Amiga 600 Video Output Repair

U12, the CXA1145 encoder, on Commodore's A600 schematic, sheet 4: the band-pass filter Z221 and delay line Z222 above it, the RGB buffers Q231-Q233 at the right, and the composite output through C237 to CN10.

On the Amiga 600 every video output passes through U12, a Sony CXA1145 encoder. It produces the composite signal for the yellow jack CN10 and the RF modulator X2, and its R, G and B outputs, buffered by three transistors, feed the analogue colour pins of the 23-pin video port CN9. Leaking capacitors in this corner of the board cause most video faults: one colour missing from the RGB picture, composite and RF dead or flickering, or no picture at all.

Symptoms

A600 video faults reported by repairers

Symptom	What was found	Source
RGB picture with no red	The signal went into the red channel's transistor stage and nothing came out. Replacing the transistor, with wire links where corroded pads had lifted, brought red back.	The 8-Bit Manshed, via The Oasis BBS
RGB port missing the red component	Damage from leaking capacitors	robthenerd
Composite and RF dead, RGB working	The filter cans for the CXA1145, destroyed by capacitor corrosion	LinuxJedi
Composite dead	Continuity checks showed a failed section of the delay line	The 8-Bit Manshed, via The Oasis BBS
Red and green colour flicker on PAL composite, getting worse over time	The band-pass filter Z221 and delay line Z222 suspected; the picture got worse after cleaning the area with isopropyl alcohol	Waveguide
No picture at all	A member of the exxos forum notes that the A600 has no picture if the CXA1145 fails	exxos forum

How the circuit works

The video corner of a Rev. 2D board: the modulator X2 and composite jack CN10 at the top, the single filter can and the row of three cans at the lower left, and U12 at the bottom.

Commodore's schematic shows these paths round U12:

Signal	Path
R, G, B in	To pins 2, 3 and 4 through C211, C212 and C213 (0.33 µF). Composite sync enters on pin 10.
Chroma	Out on pin 15, through R221 (1 kΩ), C221, the band-pass filter Z221 and C222 (0.1 µF each), back in on pin 17.
Luma	Out on pin 16, through R222 (1 kΩ) and the delay line Z222, back in on pin 18.
R, G, B out	Pins 23, 22 and 21, through R231A-R231C (1 kΩ) to the bases of Q231, Q232 and Q233 (2N3904), then from the emitters through R234A-R234C (75 Ω) to the AR, AG and AB lines.
Composite	Pin 20, through C237 (470 µF) and R234D (75 Ω) and the filter E231 to CN10. A second path through C236 (100 µF) and R235 and R236 feeds the modulator's video input.
Audio for RF	The MONO audio line through C214 (10 µF) into pin 8; out on pin 9 through R237 and C235 (100 µF) to the modulator.

On sheet 6, AR, AG and AB pass through E260, E259 and E258 to the R, G and B pins of CN9. The RGB port therefore depends on U12 and on Q231-Q233, as well as on Denise.

Waveguide reports that Commodore used different parts for the filter and delay line from those in Sony's data sheet: the delay line is three cans joined in series and the band-pass filter is a single can. The cans have open bottoms against the board, so liquid on the board is drawn up into them.

Video parts (Commodore parts list)

Ref	Part	Commodore part no.
U12	Video encoder CXA1145M, surface mount	391084-01
Z221	Band-pass filter, PAL (BPF K5 MS) or NTSC (F-K5-MS)	391146-01 (PAL), 391146-02 (NTSC)
Z222	Delay line, PAL (F K5 MT) or NTSC (F K5 MD)	391147-01 (PAL), 391147-02 (NTSC)
Q231-Q233	2N3904, SOT-23	391121-01
C237	470 µF 16 V aluminium electrolytic	390101-03
C235, C236, C239	100 µF aluminium electrolytic, surface mount (case D)	391097-01
C214	10 µF 25 V aluminium electrolytic, surface mount (case C)	391097-07

Tools and parts

• An RGB monitor and a composite monitor, so the two outputs can be tested separately.
• Multimeter for continuity checks; an oscilloscope helps to follow the colour signals.
• Soldering equipment for surface-mount parts. LinuxJedi lifted the CXA1145 with a hot-air station.
• Spare 2N3904 transistors in SOT-23 for the RGB buffers.
• For a composite repair: a Sony CXA1645, with the parts from its data sheet circuit and a 1 % 16 kΩ resistor (20 kΩ for NTSC), or an adapter module.
• Kapton tape.

Procedure

Find the faulty output

1. Take the main board out; see the Amiga 600 Maintenance Guide. Check for shorts between the supply rails before powering a corroded board.
2. Connect RGB and composite in turn and note which works and whether a colour is missing.
3. Look round U12, the cans and the video electrolytics for leaked electrolyte. LinuxJedi notes that even a small amount of capacitor corrosion gets into the cans and destroys them.
4. Recap the board first if it has not been done; see the Amiga 600 Capacitor Replacement Guide. LinuxJedi recapped and cleaned this area before replacing the encoder, to rule the capacitors out.

One RGB colour missing

1. Trace the missing colour from U12's output pin (23 for red, 22 for green, 21 for blue) through R231, the transistor and R234 to the line that goes to CN9.
2. Probe the base and emitter of the transistor for that colour. In the 8-Bit Manshed repair there was activity going into the red stage and nothing coming out; a new transistor fixed it.
3. Check the tracks and pads round the stage. On the same board several corroded pads lifted during capacitor removal and were bridged with small wire links.

Composite or RF dead or flickering

1. Check continuity through the delay line Z222 and the filter Z221. The 8-Bit Manshed found a failed delay line section this way.
2. Keep cleaning fluids away from the cans. Waveguide suspects that cleaning with isopropyl alcohol made his picture worse, because the fluid creeps up into the open-bottomed cans.
3. No new cans are available, so repairers replace the CXA1145 with the later Sony CXA1645, which has the filter built in and is almost pin compatible. There are two ways to do it, described below.

Piggy-back CXA1645 (Waveguide)

Waveguide describes this as not a beginner's modification; it needs proper equipment and good surface-mount soldering skills.

1. Remove the CXA1145 from U12.
2. Cover the pads whose function changes on the CXA1645 with Kapton tape. The article's photographs show which pads.
3. Solder the CXA1645 in U12's place over the taped pads.
4. Add the new and changed parts from the CXA1645 data sheet circuit round the chip. Fit a 1 % 16 kΩ resistor from pin 18 to ground; it sets the colour signal amplitude. The author used 15 kΩ and 1 kΩ in series, measured to within 1 %. For NTSC the value is 20 kΩ.
5. Z221 and Z222 are no longer needed. The author left them in place because they are awkward to desolder, but says they can be removed.

On the CXA1145, pins 8 and 9 form the audio buffer for the RF modulator. Those pins have other functions on the CXA1645, so RF audio is lost; the author suggests that bridging pins 8 and 9 on the back of the board would probably restore it.

Adapter module (LinuxJedi)

LinuxJedi removed the CXA1145 with hot air, soldered a male header where it had been, and plugged in his own double-sided adapter. It carries a CXA1645 rewired to the CXA1145 pinout and adds the audio buffer that the CXA1645 lacks. The header means another chip, or the original, can be fitted later.

A reader of the Waveguide article reports doing the same job on an A1200 with a Sony CXA2075M, which has built-in 75 Ω drivers; the same pins still have to be lifted and insulated.

No picture at all

1. Check whether sync is present; a black screen with sync points away from the clocks.
2. Probe U12's R, G and B inputs on pins 2-4 and outputs on pins 21-23. With RGB taken from U12's outputs, a dead encoder takes the RGB picture with it.
3. If the board has not been recapped and also fails to boot, look at the reset circuit first; see Amiga 600 Reset Circuit Repair.

Reassemble

1. Clean the board and refit it. LinuxJedi gives repaired boards an ultrasonic clean; keep liquids away from any original cans still fitted.

Testing

1. Check the RGB picture for all three colours on a test screen or the start-up screens.
2. Check composite on CN10 and, if used, RF. After his CXA1645 change Waveguide reported a clean picture with no flicker; one reader saw faint vertical stripes, which the author put down to supply ripple from the power supply, bad capacitors or corroded supply tracks.
3. If the CXA1645 was piggy-backed without bridging pins 8 and 9, expect no sound through the RF modulator.

Board revisions

Commodore's schematic set is schematic 315987 Rev. C. The photograph above is of a Rev. 2D board, where U12, the cans and the video capacitors sit in the same corner beside the modulator. LinuxJedi notes the same composite failure on the Amiga 1200 and CD32, and Waveguide expects the CXA1645 change to be much the same on the A1200.

Related pages

• Amiga 600
• Amiga 600 Troubleshooting Guide
• Amiga 600 Capacitor Replacement Guide
• Amiga 600 Reset Circuit Repair
• Amiga 600 System Schematics (400422-02)