The conclusion first: at microelectronics scale, printing failure changes character. In conventional printing, "under-filled" is a parameter problem — squeegee pressure, speed, separation speed, paste condition. Adjust them and you usually get back inside the window. Below a certain aperture size something different appears: apertures that release very little paste, or none, under any combination of parameters. Continuing to hunt for a window on the printer has no exit at that point, because the limit has moved from the machine into the aperture itself.
This article is about where that transition sits, what sets it, and which decisions must be frozen before the stencil is cut.
1. Under-fill and no-release are two different failures
Both show up on SPI as low paste volume. The causes are different and the responses are opposite.
Under-fill means the aperture was never properly packed. It usually correlates with squeegee travel — positions near the end of the stroke are worse — and it moves noticeably when you change a paste jar or adjust pressure and speed. That is a parameter problem.
No-release means paste entered the aperture and stayed there on board separation. Its signature is that it is position-independent: every aperture of that size on the board is equally bad, while slightly larger apertures right next to them are fine. Parameter changes buy very little, and often cost something elsewhere — slowing separation speed to improve release can wreck line throughput.
The triage is direct: check whether the failure tracks aperture size rather than position. If all the small apertures fail and all the large ones pass, stop looking for the answer on the printer.
2. Inside a small aperture, paste is no longer a fluid
At conventional sizes, solder paste behaves approximately like a high-viscosity, shear-thinning continuum: the squeegee shears it, it thins and fills, shear stops and it recovers enough to hold shape, then the pad pulls the whole slug out on separation.
Shrinking the aperture breaks two premises of that model at once.
First, the particle count drops. Paste is metal powder in a flux vehicle, and the powder has a particle size distribution. When aperture width can only accommodate a small number of the larger particles, the contents can no longer be treated as a continuum — it is a packing of a few particles plus vehicle, governed by particle-to-particle and particle-to-wall contact rather than by bulk rheology. This is a "not enough particles" problem, not a "paste quality" problem.
Second, the surface-area balance inverts. An aperture is a cavity, and for paste to come out, adhesion on the pad side must beat adhesion on the wall side. Scale an aperture down proportionally and the pad area falls with the square of the dimension, while wall area falls linearly and is multiplied by depth. In other words, the smaller the aperture, the relatively larger its walls. That is why area ratio — aperture opening area over aperture wall area — is the governing metric here, and why it collapses so quickly at small sizes.
IPC-7525 offers rules of thumb of roughly 0.66 area ratio and roughly 1.5 width-to-thickness ratio; below those, release degrades fast. How those are calculated and which defects map to which side of the line is covered separately in stencil aperture area ratio and paste release, and this article only uses the conclusion. What matters here is what remains available once your pads are small enough that the line cannot be satisfied at all.
3. Aperture geometry: which ratio bottoms out first
The two ratios constrain different aperture shapes, and designs have to be checked against both.
Width-to-thickness ratio (minimum aperture width ÷ foil thickness) governs narrow slot-shaped apertures — split thermal pads on fine-pitch QFNs, long connector pads. These are not small in area, but one direction is narrow and paste there is pinched between two walls.
Area ratio governs near-square and round apertures — fine-pitch BGA, 01005-class chip pads. No direction is unusually narrow, but every dimension is small.
Whichever ratio is worse decides the outcome. A common misjudgment is clearing a board on area ratio alone when it is full of narrow slots. The practical method is to sort the board's apertures by shape, check the narrow ones on width-to-thickness and the near-square ones on area ratio, then see which group falls out of bounds first. That group is the board's bottleneck and where optimization belongs.
Two geometric measures help noticeably at small sizes: tapered aperture walls (opening slightly smaller at the top so paste separates toward the pad) and corner radii (so paste is not pinched in square corners). Both improve release without changing opening size, but the margin they buy is finite. Neither rescues an aperture far below the rule-of-thumb line.
4. Why "thinner is better" does not hold for foil
Foil thickness sits in the denominator of both ratios, so a thinner stencil improves both. True — but only within limits, and at three costs.
Cost one: thin foil constrains the thickest joint on the board. A stencil has one thickness (step stencils aside) and must serve every component on the board. Thin it enough for fine-pitch BGA and the large connectors, shields and power devices on the same board may be starved. Mixed-technology conflicts of that kind are handled with step stencils, and the trade-offs are in step and multilevel stencil aperture design.
Cost two: thin foil is less rigid. The thinner the foil, the more it deforms locally around apertures under squeegee pressure, and the more readily excessive pressure drives paste from the aperture mouth into the gap between stencil and board. Thinning moves some difficulty from release into print-process stability.
Cost three: thin foil is more sensitive to board flatness. The stencil has to stay in contact with the board throughout, and the thinner the foil the more it depends on how the board is held. That side of the problem is covered in board support and tooling flatness in stencil printing.
So thinning is not a free improvement; it is a redistribution of difficulty. Before committing, confirm the requirement of the thickest joint on the board and whether the thinner window still contains it.
5. Wall condition carries more weight at small sizes
Aperture wall roughness is a second-order factor at conventional sizes and a first-order one at small sizes — for the same geometric reason as above: the smaller the aperture, the larger the wall share of total adhesion. Rough walls offer more mechanical keying and the paste is more reluctant to leave.
Two consequences follow. First, for boards dominated by small apertures, stencil fabrication method and post-processing stop being optional: the as-cut wall condition, whether polishing is applied, whether a low-surface-energy coating is applied — these decide whether the stencil is usable at all. Selection is covered in aperture wall polishing and nano-coating selection. Second, small-aperture stencils degrade faster in service: coating wear and wall residue show up as declining paste volume sooner than they would on larger apertures.
6. Matching powder type to aperture size
This is the direct countermeasure to the particle-count problem from section 2: as apertures shrink, powder must get finer.
The usual check is how many of the larger particles fit across the minimum aperture width. Too few, and fill randomness rises sharply, appearing as increased scatter in paste volume between identically sized apertures on the same board — not a uniform low reading, but a noisy one. This is why small-aperture problems often surface first in SPI standard deviation rather than in the mean.
The counter-cost to weigh at the same time: finer powder has more specific surface area, is more oxidation-sensitive, imposes tighter requirements on storage, warm-up and open-jar time, and changes viscosity faster during printing. Switching to finer powder carries running cost; it is not a free upgrade. Use the technical data sheet of the paste in question for grades and particle ranges — we do not quote numbers here.
7. The pad side actually holds the decision
Six sections about stencils, and now the fact that has to be stated plainly: the numerator of area ratio is aperture opening area, and the ceiling on opening size is set by the pad and its solder mask opening. Small pad, small aperture. Small aperture, no way back to an acceptable area ratio.
So at design stage the following are more effective than anything done to the stencil:
- Do not shrink pads without reason. Narrowing pads for routing convenience is paid for in printing.
- Define the mask opening approach. Solder-mask-defined (SMD) and copper-defined (NSMD) give different usable pad areas and change where paste sits.
- Keep small pads on the same net dimensionally consistent. Equal apertures can share one process window; a board with five different small pad sizes asks one parameter set to satisfy several windows.
- Settle thermal pad split patterns together with voiding requirements. The split affects release and post-reflow voiding at the same time. SKPCB PCBA can hold soldering voiding under 10% with nitrogen reflow, but that depends on pad and aperture design working together — reflow process alone does not deliver it.
8. Real constraints near the limits
Put the above together and it explains why a capability limit is not a promise that any board reaches it.
SKPCB PCBA capability boundaries are: smallest chip component 01005; minimum pitch 0.2 mm for BGA and 0.3 mm for QFN/QFP; largest component a BGA/QFP-class package up to 74 mm diagonal; BOM capacity 1000 lines; assembly to IPC-A-610 Class 3.
But limit values are single-axis extremes, not a freely combinable set. That 01005 and 0.2 mm pitch BGA are each feasible does not mean they, plus a large connector needing a thick deposit, all come out well on one board with one stencil and one parameter set. Feasibility is decided by the span between the smallest aperture and the largest paste demand on that board. Small span, wide window. Large span, and you are into step stencils, two-pass printing or process splitting — all of which belong in quotation and process preparation, not in a discovery at line start.
9. Mixed-technology boards: handling the span
A board carrying both 01005 and a large shield is normal. The available routes are few, each with a clear cost:
- One compromise foil thickness. Simplest, but both ends sit at the window edge, giving the least tolerance to paste lot and environmental drift. Suitable when the span is modest.
- Step stencil. Different thickness by zone, both ends served. Costs stencil money, requires extra validation of squeegee behavior at the step, and demands keep-out distance around step boundaries.
- Two-pass printing. For extreme spans. Costs throughput and adds a registration risk.
- Change the design. Remove the source of the span — move the thick-deposit device to the second side, or use a different attachment method. Cheapest route, but it has to be raised at design stage.
Which route depends on how large the span is, how large the volume is, and how many revisions are still coming. For high-mix low-volume work, a compromise thickness plus tighter incoming and in-process monitoring is often better value than a step stencil; at volume with a large span the conclusion reverses.
10. SPI interpretation and printability pre-review
Small-aperture problems have two windows for early detection.
After line start, read SPI for scatter rather than mean. As noted, insufficient particle count shows up first as noisy volume between identically sized apertures. An SPI rule set watching only mean and limits misses that signal until yield suffers. Grouping statistics by component type and watching within-group scatter is more useful than whole-board pass rate.
More valuable still is pre-review before line start. With stencil data in hand, sort apertures by shape, compute area ratio and width-to-thickness ratio, and flag everything below the rule-of-thumb line. Done before scheduling, it costs only engineering time. It answers the question "does this board contain a class of aperture that is bound not to release" — and if the answer is yes, what should change is the pad, the aperture or the foil thickness, not the parameters after line start.
SKPCB quality flow carries more than 10 inspection gates, including 100% electrical test, 100% FQC and per-batch reliability testing, while PCBA-side test coverage spans SPI, inline 2D AOI, post-reflow 3D AOI, X-ray, ICT and FCT. SPI's position in that chain is immediately after printing, and its value depends on whether the interpretation rules are set correctly. Incoming and handoff checks are covered in stencil incoming inspection and handoff.
11. What to state at RFQ and process preparation
If the board carries microelectronics-scale apertures, supplying the following at quotation avoids most rework:
- Smallest pad size and minimum pitch on the board, and which package classes they belong to.
- The component with the largest paste demand (connector, shield, power device), to establish the span.
- Mask opening approach (SMD / NSMD) and whether pad adjustment is permitted.
- Whether a step stencil is acceptable, and whether layout can be adjusted for it.
- Volume and changeover frequency, which decide stencil life and whether coating is worth it.
- Voiding requirements for thermal pads, confirmed together with the split pattern.
- Whether the board is double-sided, since second-side printing constraints differ from the first.
With that information in process preparation, printability pre-review becomes real work. Handing over one Gerber set plus "build it to the tightest spec" generally means process engineering makes a set of trade-offs on your behalf that you never see.
To confirm which regime your board sits in, send the design files for a stencil manufacturing and printability assessment, or review PCBA and SMT assembly capability and manufacturing capability limits.
FAQ
Q: How small is "microelectronics scale"? Is there a definite dimension?
A: There is no absolute size independent of the component, because the criterion is a ratio, not a dimension. The same aperture width over a 100 µm foil and over a 50 µm foil differ by a factor of two in width-to-thickness ratio and behave completely differently. The practical approach is to compute area ratio and width-to-thickness ratio and see whether they approach the IPC-7525 rules of thumb (about 0.66 and about 1.5). Apertures below those lines belong to the no-release class described here.
Q: Will a better printer solve small-aperture release?
A: It will improve print process stability and repeatability, but it cannot change the physics of release. Release is a contest between wall adhesion and pad adhesion, set by aperture geometry, foil thickness, wall condition and paste. The printer can only minimize losses under given conditions. If a class of apertures sits clearly below the rule-of-thumb line, a machine change returns very little.
Q: Is thinning the stencil the most effective fix?
A: It does improve both ratios simultaneously, at three costs: the thickest joint on the board may be starved, reduced foil rigidity makes the print process more sensitive, and sensitivity to board flatness rises. Confirm the largest paste demand on the board is still satisfied before committing. For boards with a large span, a step stencil is usually a better fit than overall thinning.
Q: Does a small aperture always require finer solder paste powder?
A: Once aperture width accommodates only a few particles, finer powder is necessary, otherwise fill randomness drives up volume scatter. But finer powder has greater specific surface area, is more oxidation-prone, tightens storage, warm-up and open-jar time requirements, and shifts viscosity faster during printing. It is a trade with running cost. Use the paste technical data sheet for grades and particle ranges.
Q: How do I know before line start whether my board will have printing problems?
A: Run a printability pre-review. With stencil data, split apertures into narrow-slot and near-square groups, check width-to-thickness and area ratio respectively, and flag everything below the rule-of-thumb line. Done before scheduling it costs only engineering time; found after line start it costs a stencil or a design change. Supplying smallest pad size, minimum pitch and the largest paste-demand component at RFQ makes the review accurate.
Q: Small apertures print badly, so why does SPI pass rate still look acceptable?
A: Because insufficient particle count affects scatter before it affects the mean. Identically sized apertures read high and low, the average may still sit inside specification, and a rule set watching only mean and limits misses it. Grouping by component type and watching within-group scatter catches it earlier.