Jump to content

Commodore 64 Capacitor Replacement Guide

From RetroTechCollection

Recapping a Commodore 64 (all board revisions) restores stable power rails, lowers heat-stress on MOS chips and cures many “mystery” crashes, video glitches and SID noise. Because every C64 relies on an external +5 V DC / 9 VAC “brick” the on-board capacitors don’t fail quite as catastrophically as in computers with internal switch-modes, yet forty-year-old aluminium electrolytics are now well past their design life.

Visual Inspection & Failure Signs

[edit | edit source]
  • Bulging or leaning cans – especially the tall 4 700 µF reservoir (C88/C89).
  • Leakage residue – brown/green crust at base of C90 (colour-RAM decoupler) is common.
  • Ripple / hum on audio – audible buzz that follows screen brightness often tracks to dried-out C13 (SID analogue rail) or the RF-modulator’s 100 µF pair.
  • Intermittent reset when warm – C15 (10 µF reset RC) high-ESR causes POR threshold to mis-trip.

If one capacitor shows trouble, replace all electrolytics on the board.

Commodore 64 Logic-Board Capacitor Lists

[edit | edit source]

Unlike the VIC-20, the C64 underwent five major PCB generations. Commodore kept reference numbers fairly consistent, but always cross-check your silkscreen.

“Long boards” – Assy 326298 / 250407

[edit | edit source]
Early C64 Electrolytic Capacitors
Board ref Capacitance Voltage Purpose / rail
C88 4 700 µF 16 V Main +5 V bulk (from external PSU)
C90 1 000 µF 16 V Secondary 5 V at VIC/SID (reduces raster-noise)
C91 470 µF 25 V +12 V filter (SID 6581 & VIC 6567/9 analogue)
C13 220 µF 25 V 9 VAC ► +12 V doubler reservoir
C14 100 µF 16 V 9 VAC ► 5 V gate-array supply (PLA decouple)
C15 10 µF 16 V Power-on reset RC
C38 22 µF 16 V Audio path (SID DC-blocking)
C70 3.3 µF 50 V Chroma-coupling into RF modulator
C71,C72 100 µF 16 V RF modulator 5 V and audio buffer

Very early NTSC boards mark C90 as 2 200 µF – Commodore standardised on 1 000 µF by Rev-B.

Assy 250425 – first “cost-reduced” long board

[edit | edit source]

Same parts list as 250407 except C91 is omitted (12 V rail deleted when 8580 SID not yet adopted but cost saving began). Replace values exactly as found.

Assy 250466 – transitional board (8-chip DRAM → 2-chip DRAM)

[edit | edit source]
250466 Capacitors
Ref Capacitance Voltage Notes
C88 3 300 µF 16 V Commodore shaved value; 4 700 µF fits & improves ripple
C90 470 µF 16 V Lower current draw of HMOS chips allowed smaller part
C91 220 µF 25 V +12 V still present for 6581 SID (last board to use it)
Others identical Replace like-for-like

“Short boards” – Assy 250469 (A/B/C) & 250469-Aldi

[edit | edit source]
Short-Board Capacitor Set
Ref Capacitance Voltage Rail / circuit
C88 2 200 µF 10 V Bulk +5 V (HMOS load ≈ 600 mA)
C89 1 000 µF 10 V Local decouple for VIC-II (8562/8565)
C90 330 µF 10 V SID 8580 analogue +9 V
C91 *n/a* 12 V rail deleted; do not fit tall cans!
C15 10 µF 16 V Reset timer (value unchanged)
C38 47 µF 16 V Audio DC-block (lower because 8580 output bias differs)
C70 2.2 µF 50 V Chroma-coupling (smaller foil can type)
C71,C72 47 µF 16 V RF modulator local filters

The 250469-Aldi variant adds a linear 7812 + 470 µF 16 V can (C92) to power a 6581 SID on an otherwise 5 V-only board. If present, replace that extra capacitor as well.

Recapping Procedure

[edit | edit source]
  1. Disassemble : remove three rear screws, two under the keyboard, pop keyboard hinges.
  2. Remove RF shield : desolder the six folded-tabs or snip and replace with screws on re-assembly.
  3. Label keyboard & LED leads.
  4. Desolder each capacitor with braid + flux. Commodore used thick, high-temperature solder; 375 °C is usually required.
  5. Fit new capacitors, matching polarity and lead-spacing (5 mm on small cans, 7.5 mm on main filters). Keep height ≤ 17 mm so the metal shield clears.
  6. Clean flux, inspect for bridges.
  7. First power-up through a current-limited bench supply or a C64 Saver inline protector in case the original brick is bad.

Post-Recap Voltage / Ripple Checks

[edit | edit source]
Expected Rails – BASIC READY prompt, no cartridge
Test point (scope @20 MHz) Long board Short board Max ripple (p-p)
+5 V (U6 PLA pin 24) 4.95 – 5.10 V same < 50 mV
+12 V (SID pin 28) 11.5 – 12.6 V < 120 mV
+9 V (SID pin 28, 8580) 8.7 – 9.3 V < 100 mV
9 VAC (user-port pins 10/11) 8.5 – 10 V rms same sine-like

A VIC-II picture that jitters or vertical “rainbow lines” usually means > 150 mV of 5 V ripple.

[edit | edit source]
  • 70 W temperature-controlled iron (2 mm chisel tip).
  • Solder-wick and spring pump – double-layer ground pads are stubborn.
  • Leaded 63/37 or SAC-lead-free solder (avoid cheap flux-core that leaves residue).
  • 105 °C, low-ESR radial capacitors (Nichicon PW/PS, Panasonic FR/FC, Rubycon ZLH).
  • ask an expert or the AI about specialty caps like polyester and hybrid types. These have many advantages like longer/unlimited life and less noise, especially with regard to the SID and VIC filters.
  • IPA, nylon brush, Kapton tape to insulate cap cans that sit against the RF shield.

Extra Tips

[edit | edit source]
  • Check the brick first ! An over-voltage PSU kills fresh caps (and PLA / RAM) instantly.
  • SID heat-pad : early boards rely on C90 as a “heat-sink standoff” – keep replacement can the same diameter so the RF-shield pad still contacts the SID.
  • Retain RF modulator? If you have done an S-video or composite-bypass mod, you may delete C70/71/72 and the entire modulator 5 V line.
  • Socket inspection : reseat VIC-II, SID, PLA after the recap – temperature swings often unseat them.
  • Re-cap the 1541-II brick if bundled; its 4 700 µF can is the same age.
  • Ask an expert or the AI about specialty caps like polyester and hybrid types. These have many advantages like longer/unlimited life and less noise, especially with regard to the SID and VIC filters.
[edit | edit source]