QFP solder bridging is almost always a geometry problem wearing a process costume. Before you touch the reflow oven, check three numbers: the copper-to-copper gap between adjacent lands, the solder mask expansion your CAD tool applied, and the stencil aperture width. On a 0.5 mm-pitch part those three decide the outcome, and two of them are set in the layout, not on the line.
Key takeaways
- A 0.10 mm per-side solder mask expansion on a 0.5 mm-pitch QFP makes adjacent mask openings overlap. The dam is gone before the board reaches the printer.
- Area ratio is almost never the binding constraint on a 0.5 mm-pitch QFP. A 0.30 mm × 1.60 mm aperture in a 0.12 mm foil gives an area ratio of 1.05, far above the IPC-7525B floor of 0.66.
- Type 4 powder becomes mandatory at 0.4 mm pitch, not at 0.5 mm. The five-ball rule fails at 4.9 balls with Type 3 in a 0.22 mm aperture.
- Where the bridge sits tells you the cause. Toe bridges mean volume, heel bridges mean placement, repeating bridges on the same two pins mean data.
- Under IPC-A-610, bridging is a defect in every class. There is no Class 1 allowance to hide behind.
What causes QFP solder bridging?
QFP solder bridging happens when molten solder from two adjacent lands coalesces into one mass. The cause is excess solder volume in the gap, paste displaced into the gap, or a missing solder mask dam. Fine pitch amplifies all three because the gap between lands shrinks faster than the tolerances that feed it.
Ranked by how often each one is the actual root cause on a fine-pitch QFP, rather than by how often it gets blamed:
| Rank | Root cause | Physical mechanism | Where you fix it |
| 1 | Solder mask dam removed by mask expansion | Adjacent mask openings merge into a gang opening; molten solder has an uninterrupted copper-free path between lands | Layout: local mask expansion rule |
| 2 | Print misregistration | Paste brick lands partly on the dam instead of the land; the offset volume has nowhere to go at reflow | Printer setup, fiducials, stencil tension |
| 3 | Aperture too wide for the pitch | Paste volume exceeds what the land can hold once it collapses | Stencil data: width reduction |
| 4 | Paste smeared under the stencil | Excess squeegee pressure or a dirty stencil underside deposits paste in the gap | Understencil wipe frequency, squeegee pressure |
| 5 | Placement Z-height too low | The nozzle presses the lead through the brick and squeezes paste sideways past the land edge | Placement program: Z and dwell |
| 6 | Lead coplanarity out of spec | A high lead sits above the paste and drags the neighboring joint when it finally wets | Incoming inspection, handling, tray damage |
| 7 | Hot slump during ramp | Paste viscosity collapses before the flux activates and the deposit spreads into the gap | Paste selection, ramp rate, paste age |
Notice what is not at the top. Peak reflow temperature is the first thing most teams change and it is rarely the cause. Solder that has already been displaced into the gap will bridge at any peak temperature above liquidus.
Diagnose before you change anything: what the bridge location tells you
Pull five bridged boards and put them under a stereo microscope at 20× before adjusting a single process parameter. The position and the pattern of the bridges narrow the cause faster than any single-variable experiment.
| What you see | Most likely cause | First action |
| Bridge at the toe, outboard end of the land | Aperture too long or too much volume | Shorten aperture length 10 percent, re-print, measure with SPI |
| Bridge at the heel, close to the package body | Paste squeezed sideways at placement, or no mask dam | Check placement Z-height and the mask opening in the fab data |
| Same two pins bridged on every board | Footprint or stencil data error, not process | Diff the paste layer against the copper layer for those two lands |
| Bridges scattered randomly across all four sides | Print process variation | Increase understencil wipe frequency, verify squeegee pressure |
| Bridges on one side of the package only | Placement rotation, board warp, or oven airflow asymmetry | Check placement theta and profile that side of the package |
| Bridges with satellite solder balls nearby | Hot slump or paste past its working life | Check paste open time and ramp rate below 150 °C |
| Paste visibly on the mask before reflow | Registration, not volume | Do not reduce the aperture. Fix the print offset first |
That last row is the one that costs teams the most time. Reducing aperture size to cure a registration problem trades bridges for opens, and the opens are harder to see.
[IMAGE 1: annotated microscope photo of a fine-pitch QFP showing a toe bridge, a heel bridge and a clean joint side by side | alt: “QFP solder bridging at the toe and heel of adjacent leads compared with an acceptable joint”]
The geometry that decides everything
Start from the package drawing, not from the defect. Verified dimensions from manufacturer package outlines and the EIA-standard soldered-pad layouts published for quad flat packs:
| Pitch | Lead width b (mm) | Land width (mm) | Land length (mm) | Copper gap at nominal land | Lead coplanarity |
| 0.65 | 0.22 to 0.40 | 0.30 to 0.50 | 1.80 | 0.25 mm at a 0.40 mm land | 0.10 mm |
| 0.50 | 0.17 to 0.27 | 0.30 to 0.40 | 1.60 | 0.15 mm at a 0.35 mm land | 0.08 mm |
| 0.40 | 0.13 to 0.23 | no published EIA layout | derive per IPC-7351B | 0.15 mm at a 0.25 mm land | per datasheet |
Sources: lead widths and coplanarity from the Altera Package Information Datasheet for Mature Altera Devices, DS-PKG-16.8, December 2011, for the 100-lead PQFP (MS-022 GC-1), the 208-lead PQFP (MS-029 FA-1) and the 64-lead 0.40 mm-pitch outline (MS-026 ABD-HD). Land geometry from the EIA standard soldered-pad layout table in the Xilinx Packages and Thermal Characteristics data book, version 2.0, 20 November 1997.
The number that matters is the copper gap. At 0.5 mm pitch with a 0.35 mm land you have 0.15 mm of bare laminate between adjacent copper. Everything else in this article is a fight over that 0.15 mm.
Worked example: the 4 mil mask expansion that deletes your dam
Take the 0.5 mm-pitch case. Land width 0.35 mm, pitch 0.50 mm, copper gap 0.15 mm. Now apply the mask expansion values people actually set in CAD.
At 0.10 mm per side, a common global default: each opening becomes 0.35 + 0.20 = 0.55 mm wide on a 0.50 mm pitch. Adjacent openings overlap by 0.05 mm. There is no dam. The fabricator ships a gang opening and the bridging is designed in.
At 0.05 mm per side: openings are 0.45 mm wide, leaving a 0.05 mm dam. That is below every published fabricator minimum, so the fab deletes the dam and ships a gang opening anyway.
At zero expansion, 1:1 with copper: openings are 0.35 mm wide and the dam is 0.15 mm. That clears the published thresholds. JLCPCB documents removing dams automatically below 0.10 mm for green ink on 1 oz copper and below 0.13 mm for black or white ink, and states laser direct imaging holds alignment to ±0.02 mm, which is what makes 1:1 openings viable.
Worthington Assembly puts the same rule in one line for 0.5 mm and 0.4 mm-pitch QFP and QFN designs: the fabricator needs at least a 0.10 mm gap between mask openings to form a dam, and designers should insist on it rather than accept gang relief.
Practical rule: set a local mask expansion of zero on every QFP land at 0.5 mm pitch or finer, and choose green mask on 1 oz copper. A global 4 mil rule that is fine on an 0805 resistor is fatal here.
Stencil design for fine-pitch QFP
IPC-7525B sets two release rules: area ratio above 0.66 and aspect ratio above 1.5. Area ratio is aperture area divided by aperture wall area, or (L × W) ÷ [2 × (L + W) × T].
Run the arithmetic for a 0.5 mm-pitch QFP land of 0.35 mm × 1.60 mm in a 0.12 mm foil. At 1:1 the area ratio is 1.20. Reduce the width to 0.30 mm and it is 1.05. Drop to a 0.10 mm foil at 1:1 and it is 1.44. Every option clears 0.66 by a wide margin.
The conclusion is worth stating plainly, because the generic advice gets it backwards: on a 0.5 mm-pitch QFP you have paste volume to spare, so reduce it. The constraint is registration and dam width, not release.
| Pitch | Foil thickness | Aperture (W × L) | Area ratio | Volume per lead | Five-ball check | Paste type |
| 0.65 mm | 0.15 mm | 0.40 × 1.80 mm | 1.09 | 0.108 mm³ | 0.40 / 0.045 = 8.9 | Type 3 |
| 0.50 mm | 0.12 mm | 0.30 × 1.60 mm | 1.05 | 0.0576 mm³ | 0.30 / 0.045 = 6.7 | Type 3 works, Type 4 prints better |
| 0.50 mm | 0.10 mm | 0.35 × 1.60 mm | 1.44 | 0.0560 mm³ | 0.35 / 0.045 = 7.8 | Type 3 |
| 0.40 mm | 0.10 mm | 0.22 × 1.40 mm | 0.95 | 0.0308 mm³ | 0.22 / 0.045 = 4.9 fails | Type 4 required |
Area ratios and volumes computed from the land geometry above. The five-ball rule from IPC-7525 says the smallest printable aperture is five times the largest powder sphere, so the check divides aperture width by the top of the powder range: 45 µm for Type 3 and 38 µm for Type 4 under IPC J-STD-005A. The 0.40 mm-pitch row assumes a 0.25 mm land, since no EIA layout is published at that pitch.
So Type 4 powder is not a fine-pitch talisman. It buys you nothing measurable at 0.65 mm, it slightly improves print definition at 0.5 mm, and it becomes a hard requirement at 0.4 mm.
Print settings that create bridges
Once the geometry is right, the printer is where bridges are made. Four settings, in the order worth checking:
- Squeegee pressure. Use the lowest pressure that still gives a clean wipe. Excess pressure forces paste under the stencil edge and smears it into the gap between lands.
- Understencil wipe frequency. Paste accumulating on the stencil underside transfers to the mask between apertures and reflows into a bridge. If bridges appear on fewer than about five percent of lands, wipe before you change anything else.
- Print speed. Indium Corporation notes that excessive print speed causes a loss of viscosity that promotes slumping, and slump on a fine-pitch part goes straight into the gap.
- Paste open time. Flux volatiles leaving the deposit change its rheology. Track time on the stencil rather than assuming a shift is a shift.
Add SPI if the line has it. Volume control alone will not catch a registration problem, so watch the offset measurement, not just the volume window.
Placement and reflow contributions
Placement matters more than most process engineers credit at 0.5 mm pitch. A lateral offset of 0.075 mm puts the edge of a 1:1 paste brick over the centerline of a 0.15 mm dam. Reducing the aperture from 0.35 mm to 0.30 mm buys back 0.025 mm of offset budget per side, which is why aperture reduction and registration control solve the same defect from different ends.
Z-height is the other half. If the nozzle drives the lead through the brick, paste extrudes sideways past the land edge and reflows as a heel bridge. That is the signature the Arduino and EEVblog threads on 0.5 mm QFP bridging keep describing: bridges sitting higher up, closer to the package body, rather than at the pin tips.
Reflow is a smaller lever than its reputation suggests, with two real contributions. A ramp that is too fast below 150 °C drives hot slump before the flux has activated. And a package that heats unevenly, because of oven airflow or board warp, wets one side first and pulls solder across the cooler side. Profile the QFP body itself rather than trusting the recipe from a different assembly.
Fixing a bridge that already exists
Rework is straightforward on a QFP, which is one of the format’s few remaining advantages. Every joint is visible and individually accessible.
- Flux first, always. Fresh liquid flux on the whole affected row changes the surface tension balance and does more than any amount of extra heat.
- Drag with a cleaned chisel or spoon-shaped tip. Pace specifies a starting tip temperature of 316 °C for tin-lead alloys and 343 °C for lead-free when drag soldering fine-pitch QFPs, and notes that misalignment of as little as a quarter of a land will produce bridges in the first place.
- Wick only what you must. Fine braid pulls solder off the joint as well as out of the gap, so re-inspect for insufficient fillets afterward.
- Hot air for multiple adjacent bridges. Flux the row, heat evenly, and let surface tension redistribute the solder rather than trying to remove it.
- Clean and re-inspect at 20× or higher. A bridge that reforms in the same place on the same pins is a data problem, not a technician problem.
One caution on repeated rework: each reflow cycle adds thermal exposure to a moisture-sensitive plastic package. If a board has been through rework more than twice, treat the package as suspect rather than assuming the joint is the only thing affected.
Inspection and acceptance
IPC-A-610 classifies bridging as a defect, and the class logic runs downward: a defect for Class 1 is automatically a defect for Class 2 and Class 3. There is no consumer-grade allowance for a short.
For a QFP the inspection stack is simpler than for area-array parts, because the joints are exposed. Visual inspection with magnification matched to the pitch catches most bridges, AOI catches them repeatably in volume, and continuity testing between adjacent nets catches the ones hidden under the package body. X-ray, which is mandatory for BGA bridging, is optional here.
[IMAGE 2: AOI capture of a 0.5 mm-pitch QFP with one flagged bridge and the adjacent acceptable joints | alt: “Automated optical inspection image showing QFP solder bridging flagged between two adjacent leads”]
Frequently asked questions
What causes solder bridging on QFP components?
Excess solder volume in the gap between adjacent lands, paste displaced into that gap by print misregistration or placement pressure, or a missing solder mask dam. On 0.5 mm-pitch parts the copper gap is only about 0.15 mm, so small tolerance stacks that are harmless on larger packages consume the whole margin.
How do you fix a solder bridge on a QFP?
Apply fresh liquid flux across the affected row, then drag a cleaned chisel or spoon tip along the leads so excess solder wicks onto the tip. Braid works but removes solder from the joints too, so re-inspect fillets afterward. Hot air is better when several adjacent leads are bridged.
Does solder mask between pins prevent bridging?
A real dam helps, but only if it survives fabrication. At 0.5 mm pitch a 0.10 mm per-side mask expansion makes adjacent openings overlap, so the dam disappears and the fabricator ships a gang opening. Use zero local expansion on fine-pitch QFP lands and green mask on 1 oz copper.
Should I use Type 4 solder paste for 0.5 mm-pitch QFP?
It is optional at 0.5 mm and required at 0.4 mm. Under the five-ball rule, a 0.30 mm aperture holds 6.7 Type 3 spheres and prints acceptably. A 0.22 mm aperture at 0.4 mm pitch holds only 4.9, which fails, and Type 4 at 38 µm restores the margin.
Is solder bridging always a defect?
Yes, under IPC-A-610. Bridging is a short between conductors that should be isolated, and a condition classified as a defect for Class 1 is a defect for Class 2 and Class 3 as well. It requires rework or rejection regardless of product class.
What to do next
If bridges are appearing now, work in this order: inspect five boards under magnification to locate the bridges, check the fabrication data for a gang opening where you specified a dam, then check print registration, then reduce aperture width. Changing the reflow profile first wastes a shift.
If you are designing a board with 0.5 mm-pitch QFPs, set three rules before layout release. Local solder mask expansion of zero on those lands. Aperture width at roughly 85 percent of land width with the length reduced at the toe. Foil thickness at 0.10 mm to 0.12 mm, and check the area ratio only to confirm it is comfortably above 0.66, which it will be.
At 0.4 mm pitch, add Type 4 paste and negotiate the dam with the fabricator in writing before the first build. That is the pitch where the geometry stops forgiving anything.
External references: Altera Package Information Datasheet for Mature Altera Devices (https://www.mouser.com/datasheet/2/612/pkgds-1299398.pdf); Xilinx Packages and Thermal Characteristics data book v2.0 (https://studylib.net/doc/18690369/xilinx-packages-and-thermal-characteristics–data-book-se); IPC-7525B Stencil Design Guidelines preview (https://webstore.ansi.org/preview-pages/IPC/preview_IPC+7525B-2011.pdf); IPC J-STD-005A Requirements for Soldering Pastes (https://www.electronics.org/TOC/IPC-J-STD-005A.pdf); Worthington Assembly, solder mask dams between fine-pitch devices (https://www.worthingtonassembly.com/blog/2021/11/19/solder-mask-dams-between-fine-pitch-devices); Indium Corporation on solder paste slump (https://www.indium.com/blog/stuck-in-a-slump-factors-impacting-solder-paste-slump-susceptibility/).
Internal links: [INTERNAL LINK: PQFP package dimensions and JEDEC outlines → PQFP package guide]; [INTERNAL LINK: Xilinx FPGA series list and pricing → https://fpga.io/xilinx-fpga/]; [INTERNAL LINK: FPGA brands and manufacturers guide → https://fpga.io/fpga-brand/]; [INTERNAL LINK: Gowin FPGA package and series guide → https://fpga.io/gowin-fpga/]; [INTERNAL LINK: Efinix FPGA sourcing and package trade-offs → https://fpga.io/efinix-fpga/].
