Famicom Family and PAL NES Repair Reference

This is my companion page to the NES front-loader reference. The Japanese Famicom and its relatives share the same brains as the American NES: the same RP2A03 CPU, RP2C02 PPU, the same address decode and controller readout. So instead of repeating what I already covered there, this page documents only what is different on the siblings, and there is more of it than people expect. If a procedure is shared (corrosion cleanup, general recap approach, bench testing method), I link back to the front-loader page rather than saying it twice.

The machines covered here:

Famicom HVC-001, the original 1983 red-and-white console with hardwired controllers and RF-only output.
AV Famicom HVC-101, the 1993 redesign that adds composite video and drops the hardwired pads.
Sharp Twin Famicom, an all-in-one Famicom plus Disk System in one case.
Famicom Disk System (FDS), the RAM adapter and Mitsumi disk drive that plug into a Famicom.
PAL NES-001, the European front-loader, which is mostly the American machine with a few region traps.

I want to be honest about scope up front. The American front-loader is my daily bench work; these Japanese and PAL siblings come across it less often, so the procedures below lean harder on documented references than on a stack of my own repairs, and I have cited them as such. The method is the same method I trust; the failure-rate claims are the community’s, and I say so where it matters.

Safety first: the DC power trap

Read this before you plug in anything. This is the single most important difference from the American NES, and getting it wrong is instantly fatal to the console.

The American NES-001 takes 9 V AC and rectifies it onboard with a full bridge, so it does not care about polarity and shrugs off a lot of adapter mistakes. The Famicom family does none of that. It takes DC input, it has no onboard bridge rectifier, and the polarity is not even uniform across the family. The Twin Famicom is wired the opposite way from everything else.

Console	Adapter	Rating	Polarity	Barrel
Famicom HVC-001	HVC-002	DC 10 V, 850 mA	center-NEGATIVE	5.5 / 2.1 mm
AV Famicom HVC-101	HVC-002	DC 10 V, 850 mA	center-NEGATIVE	5.5 / 2.1 mm
Famicom Disk System	HVC-025	DC 9 V, 400 mA	center-NEGATIVE	5.5 / 2.1 mm
Twin Famicom	AN-500 (part UADP-0041CEZZ)	DC 7.6 V, 1.25 A	center-POSITIVE	5.5 / 2.5 mm
PAL NES-001	NES-002E	9 to 9.8 V AC, 1.3 A	none (AC, onboard bridge)	(barrel N/A)

Two things will hurt a Famicom:

Reversed DC. On a bare HVC-001 the DC jack feeds the 7805 regulator almost directly, so reversed polarity is destructive. The AV Famicom and the FDS power board each carry a single series protection diode, which usually makes a reversed hookup inert rather than fatal on those two. That diode is a single diode, not a bridge, so it protects against reverse DC but does not make the console AC-tolerant.
AC into any of them. Feeding an American NES adapter (which is AC) into a Famicom is the classic mistake, and it destroys all of them. Community references describe the regulator failing violently and instantly, because the AC hits the 7805 unrectified. That an AC brick is not harmless is itself the proof that no bridge rectifier exists anywhere in this family.

No OEM document I could find states the polarity of any of these consoles. The Famicom manual lists “DC10V 850mA” and shows no polarity glyph at all. Every polarity fact in the table above is from secondary sources, but each one is corroborated across several independent references and I found no dissent. Even so, the bench rule stands on its own:

Meter the barrel of any unknown import adapter before the first power-on, every time. Do not trust a label, a bin, or my table alone. A modern regulated replacement supply is the safe path: 9 V, center-negative for the Famicom, AV Famicom, and FDS, and the correct polarity and 2.5 mm pin for the Twin.

Common problems and fixes

Dead, or died right after a power swap

If a Famicom died the moment someone changed adapters, suspect polarity or adapter type first (see the safety block), then walk the power rail. The path is similar to the NES but the parts live in different places.

On the HVC-001 the regulator is not on the mainboard. The DC jack, power switch, and 7805 all live on the RF and power daughter-board, and regulated 5 V arrives at the mainboard through the P3 harness on pins 2 and 3. So if you are probing for 5 V, probe P3 first. If those pins read 0 V, the fault is upstream on the power board (the 7805, its input, or the jack), not on the mainboard. A 7805 that reads roughly the input voltage on all three pins has failed short and is passing its input straight through, which can put an over-voltage on the logic.

The Twin Famicom does not use a 7805 at all. Its regulator is an M5236L (IC201) and R211 is the 5 V adjustment pot, so on a Twin you set 5.0 V rather than swapping a fixed regulator.

Won’t boot, garbage, or works only after reseating a cart

The Famicom uses a 60-pin cartridge slot, not the American 72-pin ZIF connector. That is good news: it does not have the front-loader’s downward-push pin-bending problem, so the fix here is cleaning, not replacing. Clean the connector and the cartridge edge fingers with high-purity isopropyl alcohol, and look for corrosion under the connector. There is no equivalent to the front-loader connector refurbishment because there is no ZIF fatigue to correct. If a clean, known-good cart still fails, move on to the power rail and the logic, using the same testing approach from the front-loader page.

One RF-Famicom quirk that gets misdiagnosed as a fault: on some RF board revisions there is no picture at all unless a cartridge is inserted, because the video path is gated by a cart bridging two of the 60-pin contacts. A blank screen with no cart in is stock behavior on those units, not a failure.

Famicom Disk System will not load a disk

This is the dominant FDS complaint, and community references are unanimous that the usual cause is the drive belt. The Mitsumi Quick Disk mechanism uses a thin rubber belt, and after 30-plus years that belt stretches, goes gummy, or fully melts and half-welds itself to the pulleys. Sources call the belt the single most common failure in the whole family, and I flag that as a claim worth confirming on your own bench rather than assuming: open the drive and look. A healthy belt is a taut, clean loop. A dead one is slack, sticky, split, or shows up as a black smear on the pulleys.

The tell that separates a belt fault from an alignment fault: a belt problem means the disk barely spins or does not spin, and you often get ERR.02 at boot even with power confirmed. An alignment problem means the disk spins fine but misreads. If the motor energizes and the disk still will not turn, it is the belt.

Belt replacement is its own small project, covered below under “FDS drive service,” because a belt swap always disturbs the alignment and you have to re-verify it afterward.

FDS spins fine but throws read errors (ERR.21, ERR.22, ERR.27)

If the disk spins normally but the drive errors out, and especially if this started after a belt swap, you are into alignment, not the belt. But before you touch a single alignment screw, prove the media. Keep a known-good reference disk on the bench and test with it first, because a drive fault fails every disk while a media fault follows one disk from drive to drive. FDS disks have no shutter, the magnetic surface is exposed to dust and fingerprints, and the magnetic layer oxidizes and demagnetizes with age. Chasing alignment on a rotten disk is the most expensive time sink in FDS repair. If several disks read and one never does, stop working on the console and clean or rewrite the disk.

The FDS error codes have two published readings that do not fully agree (the BIOS-literal list versus the repair community’s bench heuristics), so treat a few codes as ambiguous. Broadly: ERR.21 and ERR.22 mean the drive cannot find the start of the data; ERR.23 through ERR.25 mean it found the start but is losing data further into the spiral; ERR.27 is commonly a worn felt pressure pad on the bench, though the ROM itself just means a premature block end. The alignment procedure below addresses all of these.

Wavy or dim video, hum, or buzz

As on the NES, the usual suspect is the electrolytic filter capacitors. The critical difference: do not reuse the NES cap list here. The Famicom family capacitor values and designators are board-specific, and I cover them under “Recap and parts” below. Community references single out particular caps as the common failures, and I treat those as leads to check rather than parts to replace on faith: inspect for bulging or leaking, scope the rail for ripple, and replace what actually reads bad.

One dim-picture case that is not a fault at all: early RF Famicom boards use a different video output resistor value than later boards, so a slightly dim color on an early board is stock. And the Twin Famicom has a factory audio layout that makes it sound muffled out of the box (a capacitor and resistor forming a low corner frequency). That dull sound survives a recap because the offending part is a ceramic, not an electrolytic, so a recap that “did not fix the dull audio” is not a failed recap.

Internal corrosion

Battery leaks and old liquid exposure leave corrosion just like on the front-loader, most often around the power and RF sections. I handle it the same way I describe on the front-loader page: neutralize the active corrosion, clean thoroughly with isopropyl alcohol, and seal the treated area.

FDS drive service: belt and alignment

The Famicom Disk System drive is the one genuinely different repair in this family, so it gets its own section. Twin Famicom owners: your integrated drive is the same mechanism and the same procedure.

The belt. No OEM belt spec exists, so this is derived from the drive geometry rather than measured off an original belt, and the published numbers are easy to misread. There are two different circumference figures floating around and they are not in conflict: one is the belt stretched and installed, the other is the belt relaxed. The stretched, installed path works out to roughly 245 mm (a fold of about 122 mm), while the relaxed loop is roughly 229 to 233 mm (a fold of about 115 to 116 mm). Buy an FDS-specific belt sold with a stated fold length, because unlabeled “FDS belts” have been sold at wrong sizes, and do not substitute generic Mobilon bands, which one detailed source found to be both too small and too rigid. Clean every trace of the old belt off the pulleys with isopropyl on a swab, and keep finger grease off the new belt and the pulleys or it will slip immediately.

The alignment. A belt swap always moves the calibration, so plan on re-checking four adjustments afterward. In order:

Spindle hub timing, held by a 1.5 mm hex set screw. This is the “which part of the spiral” adjustment.
Drive speed, set by a pot on the motor. The corrected target is 400 RPM measured at the disk-table shaft (roughly 800 RPM at the motor belt holder). Note that older references quote figures around 820 to 873 RPM at the shaft; those were retracted by their own author as a stroboscope-app harmonic error, so treat any 800-to-900 RPM reading at the shaft as a doubled artifact and calibrate to 400 with a real tachometer.
Head position, adjusted at the head screw to roughly 10.68 to 10.72 mm with a working tolerance around plus or minus 0.05 mm. A quarter turn moves it about 0.1 mm, so work in eighth-turn steps.
Pressure pad. Replace the felt if it is thin or worn (a common ERR.27 cause).

The one hard rule every source agrees on: adjust the head screw only. Do not touch the disk-clamp or spindle set screw once the drive is factory-aligned. A practical technique is to iterate the head in eighth-turn steps in one direction until ERR.21 flips to ERR.22, which marks one boundary of the good window, then reverse and bracket back into the center. Scrape any locking glue off the head screw with a perfectly fitting driver first, or you will strip it.

If you would rather remove the whole media-and-belt variable, an FDSKey or FDSStick emulates the drive from flash storage, which I cover under Mods.

Inside the family: what differs

A short tour of the deltas from the American front-loader, which makes the faults above make sense. Everything not listed here (the CPU, PPU, RAM map, controller readout) is identical to the front-loader and lives on that page.

Power is DC with no onboard bridge, and the regulator is not always on the mainboard (it is on the RF and power board on the HVC-001, and it is an adjustable M5236L on the Twin). This is the whole reason the polarity trap exists.
The cartridge bus is 60-pin, and it carries an expansion audio loop that the American 72-pin slot does not have. Pin 45 carries the console’s amplified audio out to the cartridge, and pin 46 carries audio back to the RF and audio section. Most carts simply bridge these two pins. Carts with their own sound hardware (the FDS RAM adapter, and mappers like VRC6, VRC7, MMC5, N163, and 5B) inject their audio into this loop, which is why Famicom expansion audio “just works” on real hardware and needs a mod on an American NES. If those two pins are open on the cart or the trace is broken on the console, the audio is silent, so check the loop before suspecting the sound chip.
Reset works differently. On the Famicom the PPU reset pin is tied to +5 V, so pressing Reset does not clear the screen the way it does on some machines. That is normal, not a stuck reset.
Controllers on the HVC-001 are hardwired, and controller 2 legitimately has no Select or Start buttons because a microphone occupies that space instead. A dead hardwired pad is usually harness flex fatigue where the cable enters the case. The AV Famicom’s detachable controller ports omit two data lines, so peripherals like the Zapper and Power Pad do not work on a stock AV Famicom. That is a design limitation, not a fault to chase.
Clocks are region-specific. NTSC-J machines run a 21.477272 MHz master crystal with the RP2A03 and RP2C02; PAL runs a 26.601712 MHz crystal with the RP2A07 and RP2C07. The region lock is baked into the chip divisors, so mixing region parts does not “mostly work,” it detunes the CPU speed, dot clock, and colorburst together. The NTSC-J oscillator has a real, adjustable trimmer (TC1), so a Famicom that will not color-lock on a picky TV may just want the trimmer set with a frequency counter, not a recap.
The PAL NES-001 is otherwise the American front-loader with two region traps: PAL-A and PAL-B use different, mutually incompatible lockout chip codes (a 1 Hz blink on the “wrong” PAL cart is a region mismatch, not hardware), and some PAL decks have controller-port protection diodes that can make otherwise-good NTSC pads read as dead. Its power path is the American AC-plus-onboard-bridge design, so do not carry Famicom DC and polarity logic over to a PAL NES.

Mods worth knowing

I do not reproduce anyone’s install guide. This is an orientation to what is worth doing and where to get the real instructions.

Composite / AV mod (RF Famicom only). The HVC-001 has no composite output at all, so adding it is the headline mod. It taps composite video off PPU pin 21 through a small transistor amplifier, with audio off the cart-audio pin. Do not ever destroy PPU pin 21, because a future RGB install needs it too. The low-labor path is a drop-in power-and-AV replacement board (for example the Backoffice Power VAMP), which is fully reversible and needs no case cutting. The AV Famicom and Twin already have composite, so this mod does not apply to them. Reference: nesdev PPU pinout (VOUT / composite amp).
RGB output: NESRGB. Tim Worthington’s board does RGB, S-video, and composite with no lag, and it fits every machine in this family via a per-console adapter board (the main board is the same across consoles; only the small adapter differs). The AV Famicom is the cleanest and most profitable host because its Multi Out needs no case cutting. As of the version 5 board (2026), the encoder moved into FPGA logic and the output set grew to composite, S-video, RGB, and component. One caution that only becomes visible after an RGB install: a G-revision-or-earlier PPU has a known speckle defect on RGB, so read the PPU revision before quoting the job. Reference: etim.net.au NESRGB and its Famicom install guide.
HDMI: Hi-Def NES. Kevtris’s kit gives a lag-free digital picture, but it does not fit every machine in this family: the original Famicom’s chip revisions are not supported, the AV Famicom and PAL NES are, and the Twin is untested. Confirm compatibility for your exact machine first. Reference: ConsoleMods: Hi-Def NES.
Region and lockout. This is a non-issue on the Famicom family, which has no lockout chip at all. It only matters on the PAL NES, where the fix is the standard CIC handling covered on the front-loader page.
FDS drive emulation. An FDSKey or FDSStick replaces the disk media entirely with flash storage, which removes the belt, alignment, and rotting-disk variables from a resale unit in one move. Reference: FDSStick.

For a broader mod orientation, RetroRGB’s NES mods index is the best single jumping-off page and covers the Famicom family too.

Recap and parts

The one rule to carry away from this section: the Famicom family caps are board-specific, and reusing an American NES cap kit will put wrong values in the wrong places. Identify your exact board by its silkscreen before you order.

Community references point at a handful of usual-suspect electrolytics per board: the HVC-002 power brick’s large filter cap, roughly 1000 uF filter caps on the Famicom and AV Famicom mainboards, and a 2200 uF main filter on the Twin’s power board. I treat these as parts to inspect and test, not parts to swap on faith. There is one value trap on the AV Famicom worth calling out: at least one position must stay at its small original value (a 47 uF), so do not “round up” every cap when you recap.

A couple of practical notes that carry over from my general recap and testing method (full version on the front-loader page):

When you replace a ceramic capacitor with a modern MLCC, remember that MLCC capacitance drops under DC bias, sometimes dramatically. Do not blindly swap a ceramic for a same-nameplate part in a filtering or timing position without accounting for the derating.
The Twin Famicom’s regulator is adjustable (R211), so after any power-board work, set 5.0 V rather than assuming it.

If you would rather buy a console that has already had this work done, everything I restore is in the shop. I will link specific Famicom-family service and RGB-mod listings here as those pages firm up.

Sources and further reading

These are the outside references I trust for the Famicom family and PAL NES. I link them rather than copy them, and where a source is a community schematic redraw or a photo I link it rather than reproducing the image.