Logic Chip Testing

A logic probe with HI and LO lamps, a TTL/CMOS switch, a MEM/PULSE switch and clips for the circuit's supply (photo: Christian Taube, CC BY-SA 2.5).

Logic chip testing is finding out whether a digital integrated circuit (a gate, counter, buffer, RAM, ROM or custom chip) is doing its job, either in the machine or on a tester. On an old board a missing picture, a crash or a wrong memory count usually comes down to one chip among dozens, and most of the methods here narrow the list without unsoldering anything. A chip should only be condemned after its supply, clock and inputs have been checked, because a good chip with a bad input gives a wrong output.

When to use it

• The machine's symptoms or its diagnostic ROM point at a group of chips. The Commodore 64 PLA Replacement and Commodore 64 RAM Replacement guides are examples.
• A signal is stuck high or low, or missing, at a chip's output while its inputs are correct. Tracing the signal to that chip is covered in Continuity Testing and Signal Tracing.
• A fault appears only after the machine warms up, or clears when it is cooled.
• A chip already out of the board needs checking before it is thrown away or refitted, or a second-hand replacement needs checking before it is fitted.

If a supply rail is pulled down or a chip is running very hot, start with Finding Short Circuits.

Safety

• Logic chips are damaged by static. Use an earthed wrist strap with its built-in resistor.
• Switch off before fitting, removing or piggybacking a chip. Carlsen warns that a piggybacked test chip fitted the wrong way round will probably be damaged when power is applied.
• Freeze spray puts a static charge on what it hits; use an anti-static grade on static-sensitive boards.

Tools and materials

• The datasheet or pinout for every chip under test. Most 14-pin 74-series chips have their supply on pin 14 and ground on pin 7, as on the 74LS00, but not all: the 7490 and 74LS90 counters take their supply on pin 5 and ground on pin 10.
• A multimeter and an oscilloscope.
• A logic probe. Hewlett-Packard's 545A used TTL thresholds of 2.0 V and 0.8 V, matching the 74LS00's specified input levels, or CMOS thresholds of 70 % and 30 % of its supply.
• A logic pulser, to drive a node to the opposite state on demand.
• A logic clip, which shows the state of all 16 pins of a DIP chip at once.
• Known-good chips of the same types, for substitution and piggybacking.
• Optionally an out-of-circuit tester. The Retro Chip Tester Professional from 8Bit-Museum.de tests static and dynamic RAM, ROMs and EPROMs, and over 1,200 TTL and CMOS logic chips; it also saves ROM contents to a memory card and names known ROMs from a database of 400,000 entries.
• Optionally an analogue signature analyser, such as a Huntron Tracker, for comparing a dead board with a good one with the power off.
• Electronics freeze spray, and a hair dryer, for heat-related faults.

Procedure

Check the chip's surroundings first

1. Measure the supply at the chip's own supply and ground pins, taken from its datasheet. 74-series TTL is specified for 4.75 to 5.25 V.
2. Check the clock and reset signals that reach the chip. Commodore lists the power supply and the reset and clock circuits alongside the chips themselves as causes of a C64 that will not run its diagnostic cartridge.
3. Check each input with the logic probe. A dim lamp on an input means it is floating or sitting between logic levels, which is a fault in whatever drives it, or an open track.

Check the logic in circuit

1. For a gate or other combinational chip, read every input and output with the logic probe and compare the output with the chip's truth table for those inputs. HP describes the logic probe being used in this way to verify combinational relationships.
2. For a counter, latch, shift register or flip-flop, use the logic pulser to clock it or to set and clear it, and watch the outputs with the logic probe or a logic clip. HP's pulser and clip were designed to be used together to step a chip through its truth table.
3. Where nothing on the board is moving, the pulser can stand in for the clock, a pulse at a time.
4. If an output is stuck, decide whether the chip is failing to drive it or something else on the node is holding it. HP's current tracer shows a large current from a working driver into a shorted input, and almost none from a dead driver. Commodore's manual uses a cut instead: it cuts the PLA's CASRAM pin and measures the cut pin, so that a good signal on the pin clears the PLA and points at the RAM.

Compare against a reference chip in circuit

HP's 10529A logic comparator clipped onto a 14- or 16-pin TTL or DTL chip in a powered circuit and compared it pin by pin with a known-good reference chip of the same type fitted in the comparator. It lit a lamp for any pin that differed, and stretched errors as short as 200 to 300 ns so they could be seen. HP points out that a failed input pin shows up as an error on the chip that drives it, so the comparator points to a faulty node, not always to the faulty chip.

Piggyback a known-good chip

Terry Stewart's piggyback test on an Atari 800XL: a known-good NEC 4164 pressed on top of the RAM chip at U14.

Piggybacking puts a known-good chip on top of a suspect one, pin to pin, so that the two work in parallel. It is used mostly on RAM, where a board has many identical chips. Ray Carlsen's method for the Commodore 64:

1. Press the pins of the known-good chip against a flat surface to bend them in slightly, so that it grips the chip below.
2. Switch off. Fit the test chip on top of the first suspect chip, notch over notch, with every pin touching its partner and none touching a neighbour.
3. Switch on and watch the symptom. On the C64 Carlsen watches the bytes-free figure on the start-up screen; if it changes, even without becoming correct, the chip underneath is the suspect.
4. Switch off and move the test chip to the next chip, until every chip has been tried.
5. Replace the suspect chip. Never leave the piggybacked chip in place: Carlsen warns that it can cause timing problems or stop working, and that the board chip can go on deteriorating until it shorts.

Piggybacking has limits. It does not work on a chip that has failed short and holds the machine at a blank screen. Terry Stewart found that it often, but not always, finds the failed RAM in an Atari 800XL, and another owner's machine stopped working when he tried it; a chip can only be piggybacked on one with the same pin count and pinout.

Heat and cold

1. Feel the chips with the machine running. Carlsen finds that C64 RAM normally runs barely warm, and that a chip that gets hot quickly, or noticeably warmer than the others of its type, is bad. Some chips run hot when working: an original C64 PLA does, so its temperature alone proves little.
2. For a fault that appears after warm-up, let the machine run until it fails, then spray the suspect chips one at a time with freeze spray. The chip whose cooling brings the machine back is the faulty one.
3. Goldklang also uses a hair dryer to locate thermal problems, warming the area where the fault is suspected.

Substitute

1. Where the chip is socketed, swap it for a known-good one of the same type. Carlsen notes that substitution is the only practical way to test many of these chips.
2. Where Atari's or the manufacturer's flowchart gives an order, follow it. Atari's 2600 procedure swaps the 6507, the 6532 and the TIA in turn, leaving each known-good chip in place when it does not cure the fault.
3. To check a suspect chip, fit it in a known-good machine. Sinclair's QL manual says to test a suspect chip in a known-good QL, not to put a good chip into the faulty one.
4. Where the chip is soldered in, desolder it intact unless you are sure it has failed, because it may turn out to be good, and fit a socket. See IC Removal and Socketing and Through-Hole Desoldering.

Test out of circuit

1. Fit the chip in the tester's socket, the right way round, and choose its type or let the tester identify it.
2. For RAM, the Retro Chip Tester reports the failing memory cells, and can store the result on a memory card.
3. For a ROM or EPROM, read its contents and compare them with a known-good image; the Retro Chip Tester names ROMs it recognises from its database.
4. If the fault in the machine was intermittent or heat-related, put the chip back and repeat the heat and cold tests above before trusting a pass.

Compare a dead board with a good one, power off

Analogue signature analysis applies a current-limited AC sine wave across two points of a component or circuit, and displays current against voltage. The pattern is compared with the same point on a known-good board. Huntron, whose Tracker instruments do this, lists the advantages: the board need not power up, the test cannot damage it further, and no schematic is needed. It does not test function, and reading the patterns takes experience.

1. Switch off and disconnect the board under test and the known-good board.
2. Put the instrument's leads on the same pin of the same chip on each board in turn.
3. Look for a pin whose pattern differs. Huntron's examples include a damaged input junction on a 74S04 and a capacitor whose connection had come open.

Which method suits which fault

Fault	Methods	Notes
Wrong output from a gate or counter	Logic probe, pulser and clip; comparator	Check supply and inputs first
Node stuck high or low	Current tracer; cutting or lifting a pin	Tells a dead driver from a shorted input
One of a bank of identical RAM chips	Piggyback; diagnostic ROM; heat	Piggybacking fails on a chip that has failed short
Fault after warm-up	Freeze spray; hair dryer	Cool or warm one chip at a time
Socketed chip under suspicion	Substitution	Test the suspect in a known-good machine
Board will not power up	Analogue signature analysis	Needs a known-good board to compare
Removed chip or new stock	Out-of-circuit tester	Confirm in the machine if the fault was intermittent

Checking the repair

1. Remove any piggybacked test chips.
2. Power up and repeat the test that showed the fault: the start-up screen, the memory count or the machine's diagnostic ROM.
3. Run the machine long enough to warm up, and switch it off and on several times. Carlsen's PLA check on the C64 is to switch off and on rapidly several times and watch for a bad screen.
4. Check the power supply that feeds the board. Carlsen finds that shorted C64 RAM usually points to a failing power supply, which must be checked before the machine is used again.

Common mistakes

• Condemning a chip whose supply, clock or inputs are wrong.
• Taking the supply pins of one chip for those of another. Not every 14-pin chip has its supply on pin 14.
• Using TTL thresholds on a CMOS circuit, or the other way round, on the logic probe.
• Fitting a piggyback chip backwards, or leaving it in place after the test.
• Expecting piggybacking to find a chip that has failed short.
• Reading a hot chip as a bad one when that type runs hot when working.
• Fitting a good chip into a board that has just destroyed one, before finding out why.
• Desoldering a chip that turns out to be good, and damaging its pads in the process.
• Blaming the chip a logic comparator flags when the fault is an input on another chip on the same node.

Worked examples on this wiki

• Commodore 64 PLA Replacement: substitution, Commodore's signal checks with the diagnostic cartridge, and the cut-pin test on the PLA's CASRAM pin.
• Commodore 64 RAM Replacement: the Dead Test flash codes, the heat check, piggybacking and Commodore's checks on each RAM chip.
• Atari 800XL RAM Replacement: Terry Stewart's piggyback test that found the failed 4264 at U14.
• Sinclair ZX Spectrum RAM Replacement: substitution and holding a known-good chip on each suspect chip.
• Atari 2600 TIA, 6507 and 6532 Replacement: Atari's order for swapping the three main chips with known-good parts.
• TRS-80 Model 16 Troubleshooting Guide: freeze spray or a hair dryer for intermittent faults.

Related pages

• Continuity Testing and Signal Tracing
• Finding Short Circuits
• Testing Diodes, Transistors and Regulators
• IC Removal and Socketing
• Through-Hole Desoldering
• Recommended Tools