Menu

The Profile That Doesn't Tombstone

The Profile That Doesn't Tombstone

Notes · Circuit boards and EMCNovember 19, 202617 min readSenova Systems Research

Your boards come back with tombstoned parts, cold joints, and voids, and the assembler says the profile is fine, because the oven's setpoints are fine and the board's temperatures are not. Here is the reflow profile as four zones, measured on the board, with the numbers.

Picture the batch. Two hundred boards come back from the assembler, and six to ten of them have a part standing on one end, a little resistor or capacitor, lifted off one pad, standing upright like a gravestone. The assembler says the profile is fine. The oven's setpoints are fine. The oven's setpoints are not the profile. The profile is what the board actually felt, measured with thermocouples on the board itself, and the board's temperatures are not the oven's temperatures.

A reflow profile is a curve: temperature against time, four zones, lasting four to seven minutes. Every solder joint on the board, from an 0201 passive to a 1,500-ball BGA, forms simultaneously during that one curve. Get it right and every joint forms.

Get it wrong and the same curve that makes the joints can tombstone the small parts, crack the ceramic ones, warp the board, and leave voids inside the joints you cannot see. This article is about the curve, the four zones, the numbers each one has to hit, and the one defect that tells you the curve is wrong.

The four zones

A standard reflow profile has four zones, and each one has a metallurgical job. The numbers below are consistent across four sources, LTP CBA, AtlasPCB, JLCPCB, and the BME course notes, and the article gives the range and the recommended value every time, so you can see where the window is and where the middle of it is.

Zone 1, Preheat: 25 °C to 150 °C, ramp 1–3 °C/s (recommended 1.5–2.5 °C/s), typically 60–90 seconds. The board goes from ambient to the flux activation range. The ramp rate is the number that matters, and it is the tightest number in the whole profile. "Run the preheat ramp at 1–3°C/s. Faster ramps splatter paste and crack ceramic parts." (LTP CBA) A ramp above 3 °C/s builds thermal gradients across the board, and four failures follow: paste splatters as volatiles flash off; the board warps from uneven expansion; ceramic capacitors and BGAs crack; small passives at 0201 and 0402 tombstone from unequal pad heating. A ramp below 1 °C/s brings its own cost: the profile stretches out, flux burns off before the reflow zone, and throughput drops. JLCPCB's troubleshooting guide puts it the same way: too fast means thermal shock and tombstoning; too slow means paste slump and bridging.

Zone 2, Soak / thermal equalization: 150–200 °C for 60–120 seconds (recommended 160–190 °C, 75–90 seconds). The soak is the most critical part of the curve and the most often misunderstood. It does three jobs at once. First, flux activation, the flux chemistry starts reducing copper oxides on the pad and tin oxides on the solder particles, and that reaction needs both time and heat. Second, thermal equalization, the soak lets board temperatures converge, so every joint enters reflow inside a narrow window. Third, volatile evacuation, flux solvents and activator byproducts must leave during the soak; trap them until peak and they form voids inside the joint, and BGA packages concentrate stress at those voids. One common error is running the soak too hot, above 200 °C: flux activators burn out early, and too little remains to wet the pads at peak. The ramp inside the soak is gentle, 0.3–0.5 °C/s recommended, 1 °C/s maximum.

Zone 3, Reflow: above 217 °C, time above liquidus 40–90 seconds, peak 245–250 °C. This is where the solder melts and the metallurgical bond forms. Two numbers define it. The peak temperature is the highest reading on the BGA joint. For SAC305, Sn96.5/Ag3.0/Cu0.5, the alloy modern manufacturing relies on almost exclusively, the liquidus is 217–220 °C; the minimum peak is 235 °C (so all joints exceed liquidus even with the temperature spread across the board); the recommended peak is 245–250 °C; the maximum peak is 260 °C (the MSL-3 limit per J-STD-020). The peak has to clear the liquidus by enough to wet the joint fully and grow intermetallic compound, the thin layer of Cu₆Sn₅ or Ni₃Sn₄ that is the actual bond. A healthy IMC layer is 1–3 microns. The time above liquidus (TAL) is the span where the joint sits above 217 °C: minimum 30 seconds (the absolute minimum for IMC formation), recommended 40–70 seconds, maximum 90 seconds, beyond that the IMC grows excessively thick, and at 110 seconds or beyond it grows thick enough to weaken the joint. For BGAs, 60–90 seconds is preferred, because the solder balls under the package heat slower than the exposed joints. Lead-free made all of this harder: SAC305 melts at 217–220 °C against 183 °C for eutectic tin-lead, a 34 °C increase that pushed peak temperatures dangerously close to the thermal limits of the components and the board.

Zone 4, Cooling: peak down to below 100 °C, 2–6 °C/s (JLCPCB 2–4 °C/s). Cooling sets the microstructure of the solid joint, and that structure drives its mechanical behaviour. Fast cooling at 4–6 °C/s gives a fine-grained SAC structure. Slow cooling below 2 °C/s gives the opposite: large Ag₃Sn platelets form and act as crack starters, fatigue life drops, and thermal cycling failures rise. Too fast is no better, above 6 °C/s risks thermal shock, component cracking, and pad cratering. The cooling rate is the zone most teams never measure, and it is the one that decides how long the joint lasts in the field.

Measure the board, not the oven

The profile is measured with thermocouples attached to the board at critical locations, "typically the coldest point (usually the center of the largest ground plane), the hottest point (usually a small component near the board edge), and on BGA solder joints themselves." (AtlasPCB) The placement, in full: a K-type thermocouple (36 AWG) on a corner pad of the largest BGA, attached with high-temperature solder or Kapton tape; one on the board centre, on the largest ground plane area (the coldest point); one at the board edge, near a small component (the hottest point); and one in the connector area, if through-hole parts will see wave or selective soldering later.

AtlasPCB's optimisation guide wants a minimum of five thermocouples, preferably seven to nine, the hottest expected location, the coldest expected location, the most expensive or failure-prone part, the board centre, the board edge, under the large QFN or BGA, and near the temperature-sensitive component.

The number that comes out of those thermocouples is the temperature difference between the hottest and coldest points, ΔT, and it is the primary metric for profile optimisation. For BGA boards the limits are: standard, less than 15 °C; fine-pitch BGA, less than 10 °C; high-reliability, less than 7 °C. The gap is real and it is large: "The temperature difference between a 0402 capacitor on a signal layer and a 1156-ball BGA on a ground plane can exceed 30°C during ramp, but you need both to reach peak within 10°C of each other."

(AtlasPCB) The soak is what closes that gap. A board that goes into reflow with a 30 °C spread across it does not have one profile; it has two, and the cold end gets cold joints while the hot end gets damaged components. Do not approve a profile based only on the oven's centreline temperature, the supplier should attach thermocouples to the coldest heavy thermal mass and the hottest heat-sensitive package.

(YourPCB) The profile is release evidence: freeze it before pilot approval, not after the first production defect.

Tombstoning: the tug-of-war

Tombstoning, also called drawbridging, or the Manhattan effect, is the defect where a chip component stands up on one end during the reflow cycle, like a gravestone. It is one of the most common failure issues in SMT reflow soldering of miniature surface-mount components such as 0201 and 0402 packages, and the core cause is the imbalance of solder joint forces at both ends of the component. (LeadsIntec)

The mechanism is simple enough to draw on a napkin. During reflow, the solder paste under each end of the component melts and starts wetting to the termination. If the paste under one termination melts and wets before the other, the wetting force pulls the component up on its end.

The force is surface tension, the same force that pulls a slightly misplaced component to the centre of its pads, which is a benefit on a big part and a defect on a small one. "As the flux and solder alloy liquefy and wet to each side of a component, they apply small amounts of torque through surface tension... tombstoning is caused by minute differences in wetting force from one side of a component to the other.

When there is a sufficient imbalance in torque, relative to the mass of the component, the component is tipped upright." (I-Connect007) The smaller and lighter the component, the more susceptible it is. "Chip components have gotten much smaller over the years with many weighing in the milligrams.

The same torque that once helped alignment now has the power to tombstone components."

The numbers make the point sharper than the words. Measured data: for a 0603 MLCC component (terminal electrode 0.3 mm × 0.15 mm), when the pad length on one side is 0.08 mm longer than the other (0.45 mm to 0.53 mm), under conventional SAC solder paste conditions (surface tension γ ≈ 0.62 N/m, wetting angle θ ≈ 30°), the surface-tension difference between both ends can reach 12–15 µN. "This force difference is more than three times the self-weight of the 0603 component (approximately 3.2µN), completely overcoming the component's gravitational resistance and triggering component rotation and tombstoning."

(LeadsIntec) Eight hundredths of a millimetre of pad, less than the width of a human hair, produces a force nearly five times the part's own weight. That is why pad design is a tombstoning cause, not a detail: a pad that is not per the specification for the package deposits excess paste, and excess paste is the precursor to tombstoning. When a 0402 is placed on pads sized for a 0603, the solder volume is greater than required, and the part tombstones.

The other causes all work the same way, by making one end wet before the other. Uneven heating: if the thermal mass distribution is uneven, one pad heats first and the second pad heats a few seconds later, and the surface tension pulls on the side that melted first. A pad sitting over a huge copper plane is like attaching it to a giant ice cube, that pad stays cooler longer, the partner pad heats up fast, melts quickly, and develops strong surface tension, and the imbalance can flip the little part on its end before the cooler pad even gets warm.

(South Electronic PCB) The fix is often to break that copper pour connection with thermal spokes. Large pad-to-pad distance, thick soldermask acting as a fulcrum under the part, paste tackiness that cannot hold the part while one end melts, all of them are the same defect wearing a different cause. And the reflow profile itself is one of the causes: a heating ramp that is too fast makes one pad reach the liquidus before the other, and JLCPCB's troubleshooting guide gives the fix directly, reduce the initial ramp rate to approximately 1.5 °C/s, and extend the soak zone to achieve thermal equilibrium across the entire board.

The 0201 study that settled which causes matter most is the SMTA International 2007 paper, "A Study of 0201s and Tombstoning in Pb-free Systems" (electronics.org). A review of published studies and high-volume production experience uncovered 49 potential parameters that can influence tombstone defects. The investigators used a cause-and-effect matrix (rating influence 0, 3, or 9) to identify the 12 most influential: padstacks, solder paste volume, print offsets, component type, orientation, reflow profile, reflow atmosphere, placement offset, solder paste, and three more.

The largest contributor to defect generation was the combination of 0.1 mm print and placement offsets, a tenth of a millimetre of mechanical slop. The next three, in descending order, were component type, orientation, and reflow atmosphere. And the payoff of removing the slop: the overall defect rates fell from 11,823 to 2,513 ppm, a 78.7% reduction, and from 7,845 ppm to 156 ppm excluding the solder ball count, a 98.0% reduction.

The profile is on the list of the twelve. It is not the largest contributor. The largest contributor is the print and the placement, and both are almost completely controllable at the assembly level.

The single documented case that shows what the defect costs in production is the root-cause report for an electric-engine inverter PCBA: tombstoning on 0402 and 0603 passive resistors, a defect rate of 3–5% per board and 75% per batch, with the impact stated plainly, open circuits, rework delays, reliability. The PFMEA scored the reflow step S9 O4 D4, RPN 144, severity 9, occurrence 4, detection 4. The corrective actions were a ramp of 1.5 °C/s and a peak of 245–250 °C, the middle of the window, not the edge of it.

One case, labelled as such. But it is the case that makes the arithmetic in the demo below concrete: three boards in every four batches, and every one of them reworked by hand.

Example Labs is a made-up company we use for worked examples.

Example Labs' sensor hub board, through the oven

Example Labs' sensor hub board, 100 × 60 mm, 4-layer, 120 nets, 114 test points (from Article #32), goes through the oven on a SAC305 profile. The board carries 0402 and 0603 passives, a BGA, and a ground plane that covers most of the board. Here is the profile, zone by zone, with the arithmetic run in Python before writing.

Step 1, the preheat. The board has to rise from 25 °C to 150 °C, a 125 °C climb. At the demo's ramp of 2.0 °C/s, that takes 62.5 seconds, inside the typical 60–90 second window. At the recommended 1.5 °C/s it takes 83.3 seconds, also inside. At 3.5 °C/s, above the 3 °C/s limit, it takes 35.7 seconds, and the paste splatters, the board warps, the ceramics crack, and the 0201s and 0402s tombstone. At 0.8 °C/s, below the 1 °C/s floor, it takes 156.2 seconds, and the flux burns off before the reflow zone. The window is 1–3 °C/s, and the demo sits at 2.0 °C/s, in the middle of it.

Step 2, the soak, and the 30 °C gap. During the ramp, the 0402 capacitor on a signal layer and the BGA on the ground plane differ by more than 30 °C (AtlasPCB). At peak they have to be within 10 °C of each other (fine-pitch BGA). The soak has to close 20 °C of gap. The demo soaks at 160–190 °C for 90 seconds, the recommended 75–90 second window, with a ramp inside the soak of 0.3–0.5 °C/s. A soak above 200 °C would burn out the flux activators early; a soak shorter than 60 seconds would leave the volatiles trapped, and they would form voids inside the joints at peak.

Step 3, the reflow, and the tombstone. The peak is 247 °C, inside the recommended 245–250 °C, 27–30 °C above the 217–220 °C liquidus, inside JLCPCB's 20–30 °C above the melting point. The time above liquidus is 60 seconds, inside the 40–90 second window, and the 60–90 second preference for BGAs. Now the tombstone: a 0603 decoupling capacitor next to the BGA has one pad over the ground plane (cold) and one over a signal trace (hot). The pad on the hot side is 0.08 mm longer than the other, 0.53 mm against 0.45 mm, a layout error of less than a hair's width. The surface-tension difference is 12–15 µN against the part's 3.2 µN self-weight, 3.75 to 4.69 times the weight of the part. The hot end wets first, the force pulls, and the part stands up. The fix is not a different paste. It is the pad, the thermal spokes, and the ramp at 1.5 °C/s.

Step 4, the cooling, and the joint's life. From the 247 °C peak down to 150 °C is a 97 °C drop. At the demo's 3 °C/s, that takes 32 seconds. At 1 °C/s it would take 97 seconds, and the slow cooling would grow large Ag₃Sn platelets that act as crack starters, the fatigue life drops, and the thermal cycling failures rise. The total profile time is 4.1 minutes, preheat 62.5 seconds, soak 90, reflow about 60, cooling about 32, inside the 4–7 minute range and under the 7-minute limit.

Step 5, what the defect costs. The inverter case, applied to Example Labs' batch: a 200-board batch at 3–5% per board is 6–10 boards with tombstoned parts, and 75% of batches are affected, three batches in every four. At the article's own assumption of 15 minutes of manual rework per board, that is 2 hours of rework per 200-board batch, and the rework is the cheap part. The expensive part is the open circuit the test catches late, or the one it does not catch: Article #32's numbers are ICT at 95–98% coverage with 10–30 ppm escapes against visual-only at 40–60% coverage and 500+ ppm escapes, and a field fix costs 10–100 times a factory fix. The PFMEA said S9 O4 D4, RPN 144. The profile is the corrective action, a ramp of 1.5 °C/s and a peak of 245–250 °C, and it costs nothing, because the oven was already there.

The decision test

Five questions. If you can answer all five with a number measured on your board, the profile is a system. If you cannot, the oven's setpoints are doing the deciding.

  1. What is the ramp rate, measured on the board, not set on the oven? 1–3 °C/s, recommended 1.5–2.5 °C/s. Above 3 °C/s the paste splatters, the board warps, the ceramics crack, and the 0201s and 0402s tombstone. Below 1 °C/s the flux burns off before reflow. If the answer is the oven's setpoint, you do not know the ramp rate.
  2. What is the ΔT across the board at peak, and where are the thermocouples? Less than 15 °C standard, less than 10 °C for a fine-pitch BGA, less than 7 °C for high reliability. The thermocouples go on the coldest point (centre of the largest ground plane), the hottest point (a small component near the board edge), and the BGA joints themselves. If the profile was measured with one thermocouple in the middle of the board, the ΔT is unknown, and the cold end of the board is unprofiled.
  3. How long is the soak, and how hot? 150–200 °C for 60–120 seconds, recommended 160–190 °C for 75–90 seconds. Above 200 °C the flux activators burn out early. Below 60 seconds the volatiles stay trapped and form voids. The soak is the zone that closes the 30 °C gap between the 0402 and the BGA, if the soak is short, the gap survives into reflow.
  4. What is the peak, and what is the time above liquidus, on the coldest joint? Peak 245–250 °C, TAL 40–90 seconds (60–90 on BGAs). The minimum peak is 235 °C, and the coolest joints on the board should generally reach at least 228 °C, preferably 235 °C. If the peak is measured on the hottest joint, the coldest joint may never have cleared the liquidus, and a joint that never cleared the liquidus is a cold joint, not a joint.
  5. What is the cooling rate, and what does the joint look like? 2–6 °C/s, JLCPCB 2–4 °C/s. Below 2 °C/s the grain coarsens, large Ag₃Sn platelets form, and the fatigue life drops. Above 6 °C/s the thermal shock cracks components and craters pads. If nobody measures the cooling rate, nobody knows how long the joint will last in the field, and the field is where the profile is finally tested.

Objections

"The assembler says the profile is fine." The assembler is looking at the oven's setpoints. The profile is what the board felt. The two differ by the board's thermal mass, the copper distribution, and the airflow, and on a board with a large ground plane and a BGA, they differ by more than 30 °C during the ramp. Ask for the thermocouple data on the real board, at the coldest point, the hottest point, and the BGA joints. If the assembler cannot produce it, the profile has not been measured, and "fine" is a guess.

"We don't have tombstoning, we have cold joints." They are the same curve, failing at different ends. A cold joint is a joint that never cleared the liquidus, or never stayed above it long enough, the peak too low, or the TAL too short, or the coldest joint never reaching 235 °C. A tombstone is a joint where one end wet before the other, the ramp too fast, the soak too short, the pads asymmetric. Both are the profile. Fix the curve, and both go away. Fix one joint at a time, and neither does.

"The board passed the bench test." The bench test catches the open circuit. It does not catch the void inside the BGA joint, the IMC layer grown too thick by a 110-second TAL, or the coarse grain from a 1 °C/s cool-down. Those fail in the field, under thermal cycling, months later, and Article #40's post-mortem is where they end up. Article #32's numbers are the reason: ICT at 95–98% coverage with 10–30 ppm escapes, against visual-only at 40–60% and 500+ ppm. The profile is the cheapest reliability engineering there is, and it is decided before the board ever leaves the oven.

"Our boards are small, the profile doesn't matter for us." Small boards tombstone more, not less. The 0201 study's largest contributor was 0.1 mm of print and placement offset, and the smaller the part, the less mass resists the same wetting force. A 0603's self-weight is 3.2 µN against a 12–15 µN imbalance. An 0402 weighs less than a milligram. The smaller the part, the tighter the window, and the more the profile matters.

"We'll just run the oven hotter, that fixes everything." Hotter fixes the cold joints and breaks everything else. Above 260 °C you pass the MSL-3 limit. Above 250 °C on standard FR4 (Tg 130 °C) you risk delamination and warpage. Above 200 °C in the soak you burn out the flux activators. A peak of 245–250 °C is the middle of the window, not the edge of it, and the window is narrow because lead-free raised the liquidus 34 °C and left almost no margin between the solder melting and the components breaking.

Sources

  1. LTP CBA, "Reflow Oven Profiling: Zones, Curves, and Common Errors"
    What we used

    "Run the preheat ramp at 1–3°C/s. Faster ramps splatter paste and crack ceramic parts." "Hold the soak at 150–200°C for 60–120 seconds to even out thermal mass." "Target a peak of 245–250°C and never pass the component MSL rating." "Keep time above liquidus at 40–90 seconds for SAC305, with 60–90s on BGAs." "Cool at 2–6°C/s. Slow cooling coarsens the grain structure and shortens fatigue life." Preheat: "Target ramp rate: 1–3°C/s (recommended: 1.5–2.5°C/s). Duration: Typically 60–90 seconds." Soak: temperature 150 °C minimum, 160–190 °C recommended, 200 °C maximum; duration 60 s minimum, 75–90 s recommended, 120 s maximum. Reflow: "SAC305 liquidus: 217–220°C. Minimum peak: 235°C (ensures all joints exceed liquidus even with ΔT). Recommended peak: 245–250°C. Maximum peak: 260°C (MSL-3 limit per J-STD-020)." TAL: minimum 30 s, recommended 40–70 s, maximum 90 s; "For BGA packages: 60–90s preferred (solder balls under the package heat slower than exposed joints)." Cooling: target 2–6 °C/s, maximum 6 °C/s, minimum 1 °C/s. "Fast cooling at 4–6°C/s gives a fine-grained SAC structure." "A ramp above 3°C/s builds thermal gradients across the board. Four failures follow. Paste splatters as volatiles flash off. The board warps from uneven expansion. Ceramic capacitors and BGAs crack. Small passives at 0201 and 0402 tombstone from unequal pad heating." "A ramp below 1°C/s brings its own cost. Profile time stretches out, flux burns off before the reflow zone, and throughput drops." "One common error is running the soak too hot, above 200°C. Flux activators burn out early, and too little remains to wet the pads at peak." "Slow cooling below 2°C/s gives the opposite. Large Ag₃Sn platelets form and act as crack starters. Fatigue life drops. Thermal cycling failures rise."

  2. AtlasPCB, "PCB Reflow Oven Profiling"
    What we used

    "Reflow soldering is the dominant assembly process for surface-mount technology, and getting the thermal profile right is the single most important factor in achieving consistent solder joint quality." "The profile is measured using thermocouples attached to the board at critical locations, typically the coldest point (usually the center of the largest ground plane), the hottest point (usually a small component near the board edge), and on BGA solder joints themselves." Thermocouple placement: "1. BGA joint, attach K-type thermocouple (36 AWG) to a corner pad of the largest BGA using high-temperature solder or Kapton tape. 2. Board center, on the largest ground plane area (coldest point). 3. Board edge, near a small component (hottest point). 4. Connector area, if through-hole components will undergo wave or selective soldering later." "The temperature difference between the hottest and coldest points (ΔT) is the primary metric for profile optimization. For BGA boards, ΔT should be: Standard: <15°C; Fine-pitch BGA: <10°C; High-reliability: <7°C." "For SAC305 solder (liquidus 217–220°C), the ideal peak temperature is 245–250°C for most BGA packages." "The peak temperature must exceed the liquidus by enough to ensure complete wetting and intermetallic compound (IMC) formation. For SAC305, the solder must be fully liquid for tin to diffuse into copper (or nickel for ENIG-finished boards) and form Cu₆Sn₅ or Ni₃Sn₄ intermetallic layers."

  3. AtlasPCB, "Reflow Soldering Profile Optimization"
    What we used

    "SAC305 alloy melts at 217-220°C versus 183°C for eutectic SnPb, a 34°C increase that pushed peak reflow temperatures dangerously close to component and PCB material thermal limits." "Every solder joint on a modern PCB, from 0201 passive components to 1500-ball BGAs, forms simultaneously during a precisely controlled thermal cycle lasting 4-7 minutes." TAL: "40-90 seconds for SAC305. Minimum 40s ensures complete wetting of all joint surfaces. Maximum 90s prevents excessive intermetallic compound (IMC) growth and copper dissolution. Optimal range: 50-70 seconds for most assemblies." Peak: "240-250°C. Must exceed liquidus by ≥23°C for reliable wetting." "Delta-T (ΔT): <10°C across the board at peak." Cooling: "2-4°C/s from peak to 150°C." "The temperature difference between a 0402 capacitor on a signal layer and a 1156-ball BGA on a ground plane can exceed 30°C during ramp, but you need both to reach peak within 10°C of each other." "Total profile time: <7 minutes (to prevent excessive board exposure)." "Slower ramp rates (<1.0°C/s) waste oven capacity and may cause oxidation of exposed copper and component leads before flux activates."

  4. JLCPCB, "Reflow Soldering Profile Explained"
    What we used

    "The four stages of a reflow temperature profile: preheat, soak, peak reflow, and controlled cooling." Preheat: "Ambient to 150°C, 1.0–3.0°C/sec ramp." Soak: "150°C–200°C, 60–120 seconds." Reflow: "230°C–250°C, 45–90 seconds (Time Above Liquidus)." Cooling: "Peak down to Ambient, -2.0 to -4.0°C/sec slope." "Modern manufacturing relies almost exclusively on SAC305 (Sn96.5/Ag3.0/Cu0.5) alloys." "The melting point of SAC305 is ~217°C, meaning the peak oven temperature must hit 240°C-250°C to achieve proper wetting." "A healthy IMC layer (1-3 microns) is necessary for a mechanical bond. However, if TAL extends to 110 seconds or beyond, the IMC layer grows excessively thick." 8-zone oven table: zones 1–3 preheat 120/140/160 °C, zones 4–5 soak 180/190 °C, zone 6 pre-reflow 210 °C, zone 7 peak 255 °C, zone 8 cooling. "If the ramp is too aggressive, the volatile solvents in the solder paste will boil rapidly, causing paste splattering (solder balls) and thermal shock to ceramic capacitors."

  5. JLCPCB, "Reflow Soldering Profile Troubleshooting Guide"
    What we used

    "Ramp Rate (Preheat): Recommended range: 1-3 °C/sec. Too fast → thermal shock, tombstoning. Too slow → paste slump, bridging." "Soak Temperature & Time: Recommended range: 150-200 °C for 60-120 seconds. Too hot/long → premature flux exhaustion, oxidation. Too cold/short → trapped volatiles, voiding." "Peak Temperature & TAL: Recommended range: 235-250 °C (SAC305) with 45-90 seconds TAL. Too high/long → PCB warpage, component damage, charred flux. Too low/short → cold solder joints, incomplete wetting." "Cooling Slope: Recommended range: -2 to -4 °C/sec. Too fast → component cracking, pad cratering. Too slow → coarse-grain structure, dull/weak joints." "A heating ramp that is too fast, causing one pad to reach the alloy's liquidus temperature before the other." "Recommended profile adjustments include: Reduce the initial ramp rate to approximately 1.5 °C/sec to prevent rapid surface tension imbalances. Extend the soak zone duration to achieve thermal equilibrium across the entire board."

  6. LeadsIntec, "Mechanical Impact of PCB Pad Dimension Design on Tombstoning During SMT Reflow Soldering"
    What we used

    "Tombstoning (Manhattan Effect) is one of the most common failure issues in SMT reflow soldering of miniature surface-mount components such as 0201 and 0402 packages. The core cause is the imbalance of solder joint forces at both ends of the component." "Measured data shows that for a 0603 MLCC component (terminal electrode size: 0.3mm × 0.15mm), when the pad length on one side is 0.08mm longer than the other side (increased from 0.45mm to 0.53mm), under conventional SAC solder paste conditions (γ≈0.62N/m, θ≈30°), the surface tension difference between both ends can reach 12–15μN." "This force difference is more than three times the self-weight of the 0603 component (approximately 3.2μN), completely overcoming the component's gravitational resistance and triggering component rotation and tombstoning."

  7. SMTA International 2007 (electronics.org), "A Study of 0201s and Tombstoning in Pb-free Systems"
    What we used

    "A review of published studies and high-volume production experiences uncovered 49 potential parameters that can influence tombstone defects. The investigators then used a cause-and-effect (C & E) matrix approach (rating influence 0, 3 or 9) to identify the following 12 parameters as the most influential." "The largest contributor to defect generation was the combination of 0.1 mm (0.004\") print and placement offsets." "In descending order of influence, the next three major factors were component type, orientation, and reflow atmosphere." "The impact of removing mechanical slop from the assembly process brought the overall defect rates from 11,823 to 2,513 ppm, or from 7,845 ppm to 156 ppm excluding the solder ball count." "The mechanics of tombstoning are based on the wetting forces of molten solder and the mass and geometry of the component. If the solder paste under one termination melts and starts wetting to the termination before the other side does, the force applied by the wetting action can pull that component up on its end. Typically, the smaller and lighter the component is, the more susceptible it is to tombstoning."

  8. Electric-engine inverter PCBA root-cause report (single documented case), "Components lifting on one side (tombstoning during reflow). Affected components: 0402 & 0603 passive resistors. Defect Rate: 3, 5 % per board, 75% per batch. Impact: Open circuits, rework delays (engineering), reliability."
    What we used

    PFMEA: reflow step S9 O4 D4, RPN 144

  9. I-Connect007, "Tombstone Troubleshooting"
    What we used

    "As the flux and solder alloy liquefy and wet to each side of a component, they apply small amounts of torque through surface tension... tombstoning is caused by minute differences in wetting force from one side of a component to the other. When there is a sufficient imbalance in torque, relative to the mass of the component, the component is tipped upright (tombstoned)." "Chip components have gotten much smaller over the years with many weighing in the milligrams. The same torque that once helped alignment now has the power to tombstone components." "When board designers are laying out a circuit board, manufacturability considerations are often unknown or ignored. Having established that wetting force imbalance causes tombstoning, one of the primary causes for imbalance is the difference in temperature and, correspondingly, the difference in time of reflow between the two pads the chip sits on."

  10. Singh (2005), "Factors influencing tombstoning"
    What we used

    "As component size decreases from 0603 to 0402, 0201 and even 01005, solder paste surface tension has catastrophic effect. Surface tension forces acting on two sides of the pads, if not balanced by component weight, can lift components from the circuit board." "The defect is caused mainly by uneven wetting in which one termination solders before the second, causing the surface tension of the solder to pull the device in one direction." "Reflow profiles play a major role; if the thermal mass distribution is uneven, one pad will be heated first and the second pad will be heated a few seconds later."

  11. South Electronic PCB, "Stuck with SMT Assembly Defects?"
    What we used

    "A pad sitting over a huge copper plane is like attaching it to a giant ice cube. That pad stays cooler longer. The partner pad heats up fast, melts quickly, and develops strong surface tension. This imbalance can easily flip the little part on its end before the cooler pad even gets warm. Fixing this often means breaking that copper pour connection with thermal spokes."

  12. YourPCB, "SMT Reflow Profile for PCB Assembly"
    What we used

    "Freeze the reflow profile before pilot approval, not after the first production defect." "Require measured thermocouple data on the real board, including the hottest and coldest risk locations." "Do not approve a profile based only on the oven's centerline temperature. The supplier should attach thermocouples to the coldest heavy thermal mass and the hottest heat-sensitive package." "For many SAC305 lead-free SMT builds, the supplier may target a peak around 235 to 245 degrees C and a time above liquidus around 45 to 90 seconds."

  13. BME MECA course notes, "Reflow soldering ovens and reflow profiling"
    What we used

    "With SAC305 or SAC405 lead-free alloys, the coolest joints on a board should generally reach at least 228 °C, preferably 235 °C. For the lead-free alloys, the temperature 228 °C represents smaller than 10 °C superheating above the initial melting point (219-221 °C). The 235 °C represents 15 °C superheating." "250 °C is recommended as the maximum temperature for all solder joints on the board, except for components with temperature ratings lower than 250 °C. If maximum solder joint temperatures exceed 250 °C, PCB damage such as delamination and warpage may result when standard FR4 (Tg = 130 °C) material is used."

  14. Demo arithmetic verified in Python before writing: preheat 25 → 150 °C = 125 °C rise; at 2.0 °C/s = 62.5 s (in
    What we used

    The 60–90 s window); at 1.5 °C/s = 83.3 s; at 3.5 °C/s = 35.7 s (too fast); at 0.8 °C/s = 156.2 s (too slow, flux burns off). Peak 247 °C = 27–30 °C above the 217–220 °C liquidus (JLCPCB: peak 20–30 °C above the melting point). Cooling 247 → 150 °C = 97 °C drop; at 3 °C/s = 32 s; at 1 °C/s = 97 s. Total profile: preheat 62.5 + soak 90 + reflow ~60 + cooling ~32 = 246 s = 4.1 min (4–7 min range, <7 min limit). Tombstone force: 12–15 µN vs. 3.2 µN self-weight = 3.75× to 4.69×. 0201 study: 11,823 → 2,513 ppm = 78.7% reduction; 7,845 → 156 ppm = 98.0% reduction. Inverter case on a 200-board batch at 3–5% = 6–10 boards; 75% of batches affected = 3 of 4. PFMEA: 9 × 4 × 4 = 144 RPN. ΔT: 30 °C during ramp → <10 °C at peak; the soak must close 20 °C of gap. Rework: 15 min/board × 8 boards = 120 min = 2 h per 200-board batch (article's own assumption, labelled)

More in design notes

Aluminium, Steel, or PlasticAluminium, steel, or plastic? The material decides the cost, the weight, the heat, the EMI, and whether the product survives outdoors, and the wrong one fails in under two years. Three axes, four plastics, three metals, and the arithmetic for Example Labs' field sensor hub.20 minCNC Machining DFMA 30 mm deep pocket with R2 corners forces a 4 mm end mill that deflects 16 times more than the tool the job needs, and a drawing with ±0.01 mm on every dimension costs eight times the same part at ±0.5 mm. Eight design rules, each with a number, and a machined sensor hub worked from $1,370 a part to $430.25 minEnclosures That Survive the Real WorldIP67 is a laboratory test on a new unit, for thirty minutes. It says nothing about year two, when the gasket has taken a compression set and the box has baked in the sun. A sealed 100 × 60 × 40 mm box with 5 W inside runs 25 °C hotter than the air around it. The enclosure is not a box, it is a promise: keep the water out, let the heat out, and open again when something breaks.14 minInjection Moulding DFMInjection molding design determines 70–80% of final part cost and quality before production begins. The five rules are simple: a uniform 1.5–3.0 mm wall, 1–3° of draft per side, ribs at half the wall, bosses at six-tenths, and shrinkage built into the cavity, PP shrinks 1.0–2.5%, so a 100 mm part needs a cavity of 101–102.5 mm. Roughly 70% of first-submission designs contain at least one feature that would cause a defect or force an expensive mould action, and one post-tooling modification typically costs $2,000–$20,000 and 2–6 weeks. The fix is a CAD edit, not a tool edit.19 minSheet Metal DFMYour enclosure came back 7 millimetres too wide, with a cracked bend and a hole stretched into a slot. Sheet metal is not machining, the flat pattern is the part, and the bend is where the size lives. Twelve rules, in the order you meet them on a model.17 min