Commodore PET 2001 Troubleshooting

A PET 2001 showing random symbols. On this machine the video circuitry reads screen RAM independently of the processor, so this display means the video path works — and, if it never clears, that the processor is not running.

This guide is for fault-finding on the Commodore PET 2001 — the original static board, with the 6502 at UF3, 6540 or 2316B ROMs and 6550 or 2114 RAM. For cleaning, socket work and the supply figures see Commodore PET 2001 General Maintenance; Commodore's own parts lists and schematics are all hosted here and indexed at Commodore PET 2001 Service Documentation.

Do not carry figures across from the Commodore PET 4000 Series. That is a dynamic board with a 6545 CRT controller, three on-board regulators and a −5 V rail. This machine has none of those.

Which board do you have

Commodore built four variants of the static board with identical component locations and different fitted devices. The assembly number is silkscreened in the upper left corner.

Assembly	ROMs	RAM
320008	6540	6550
320081	2316B	6550
320132	6540	2114
320137	2316B	2114

Power supply and startup

There is no separate power-supply box and no regulator of any kind on the main logic board. The transformer, rectifiers, reservoir capacitor, the +12 V regulator and the flyback are all on the display assembly, centimetres from the CRT neck.

• Mains voltage is on the display assembly whenever the machine is plugged in, and on the primary side of T1 regardless of the switch position.
• The CRT anode runs at roughly +10 kV from flyback T2 and holds its charge long after power-off — CRT Discharge Procedure.
• The same board carries +85 V and −30 V rails.
• C1, the 4700 µF 25 V reservoir, stores real energy. Bleed it through a resistor.

What generates what

Assembly	Produces	Regulator
Display assembly	+12 V, +85 V, −30 V, ≈+10 kV anode	VR1, a 7812, Commodore 901527-01
Main logic board	Nothing — it receives +5 V and +9 V unregulated	None fitted

Neither of Commodore's published parts lists for this machine contains a +5 V regulator or a +5 V reservoir, so this wiki gives no figure for where the +5 V rail is made. Earlier versions of this page said the board used a 7805, and elsewhere that it used four 78L05s; neither is in Commodore's documentation and the two contradicted each other. If your fault is a sagging or absent +5 V rail, work from the board in front of you and from measurements.

There is no −5 V rail documented anywhere in this machine.

Dead machine, in order

1. Fuse continuity, then the power switch. A fuse that blows again points at a shorted reservoir or a shorted rectifier.
2. AC at the transformer secondaries.
3. The rectifiers CR19–CR22 and reservoir C1, then VR1's input and output.
4. +5 V at 6502 pin 8 (UF3). Pins 1 and 21 are ground. Pin 40 is RESET, not a supply pin.
5. Inspect the solder joints at CR19–CR22 and VR1 — heavy-current, thermally cycled joints, and a cracked one produces an intermittent that mimics failing silicon.

Video wobble

A wobbly, warped or jumping display is classically an oxidised or cracked Molex power connector making the board voltage jump between roughly 4 and 5 V several times a second. Reseat, clean or replace the connector before suspecting the display board.

Reset and clock

The master crystal Y1 is 8 MHz.

The 6502 must see a brief low on RES at power-on, then a steady high. The system clock runs at about 1 MHz: pin 37 is φ0 in (the clock arriving from the board), pin 39 is φ2 out and pin 3 is φ1 out.

A voltmeter is enough for a first pass: a healthy φ2 averages around 2–3 V because it spends half its time high. A steady 5 V or a steady 0 V means it is dead.

If clock and reset are both correct and the machine is still unresponsive, move to bus activity.

The garbage screen

On this machine the video circuitry reads screen RAM continuously and independently of the processor. At power-on that RAM holds random data, so a healthy PET 2001 briefly shows garbage and then clears the screen once the processor runs its start-up code.

Two readings follow:

• A screen full of static garbage that never clears means the processor is not running — almost always a ROM or RAM fault, or a chip out of its socket.
• A screen of garbage is also proof that the video path works. Nothing downstream of the screen RAM is at fault.

Work in this order:

• Reseat every socketed chip. Chips working loose in PET sockets is one of the leading faults on any PET. The ceramic 6540 ROMs and 6550 RAMs are the worst offenders.
• Bad RAM is common, but the 74LS244 bus drivers fail at least as often. Check them early.
• Then the ROMs and the address decoding.

Processor, ROM and RAM

Check for address and data bus activity with a logic probe. Pulsing address lines mean the processor is fetching; static lines mean it is halted or held in reset. The SYNC output indicates an instruction fetch — no SYNC, no execution.

• 6540 ROMs are unreliable and now scarce. Substitute a tested part or an adapter carrying a 2716 or 2732 EPROM.
• 6550 static RAMs fail with age and heat. A single bad chip crashes the machine. Adapter boards carrying 2114s are the usual replacement.
• Address decoding — a 74154 or 74LS138 with a stuck output makes a whole region of memory unreachable. If part of memory is dead while the rest works, probe the decoder outputs rather than swapping memory.
• A diagnostic ROM will locate a bad RAM far faster than piggy-backing.

The diagnostic sense line is worth knowing: holding user-port pin 5 low during reset drops the machine into the machine-language monitor instead of completing a normal reset. Commodore also made a diagnostic clip — a 40-pin clip over the processor with a ROM and reset switch, which forced the address bus to $9xxx so the machine initialised from the clip's ROM.

Video faults

• No raster at all, no neck glow, no EHT whine — the display assembly, not the logic board.
• Raster present, nothing on it — screen RAM or the character generator. The character generator is at UA2.
• Consistent vertical stripes or a checkerboard — stuck address lines, faulty screen RAM, or a bad data buffer.
• Characters flickering or changing shape — the character generator, not the RAM. Constant vertical lines in the same position in every character row is the specific case: a bad character generator, or the shift register that takes data from it.
• Each character line advancing the screen address by 36 instead of 40, giving overlapped or wrongly wrapped text — a weak character clock in the video circuit, not a RAM fault.
• Screen "noise" on a 2001 is screen-RAM read/write contention, and is normal for the design.

Keyboard and I/O

The 2001-4 and 2001-8 use the chiclet keyboard, the machine's weakest mechanical assembly and the cause of most dead-key and key-bounce complaints.

• Whole dead rows or columns are far more often a connector, a cracked trace or a corroded via than a bad switch. Clean the pads with isopropyl alcohol and check continuity.
• No keys at all, or phantom keypresses, points at the scanning PIA or the row decoder.
• The 6520 PIAs are at UB8 and UG8 and the 6522 VIA at UA5. The two PIAs are the same part and can be swapped to confirm a fault.

IEEE-488: watch the bus during a LOAD or SAVE. No activity points at the I/O chip or at the bus transceivers (75160/75161 on the boards that use them). A transceiver that is hot or shows no activity is the usual culprit.

Cassette and sound

The cassette motor drivers run from the +9 V unregulated rail through a 7.2 V zener — not from +12 V. The drivers are TIP29 / 2N4401 / 2N3904 transistors.

• Motor does not run when PLAY is pressed: check the rail at the driver, then the driver transistor, then that the PIA toggles the control line during LOAD.
• The sense switch must close when PLAY is pressed. If it does not, the machine will not respond at all and the motor may look healthy. Clean the contacts and check continuity.
• Read failures: clean the head, check alignment, check the belt, then the read amplifier. Write failures: check the write amplifier.

For sound, the piezo transducer is driven from a port line. `PRINT CHR$(7)` rings the bell; if nothing sounds, trace from the port output to the transducer.

Logic probing

Compare activity between similar chips to find the anomaly:

• 6502 — SYNC for instruction fetch, φ2, RES.
• ROMs and RAMs — chip-select lines should pulse as the processor accesses them.
• 74154 — a stuck output leaves the corresponding ROM or RAM permanently unselected.
• 74LS244 and 74LS245 buffers — check for stuck outputs.
• 6520 PIAs and the 6522 VIA — a shorted one can lock the bus and stop the machine entirely.

Related pages

• Commodore PET 2001
• Commodore PET 2001 General Maintenance
• Commodore PET 2001 Capacitor Replacement Guide
• Commodore PET 2001 Service Documentation
• Commodore PET 4000 Series Troubleshooting Guide
• CRT Discharge Procedure
• Capacitor Failure Symptoms
• Recommended Tools