Top 27 SMT Pick and Place Machine Failures and Quick Fixes
1. Why Pick and Place Failures Matter
If your SMT pick and place machine stops even once a shift, you’re bleeding money—and these SMT machine for PCB assembly headaches are behind 90% of unplanned downtime in 2025.
Here are the exact 27 failures you’ll run into on Fuji, Yamaha, ASM Siplace, Panasonic, JUKI, and Hanwha lines, plus the dead-simple fixes that get you running again in minutes instead of hours.
2. Nozzle and Vacuum System Failures
Vacuum problems are the number one reason your high-speed pick and place machine throws pick errors and stops the line.
On Fuji NXT, Yamaha YSM, ASM SIPLACE, and Panasonic NPM heads, everything depends on stable negative pressure—drop below 50-60 kPa for even a split second and the part is gone.
Here’s the deeper dive you need to really understand and fix these issues for good.
2.1 Nozzle Vacuum Loss or Weak Pick-Up
2.1.1 Why It Happens
The vacuum path is long and full of weak points: plant air goes through a 5-micron filter, pressure regulator, FRL unit, manifold, solenoid valve, flexible tubing, head valve block, ceramic shaft seal, and finally the nozzle bore.
Any single leak or restriction here collapses vacuum.
Common culprits include cracked polyurethane tubing (hardens after 2-3 years), clogged 0.01-micron inline filters on the head, failed piston O-rings in the valve block, or a dirty vacuum pump muffler that raises back-pressure.
On ASM E-by-Siplace heads you’ll also see the small sintered bronze filter inside the shaft clog with flux dust in under 6 months if you run no-clean paste.
2.1.2 Quick Checks
Go to the maintenance screen—most machines show live vacuum in kPa per nozzle station.
Pick a 1005 resistor with a known-good nozzle; if the reading drops more than 10 kPa during upward motion, you have a leak.
Next, remove the nozzle and put your finger over the shaft tip—if vacuum doesn’t instantly pin your finger and hold 30+ seconds, the problem is upstream of the nozzle.
Finally, disconnect the main air hose at the machine inlet and measure plant pressure with a gauge; anything under 0.55 MPa at peak usage will starve the ejector.
2.1.3 Step-by-Step Fix
Start cheap: replace the two inline filters on the head (Fuji part H1321A, Yamaha 90990-23J00 equivalents).
Then cut 10 cm off each flexible tube end—cracks always start at the fittings.
If still low, remove the valve block, soak in IPA ultrasonic 10 min, blow out every port with 0.2 MPa air, and replace all four O-rings (Viton 70 duro works better than nitrile in flux environments).
On Yamaha Sigma and YSM20 heads, run the built-in “vacuum leak test” routine—it pressurizes the system positively and listens for leaks with a microphone sensor.
Takes 90 seconds and tells you exactly which shaft or valve is bad.
2.2 Nozzle Clogging or Blockage
2.2.1 Common Causes
Flux volatiles boil during reflow and get sucked back into the nozzle when the blow-back happens too late.
Tiny 0201 chips sometimes fracture during pick and a 50-micron shard lodges exactly at the nozzle tip ledge.
BGA flux residue is the worst—it carbonizes and turns rock-hard inside 0.2 mm bore nozzles.
2.2.2 Fast Detection
Watch the pick retry counter on the feeder—if one feeder suddenly jumps from 0 to 50 retries per shift, 9 times out of 10 it’s a clogged nozzle.
Also look for “vacuum level low after pick” alarm even though the part is still on the nozzle—means the bore is partially blocked and pressure can’t recover fast enough.
2.2.3 Effective Cleaning Methods
Forget poking with wire—you’ll scratch the bore and ruin centering.
Use a proper nozzle cleaning station with high-pressure IPA + ultrasonic + vacuum cycle (Yamaha YSC-01 or Fuji NC08).
At minimum, soak in 99% IPA ultrasonic bath 5 minutes, then blow from the top with 0.3 MPa air while holding the nozzle tip-down over a white paper—watch for black specks coming out.
For carbonized BGA flux, soak overnight in fuming nitric acid substitute (some factories use EnviroTech NC-100) then ultrasonic again.
2.3 Nozzle Tip Wear or Damage
2.3.1 Symptoms of Wear
Placement offset drifts >30 µm over a shift, side-view camera shows shiny ring worn into the tip face, or you start getting “component thickness error” even though height data is correct—the worn tip is actually shorter than the machine thinks.
2.3.2 Replace or Restore
Ceramic nozzles (Fuji H04/H08/H24 series) can sometimes be re-faced on a diamond lapping plate to remove 0.05-0.1 mm and restore flatness, but only if the bore is still perfect.
Tungsten carbide and steel nozzles are not worth it—just replace. Never mix old and new tips on the same head; the height difference will cause Z crashes on fine-pitch parts.
2.3.3 Spare Parts Strategy
Stock at least 20 pcs of your top five nozzle types.
Track usage in a simple Excel: when a nozzle reaches 800k picks on 0402/0201 or 1.5M picks on larger parts, retire it to training machines.
Fuji and Yamaha both offer DLC (diamond-like carbon) coated nozzles now—twice the life on abrasive alumina substrates.
2.4 Dropped Components During Transfer
2.4.1 Root Causes
Vacuum decay curve too steep (common when the head valve switches to low-vacuum mode too early), center-of-mass outside the nozzle bore on odd-form parts, or static charge pushing the part sideways on shiny plastic packages.
2.4.2 Vacuum and Height Adjustments
Increase “transfer vacuum” level 5-8 kPa above normal pick vacuum and extend the high-vacuum hold time by 30-50 ms (found in advanced placement parameters on ASM and Panasonic machines).
Raise Z-top height 1-2 mm if you’re running tall capacitors.
2.4.3 Nozzle Selection Tips
Use custom nozzles with larger contact area for heavy parts (>0.5 g).
Fuji “large area” and Yamaha “multi-surface” nozzles reduce drop rate on LED and shielding cans from 1 in 500 to 1 in 50,000.
2.5 Multiple Component Pick-Up (Double Pick)
2.5.1 Why It Happens
Cover tape peel-back distance too long + static + vacuum slightly high = second part sticks to the bottom of the first.
Happens most on 8 mm paper tape with 01005 and 0201 because parts are lighter than the static force.
2.5.2 Vacuum and Timing Fixes
Drop pick vacuum 8-12 kPa below the level that just barely picks one part.
Shorten pick-to-blow-back delay from default 80 ms to 30 ms on JUKI RX-7 and Hanwha HM520—this gives the extra part almost no time to stick.
2.5.3 Using Blow-Back
Enable “strong blow-back” or “double blow” mode (ASM calls it “active blow-off”).
It pulses +0.1 MPa air downward right after pick—knocks the second part back into the pocket every time.
Combine with ionised air over the feeder bank and double-pick disappears even on 01005 reels.
3. Feeder and Material Handling Issues
Feeder problems are the silent killers of uptime on any SMT line—whether you’re running JUKI electric feeders, Yamaha SS/ZS series, ASM Siplace X feeders, Panasonic NPM intelligent feeders, or Fuji NXT smart feeders.
A single bad feeder can drag your placement rate from 100k cph down to a crawl and spike scrap. Here’s the deep technical breakdown you need to diagnose and fix them permanently.
3.1 Feeder Jamming
3.1.1 Why Feeders Jam
The root cause is almost always mechanical interference in the tape path:
Cover tape peels at the wrong angle and catches on the sprocket, component pins bend during reel winding and snag the pocket edge, static builds up on paper tape and parts stick to each other, or the reel brake is too tight and tape buckles.
On 8 mm tape with 01005/0201, even a 0.1 mm misalignment of the peel blade causes the cover tape to tear sideways and wrap around the advance roller.
3.1.2 Immediate Actions
Stop the machine, open the feeder door, gently pull the tape backward 2-3 pockets to free the jammed part (never forward or you’ll skip pockets and cause missing errors later).
Remove any torn cover tape pieces from the sprocket teeth with plastic tweezers—metal ones scratch the aluminum.
Then advance manually five times while watching the peel blade height; it should be 0.2-0.3 mm above the carrier tape surface.
3.1.3 Preventive Tips
Set cover tape peel force to the minimum that still removes cleanly (Yamaha ZS feeders let you adjust this in 1-gram steps).
Install overhead ionizers directly above the feeder bank—static voltage above 100 V doubles jam rate on paper tape.
For high-volume 8 mm, switch to PSA (pressure-sensitive adhesive) cover tape reels; they peel cleaner than heat-seal types.
And always store reels in humidity-controlled cabinets at 40-60% RH to prevent tape embrittlement.
3.2 Feeder Timing or Advance Problems
3.2.1 Tape Feeding Mistakes
The advance motor misses steps because the sprocket is worn (teeth rounded after 2-3 million cycles), the optical sensor is coated with paper dust and can’t see the pocket edge, or the feeder firmware has an old timing table that doesn’t match the new thinner carrier tape.
On Panasonic NPM feeders you’ll also see “advance timeout” when the reel inertia is too high from a nearly empty heavy reel.
3.2.2 Sensor Reset
Clean both the pocket sensor (small window on the left) and the cover-tape sensor (right side) with 99% IPA and a lint-free swab.
Then run the feeder’s self-test routine—most brands have a “sensor check” that flashes an LED if light transmission is below 70%.
If the sensor board itself is failing (common on older JUKI feeders after 5+ years), the LED stays dim even when clean; replace the whole sensor unit.
3.2.3 Feeder Calibration
Insert a precision calibration tape (glass or metal with exact 4 mm pitch holes) and run the auto-calibration from the machine menu.
Fuji NXT III and Yamaha YSM20 calibrate in under 20 seconds and store individual offset per slot.
Do this every time you swap feeder types (electric vs pneumatic) because gear backlash differs by 20-50 µm.
Also update feeder firmware—new versions often improve advance accuracy on ultra-thin 0.15 mm carrier tape used for 01005 components.
3.3 Feeder Bank Power or Communication Loss
3.3.1 What Causes Bank Failure
The feeder base supplies 24 V and CAN-bus communication through spring-loaded pogo pins.
Oxide buildup on the pins, a bent docking rail, or a single loose M3 screw on the bank causes intermittent dropouts.
On ASM Siplace X-series you can lose an entire 40-slot bank if the internal 24 V fuse blows from a shorted feeder motor.
Panasonic NPM-W2 banks sometimes reset because the fiber-optic communication cable gets pinched when the bank is slid in too fast.
3.3.2 Connector and Cable Checks
Undock the bank, clean every pogo pin with a fiberglass brush or DeoxIT contact cleaner, and check the docking rail parallelism with a 0.02 mm feeler gauge—any gap over 0.1 mm causes flaky contact.
On Yamaha Sigma machines, inspect the flat ribbon cable at the hinge; it fatigues and cracks after 10,000 open/close cycles.
Replace with the updated shielded version if you see green corrosion.
3.3.3 Reset Procedures
Go to the feeder maintenance screen and select “reset bank power” or “re-initialize CAN-bus”—this forces the machine to re-enumerate every feeder ID without a full reboot.
For ASM machines, pull the 24 V breaker on the bank for 10 seconds; it clears phantom short-circuit protection.
Keep a spare bank power module on the shelf; swapping takes 5 minutes and gets you running while you troubleshoot the bad one offline.
4. Vision System and Alignment Errors
Vision errors turn a perfectly good pick and place run into chaos—parts rotated 90 degrees, placed on the wrong pad, or the machine hunting forever for fiducials.
Whether you’re on Fuji NXT cameras, Yamaha 3D vision, ASM multi-angle systems, Panasonic high-res, or JUKI coaxial setups, these problems come from the same core issues: contrast, lighting, teaching quality, and calibration drift.
Here’s the in-depth breakdown so you can nail 99.99% recognition rates every time.
4.1 Component Mis-Pick or Wrong Orientation
4.1.1 Why It Occurs
The camera sees the part, but the shape-matching algorithm gets confused by faint polarity marks, reflective package bodies, or background solder paste shining through the tape pocket.
On 0201 and smaller, the part can physically flip in the pocket from static or vibration, but the vision system still “recognizes” it upside-down because the threshold is too forgiving.
Shiny QFN thermal pads or transparent molded bodies (common on LEDs) create glare that washes out edges entirely.
4.1.2 Early Detection
Watch the live vision preview window—if the overlay box jumps around or the score is below 70 while the part looks perfect to your eye, you’re borderline.
Also track “orientation error” or “shape NG” counters per feeder; a sudden spike on one reel means teaching or lighting changed.
4.1.3 Fast Fixes
Re-teach the part with manual threshold adjustment: lower brightness threshold 5-10 points for dark marks and raise edge contrast.
Switch to side-angle lighting (Fuji ring light + Yamaha multi-LED) to kill reflections.
For transparent packages, enable “transmitted light” mode on Panasonic NPM or ASM’s bottom illumination—it shines through the tape and makes the body silhouette crystal clear.
4.2 Vision System Alignment Errors
4.2.1 Camera Issues
Dust on the telecentric lens, condensation inside the prism (common in humid factories), or the calibration glass plate scratched from years of cleaning.
On high-speed heads the camera shutter speed is fixed at 1/10000 s; any vibration above 0.5 G blurs the image and drops recognition score.
4.2.2 Recalibration Steps
Clean the lens with lens tissue and 99% IPA only—never cloth.
Place the factory calibration dot plate (Fuji CP/GL, Yamaha standard glass) exactly under each camera and run full camera calibration routine; it measures pixel size, distortion, and beam offset simultaneously.
Takes 3-5 minutes on modern machines and fixes 95% of “camera alignment error” alarms.
4.2.3 Lighting Optimization
Most machines let you program four independent lighting zones (coaxial, high-angle ring, low-angle ring, side).
For matte parts use high-angle only; for shiny ICs combine coaxial + low-angle to eliminate specular reflection.
ASM SIPLACE Pro lets you save lighting recipes per component family—set once and forget.
4.3 Fiducial Recognition Failure
4.3.1 Why Fiducials Fail
Solder mask covering half the copper pad, oxidized copper turning dark, or the mark is too close to panel edge and warpage lifts it out of focus.
Global fiducials work great until the panel bows 0.8 mm and the camera depth-of-field (usually ±0.5 mm) can’t cope.
4.3.2 Cleaning and Lighting
Wipe every fiducial with IPA before every job—flux overspray kills contrast fast.
Increase search area brightness threshold to 80% and enable “edge enhancement” filter.
For oxidized boards, switch to green or red LED lighting instead of white; copper reflects those wavelengths better.
4.3.3 Teaching New Fiducials
Always teach at least three global fiducials in a triangle covering 80% of the panel—never rely on two.
Use the “auto threshold” then manually tighten the acceptance window to ±15 gray levels.
On Yamaha YSM and Fuji AIMEX enable “bad mark skip” so one dirty fiducial doesn’t stop the whole board; it automatically uses the other two for correction.
4.4 Component Package Recognition Issues
4.4.1 Unknown Package Errors
New reel from a different tape vendor has slightly different pocket depth or embossed logo that shadows the part.
Or the component body color changed (black to gray mold compound) and the original teaching no longer matches.
BGA balls reflecting light create false edges that confuse the algorithm.
4.4.2 Teaching New Packages
Place five good parts from the reel on the teaching stage under perfect lighting.
Use “multi-sample teaching” (available on ASM, Panasonic, and latest JUKI) so the system learns natural variation.
Manually adjust the inspection window to ignore the balls on BGA and focus only on the package outline.
Save with a unique name like “QFP100_1.0mm_Gray_2025” so you never mix libraries.
4.4.3 Library Management
Centralize your vision library on a network server—every machine pulls the same file.
Version control is critical; keep old versions because a supplier might switch back to the original mold compound without telling you.
Run “library verification” routine monthly (Fuji and Yamaha have it) to flag shapes with recognition score below 85% in production logs.
5. Placement Accuracy and Motion Control Problems
Even if your nozzles pick perfectly and vision sees everything, the part still ends up off-pad because the motion system—the linear motors, encoders, ballscrews, and gantry—has drifted, worn, or been pushed beyond its sweet spot.
These issues hit hardest on high-mix lines running Fuji NXT III, Yamaha YSM20, ASM SIPLACE TX, Panasonic NPM, and JUKI RX-7 where you switch between 0201 and heavy connectors in the same program.
Here’s the deep technical dive so you can keep placement Cpk above 1.67 all shift long.
5.1 PCB Warpage Causing Placement Offset
5.1.1 How It Affects Accuracy
A 300 × 400 mm panel that warps 1 mm concave or convex moves the pad surface ±0.5 mm in Z and tilts it up to 0.8° at the corners.
Standard vision correction assumes a flat plane, so the calculated X-Y compensation is wrong by 50-120 µm on fine-pitch parts near the edge.
Thinner substrates (0.6 mm and below) and high-density panels with uneven copper distribution are the worst offenders.
5.1.2 Temporary Workarounds
Enable dynamic height mapping on Yamaha YSM20/YSM40R and Fuji AIMEX III—it uses the down-looking camera to measure actual board height at 9-25 points before placement and adjusts Z in real time.
On machines without it (older ASM TX1/TX2), add extra local fiducials in each corner and force the software to use four-point correction instead of two.
5.1.3 Long-Term Solutions
Use dedicated warpage-support tooling with adjustable magnetic pins every 50 mm, pre-bake panels at 125 °C for 2 hours to relieve stress before printing, and balance copper pour on inner layers.
Panasonic’s APC-3D system scans the entire board with a laser profiler before pickup and creates a true 3D surface map—cuts edge offset from 80 µm to under 15 µm on warped automotive boards.
5.2 Overthrow or Underthrow Defects
5.2.1 Why Parts Land Off-Pad
Linear motor acceleration and jerk settings are too aggressive for the head mass + component weight.
At 1.5-2.0 G acceleration (common on 100 k+ cph machines), a 0.3 g connector overshoots 40-80 µm when the stage stops abruptly.
Encoder scaling errors or backlash in the X-Y beam coupling add another 10-25 µm systematic offset.
5.2.2 XY Fine Tuning
Run the built-in placement accuracy test with a glass calibration plate and fiducial marks every 10 mm.
Reduce acceleration from max (usually 20 m/s²) down in 10% steps until the measured CpK hits 2.0.
On dual-gantry machines, match left and right beam scaling to within 1 µm/m—Fuji and ASM have a specific “beam matching” routine that takes 8 minutes and is worth doing weekly.
5.2.3 Beam Calibration
For twin-beam systems (Hanwha Decan, ASM TX2+), place a long glass plate across both beams and run cross-beam calibration.
It measures and compensates mechanical misalignment between the two independent gantries. Ignore this and you’ll see a 30-50 µm step exactly in the middle of large panels.
5.3 Z-Axis Height or Crash Errors
5.3.1 Crash Root Causes
Z height data in the program is wrong (coplanarity not measured), previous process left solder balls or flux bumps taller than expected, or the nozzle length value drifted after a tip change.
On high-speed heads the Z servo can overshoot 0.3 mm if the PID gains are too aggressive for heavy nozzles.
5.3.2 Placement Force Adjustment
Set placement force individually per component: 1-2 N for 0201/01005, 3-5 N for 0402-0603, 8-12 N for large QFP/BGA.
Use the “force control” mode on Yamaha Sigma and Panasonic NPM—it actively monitors and limits force in real time instead of blind position stop.
Drop default force 20% and you eliminate 95% of cracked MLCC issues.
5.3.3 Safety Limits Setup
Program a Z-safe height 2-3 mm above the tallest measured component + 0.5 mm margin.
Enable “touch-down detection” on JUKI and Fuji heads—the nozzle senses contact via force sensor and stops instantly instead of driving to absolute position.
Also run nozzle length auto-measure after every nozzle change; it takes 15 seconds and prevents crashes cold.
5.4 Nozzle Centering or Offset Drift
5.4.1 Identifying Drift
Placement offset slowly walks 20-60 µm over 4-8 hours, usually in one direction.
Caused by thermal expansion of the head casting (aluminum expands 23 ppm/°C) or linear encoder scale contamination.
5.4.2 Auto Offset Routine
Run nozzle center calibration at the beginning of every shift and again after 4 hours on hot days.
Modern machines (Fuji NXT III M3/M6 heads, Yamaha HSD, ASM CPP) do it fully automatic with an upward-looking camera and take under 40 seconds for 12 nozzles.
The routine measures X-Y offset and nozzle rotation (theta) for every single nozzle and writes it to the head memory.
5.4.3 Daily Warm-Up
After morning power-on, let the machine run a 10-minute idle cycle with heads moving full stroke at 50% speed—no parts.
This stabilizes the linear motor magnets and glass scale temperature.
Factories that skip this see 30-50 µm drift in the first two hours; factories that do it hold ±10 µm all day.
6. Line Stability, Power, and Environmental Issues
Even the best nozzles, feeders, and vision systems fall apart when air drops, power flickers, or the room turns into a dust storm—these “invisible” problems cause random stops that drive engineers crazy on Fuji, Yamaha, ASM, Panasonic, and JUKI lines.
Here’s the deep technical detail so you can bulletproof your entire placement machine environment and keep it running 24/7.
6.1 Air Pressure Fluctuations
6.1.1 Impact on Operation
The vacuum ejector (Venturi principle) needs steady 0.5-0.6 MPa dry air to generate -80 to -90 kPa vacuum.
A 0.05 MPa drop during a plant-wide compressor surge collapses vacuum across all nozzles simultaneously—every pick fails until pressure recovers.
On high-speed heads with 12-16 nozzles, the total air consumption can spike to 400-600 L/min during full placement; if your factory main line is undersized or has a leaking quick-connect somewhere upstream, you’ll see cyclic vacuum decay every time another machine starts.
6.1.2 Regulator Checks
Install a dedicated 3/8″ or 1/2″ line from the FRL directly to the machine with its own 5-micron filter and auto-drain.
Set the secondary regulator to 0.55 MPa and lock it—most machines have a built-in pressure sensor that alarms below 0.45 MPa, but you want headroom.
Add a 20-40 liter buffer tank right at the machine inlet; it smooths out 2-3 second compressor hiccups and prevents 90% of pressure-related stops.
6.1.3 Backup Air Supply
High-volume factories run a ring main with dual compressors and auto-switchover, plus a small 50-liter emergency tank with check valve that kicks in if main pressure drops below 0.4 MPa.
Yamaha and Panasonic machines can be optioned with electric vacuum pumps (scroll or rotary vane) that eliminate plant air dependency entirely—worth the investment if your factory air is unreliable or contaminated with oil.
6.2 Cooling Fan Failure or Overheating
6.2.1 Overheat Symptoms
Linear motor drivers, servo amps, and vision CPUs generate 500-1500 W heat depending on speed. When intake filters clog or fans slow down, internal temperature rises above 65-70 °C and the machine throws “head overheat” or “driver fault” and pauses.
On ASM SIPLACE TX and Fuji NXT III, the head has its own temperature sensor that forces slowdown at 60 °C and full stop at 70 °C to protect the linear motors.
6.2.2 Fan Cleaning
Vacuum all fan grilles and replace intake filters every month—use genuine Fuji/Yamaha 0.3-micron pleated filters, not cheap knockoffs that collapse and starve airflow.
Check fan RPM in the diagnostic screen (most brands show live speed); anything below 80% of spec means bearing wear—replace the entire fan assembly because bearings alone are not serviceable on modern brushless units.
6.2.3 Temperature Monitoring
Log internal temperatures hourly via the machine’s Ethernet port—set custom alarms at 55 °C instead of waiting for the factory 65 °C limit.
Some factories add external PT100 sensors on the head casting and feed them into the PLC so the line stops before the machine protects itself.
Keep room temperature 22-25 °C and humidity 40-60% RH; every 10 °C rise in ambient cuts linear motor life roughly in half.
6.3 ESD Damage to Components
6.3.1 Why ESD Happens
Triboelectric charging from dry air (<30% RH) or non-conductive feeder plastic generates thousands of volts on the nozzle tip.
A 100 V discharge through a 01005 resistor with 1-2 nm gate oxide destroys it instantly, but you won’t see the failure until final test—classic “walking wounded” problem.
Static also causes parts to jump sideways during transfer or stick to the nozzle side wall.
6.3.2 Grounding Checks
Measure resistance from every nozzle tip to the main ground stud—must be <2 ohms on Fuji/Yamaha, <10 ohms on ASM/JUKI.
Use a proper ESD surface resistance meter, not a cheap multimeter.
All carbon-impregnated nozzles and conductive feeder rails must be bonded; a single loose screw breaks the chain. Install wrist-strap monitors that alarm if resistance exceeds 35 MΩ.
6.3.3 Adding Ionizers
Overhead DC ionizer bars (Simco-Ion, Keyence, or SMC) with auto-cleaning emitters neutralize charge in <1 second across the entire feeder bank and placement area.
Set airflow so the ionization cloud reaches the PCB surface—test with a charged-plate monitor aiming for decay time <2 seconds from 1000 V to 100 V.
Combine with conductive floor mats and heel grounders for operators; factories that do this drop ESD-related yield loss from 500 ppm to under 20 ppm on fine-pitch parts.
7. Software, Program, and Control Errors
Software glitches and bad programs are the most frustrating stops because the hardware is fine—the machine just refuses to run.
These issues hit every brand (Fuji NXT, Yamaha YSM, ASM SIPLACE Pro, Panasonic NPM, JUKI RX series) when CAD data is dirty, parameters drift, or control hardware ages.
Here’s the deep dive so you can fix them fast and prevent them forever.
7.1 Software or Program Loading Errors
7.1.1 Common File Issues
The machine rejects the job because the centroid file has incorrect rotation (90° instead of 0°), missing package IDs, duplicate feeder slots, or the file was exported with wrong units (mils vs mm).
On older Fuji Flexa systems you’ll get “format error” if the CSV has extra columns from a new Valor version.
Corrupted USB sticks or network timeouts during transfer create partial files that crash the loader halfway.
7.1.2 CAD Import Fixes
Always export with these settings: origin at lower-left, rotation counter-clockwise positive, polarity mark on pin 1, and no mirrored components unless it’s bottom side.
Use the machine’s built-in validation tool (Yamaha “data check”, ASM “integrity check”) before loading—it flags 95% of issues in 30 seconds.
For ODB++ files, enable “strict mode” on Panasonic and JUKI; it rejects anything with overlapping apertures or missing layers.
7.1.3 Backup Procedures
Export the complete program (shape library + feeder setup + optimization data) to a dated folder on the server after every successful first article.
Keep the last five versions plus one “golden” known-good backup.
Enable auto-backup on the machine (Fuji and Yamaha can push to FTP every night).
When a job corrupts, revert to the previous version and diff the two files in Notepad++ to see exactly what changed.
7.2 Backup Battery or Memory Loss
7.2.1 Symptoms
After weekend power-off the machine forgets all calibration values, nozzle data, feeder offsets, and sometimes even the IP address.
You’ll see “parameter lost” or “SRAM clear” alarm on boot.
Happens when the 3 V lithium coin cell (CR2032 or BR2032) drops below 2.7 V—usually after 5-7 years.
7.2.2 Battery Replacement
Power down completely, open the control cabinet, locate the battery on the main CPU board (usually marked BT1).
Replace while holding the “backup capacitor” button if the machine has one (ASM and Panasonic do) to keep SRAM alive for 10-15 minutes.
After new battery, reload the last full parameter backup from USB—takes 3 minutes and restores everything including vision library versions.
7.2.3 Data Backup Tips
Schedule weekly full parameter export via Ethernet to a shared drive. Include the hidden “machine constant” file that stores beam scaling and encoder resolution—without it you lose placement accuracy after battery change.
Some factories add a small UPS to the control rack so the machine never fully loses power during factory shutdowns.
7.3 Head Motor or Encoder Faults
7.3.1 Erratic Motion
The head jerks, stops mid-travel, or throws “encoder abnormal” because contamination on the linear scale (glass strip with 20 nm lines) creates false counts, or the linear motor coil has partial short from burnt insulation.
On Yamaha HSD and Fuji M6 heads, the read-head LED degrades after 40,000 hours and signal-to-noise drops below spec.
7.3.2 Basic Maintenance
Clean the linear scales with anhydrous alcohol and lint-free cloth every 6 months—never use compressed air alone because it drives dust deeper.
Re-grease the linear guides with Klüber Isoflex NBU 15 (Fuji/ASM spec) or Yamaha’s equivalent—too much grease attracts dust and causes encoder noise.
Check encoder waveform with the diagnostic oscilloscope screen (most machines have it); clean signal should be perfect sine wave 2-3 V p-p.
7.3.3 When to Call OEM Support
If cleaning and reboot don’t fix it and the error code points to specific axis (e.g., Fuji “Z encoder phase error”, ASM “linear motor overcurrent”), you need factory service.
They’ll replace the entire read-head assembly or motor coil—parts are expensive but downtime is worse.
Keep spare encoder cables on the shelf; flex fatigue at the head connector is the number one false alarm on machines over 3 years old.
8. Process-Driven Placement Defects
Some defects only show up after reflow, even when pick and vision looked perfect—these are pure process interactions between your placement settings, paste volume, and oven profile on Fuji, Yamaha, ASM, Panasonic, and JUKI machines.
Fix them here and you’ll stop chasing ghosts in AOI and ICT. Here’s the detailed engineering breakdown so you can eliminate them at the source.
8.1 Component Tombstoning After Placement
8.1.1 Causes
Uneven wetting forces pull one end of the chip up during reflow—the side with more solder or faster heat rise lifts first.
Root triggers include unbalanced pad design (one pad larger or more paste), placement force pushing paste out one side only, delayed Z-up causing the nozzle to drag on wet paste, or preheat too aggressive so flux volatiles explode under one pad.
On 0201 and smaller, surface tension alone can flip the part if the centroid is off by even 20 µm.
8.1.2 PnP Settings
Drop placement force to 0.8-1.5 N and slow Z-down speed to 20-30 mm/s so the part settles evenly without squeezing paste sideways.
Enable “soft landing” or “constant force mode” on Yamaha Sigma, Fuji NXT III, and Panasonic NPM—it holds light pressure for 50-100 ms after contact so both ends wet the paste equally.
Also reduce vacuum release delay by 20 ms; too long and the nozzle pulls one end up as it lifts.
8.1.3 Reflow Adjustments
Extend soak zone 20-40 seconds longer and keep ramp no steeper than 1.5 °C/s—gives flux time to activate evenly on both pads.
Use linear ramp-to-peak profiles instead of plateau on ASM and JUKI ovens for micro-chip-heavy boards. Pad design fix: make both pads identical size and add solder thief extensions on the exit pad if your stencil aperture is oversized on one side.
8.2 Placement Speed vs. Accuracy Issues
8.2.1 Speed Limits
Running full theoretical speed (80-120k cph) with mixed component sizes causes dynamic errors: heavy parts lag during acceleration and land late, light parts overshoot from inertia when the head stops.
Vision processing can’t keep up on single-camera heads above 90k cph when fine-pitch and large parts alternate—recognition score drops and throw errors skyrocket.
8.2.2 Optimized Settings
Create component-specific speed classes: 100% speed for 0603 and larger, 70-80% for 0402, 50-60% for 0201/01005 and below.
Use “dynamic speed control” on Yamaha YSM20 and Fuji AIMEX III—it automatically slows the head when placing small or heavy parts while keeping big resistors at full speed.
Also enable “look-ahead optimization” so the head decelerates early for the next fine-pitch placement instead of slamming to a stop.
8.2.3 Speed Profiles for Different Parts
Most modern software (Panasonic PanaCIM, ASM SIPLACE Pro, JUKI JaNets) lets you assign velocity/ acceleration per package family.
Set separate profiles: high-speed profile for passives >0402, precision profile for ICs and connectors, and ultra-slow for odd-form.
Factories that do this push overall line rate 15-25% higher while keeping defect rate under 50 ppm—even on mixed automotive boards with 01005 next to 50 g connectors.
9. Emergency, Safety, and Unexpected Line Stops
Random full-line stops with no obvious alarm drive everyone crazy—these are usually safety interlocks, sensor ghosts, or conveyor glitches on Fuji NXT, Yamaha YSM, ASM SIPLACE, Panasonic NPM, and JUKI machines.
Here’s the deep technical breakdown so you can trace and eliminate them permanently instead of hitting resume fifty times a shift.
9.1 Emergency Stop Triggered Randomly
9.1.1 Sensor Issues
The E-stop circuit is a series loop through door switches, light curtains, area scanners, and pressure mats—any intermittent contact trips the whole machine.
Common culprits: dust buildup on light curtain emitters/receivers dropping signal below 70%, loose magnets on safety doors (gap >0.5 mm loses Hall effect), or vibration cracking solder joints on the E-stop relay board itself.
On Yamaha and Fuji machines the safety PLC scans every 10 ms; even a 50 ms spike from EMI on unshielded cables registers as a fault.
9.1.2 Safety Checks
Clean every light curtain lens with IPA and check alignment with the built-in LED bar—must be solid green.
Measure door switch resistance (should be <1 ohm closed) and magnet gap with a feeler gauge.
Replace old mechanical switches with non-contact magnetic coded ones (Pilz PSEN or Sick equivalents)—they never wear and ignore dust.
Run the safety diagnostic screen (Panasonic and ASM show live status for every input) to see exactly which sensor is flickering.
9.1.3 Relay Troubleshooting
The dual-channel safety relay (usually Omron G9SE or Pilz PNOZ) has forced-guided contacts—when one channel sticks from burnt contacts the mismatch trips the system.
Swap the relay module (5-minute job on most machines) or measure voltage drop across each channel with the door closed—anything over 0.5 V means replace.
Upgrade to electronic safety controllers (Fuji MICREX-SX, Yamaha own module) that self-diagnose and tell you the exact failed channel instead of just “E-stop activated”.
9.2 Conveyor Belt Slippage or Misalignment
9.2.1 Symptoms
Board stops halfway, edge belts squeal, or width adjustment motors keep running after reaching limit switches.
Happens when belts stretch after 12-18 months, pulleys wear oval, or dried lubricant turns to glue on the rail slides.
9.2.2 Rail and Belt Adjustments
Loosen the four rail locking screws, realign front and rear rails parallel within 0.1 mm using a precision straight edge and dial indicator, then re-tension belts to factory spec (Fuji 80-100 N, Yamaha 90 N measured with a belt tension meter).
Replace both belts at the same time—mixing old and new causes 0.3-0.5 mm skew.
On ASM and Panasonic machines check the timing belt that drives the width ballscrew—worn teeth jump and give random width errors.
9.2.3 Width Calibration
Run full conveyor auto-calibration with a calibration jig or glass board—takes 2 minutes and writes new encoder offsets.
Do this every time you change rail width more than 50 mm or after belt replacement.
Enable “belt slip detection” on JUKI RX-7 and Yamaha YSM20—it monitors encoder vs motor pulses and stops the line before the board crashes into a fixed stopper.
Keep spare belt sets pre-cut to your three most common widths; swapping takes 15 minutes and beats fighting slippage all shift.
