts

Reference confidence key

How to read the confidence tags and citations above. The bench tag appears on exactly one entry, my supply’s dummy-load measurement, which is not a console measurement. Every GH-035 table is tagged single-source because the one Slim service manual is the only document behind it.

Sources and further reading

These are the outside references this page is built from.

• Sony PS2 service manuals, archive.org collection (GH-004 through GH-026 and GH-035; the schematics behind every pin table here)
• gamingdoc: official PS2 service manuals index
• Sony PS2 Slim instruction manuals, SCPH-70001 and SCPH-79001 (the specification pages that settle the Slim DC input)
• psdevwiki: PS2 Motherboards and Chassis types (per-board and per-sub-revision notes, including the factory-fix tables)
• psdevwiki: Pickups and CDVD Drive
• psdevwiki: Power Supply and MechaCon
• ConsoleMods: PS2 Model Differences
• ConsoleMods: PS2 Power Supply Repair and Error Screens
• ConsoleMods: PlayStation AV Multi Out
• ConsoleMods: MechaPwn and FMCB
• Console5 TechWiki: PlayStation 2 (per-board and per-PSU capacitor maps; a vendor, so read kit notes as such)
• MechaPwn (README, PIC-fix notes and wiring diagrams)
• PMAP (the MechaCon adjustment tool and its test-point list)
• PS2Ident
• FreeMcBoot installer and FAQ
• SD2PSX PCB (the open memory-card design whose console-side pads corroborate the port pinout)
• BitBuilt: SCPH-79003 board scan and SCPH-90006 / 90010 board scan (the only rail maps for boards no manual covers)

The Slim USB-C supply is in the shop.