The "Multilayer / Step" line on a stencil order tends to get ticked in one of two mindsets: someone heard that mixed-assembly boards need a step and ticked it for insurance, or nobody considered it until the first print came back with starved fine-pitch deposits and voided connector joints.
Neither is a decision. A step stencil is not standard equipment for mixed assembly, and it is not a patch you reach for after a bad print. It is a process choice with specific conditions where it applies and specific costs when it does. SKPCB's deployed SMT Stencil Manufacturing page lists it as one of four build types (Laser Cut, Electroformed, Multilayer / Step, Nano-coated) and states plainly what it does: different thickness zones on one stencil let large connector pads take more paste while fine-pitch areas take less.
That sentence already draws the boundary. What follows expands it into a check you can walk before ordering, with the costs laid out alongside.
1. Separate two classes of paste problem; only one belongs to a step
"Paste volume is wrong" looks the same on every first-article report, but on the stencil side these are two unrelated problems.
Class one: a whole-board systematic offset. Everything is light, or everything is heavy. The signature is that the deviation points the same way everywhere — large pads and fine-pitch alike. The cause usually sits in the foil thickness tier, in how apertures were compensated across the board, or in print parameters and release behaviour. None of that is local. A step will not help here, because a step changes the *relative* paste volume between zones, not the absolute level across the board.
Class two: a conflict inside one board. Large pads run short while fine-pitch runs heavy, or the reverse. The signature is both ends failing, in opposite directions. The root is geometric: foil thickness is a property of the whole stencil, while paste demand is per component. Area ratio = aperture area ÷ aperture wall area, and wall area contains thickness, so thickness only ever appears in the denominator. Go one tier thicker and every aperture on the board loses area ratio at once, with the fine-pitch group crossing below the IPC rule-of-thumb 0.66 first. Go one tier thinner and deposit volume on the large pads shrinks with it. One global parameter cannot satisfy two opposing local demands. There is no setting that works.
Only class two is what a step stencil is for. The test is direct: list the measured paste deviations grouped by component package and check whether the direction is consistent. Same direction → leave local thickness alone and go back to the thickness tier and aperture compensation. Opposite directions, both already unacceptable → then continue into the flow below.
Do not skip this. Mistaking a class-one problem for class two and ordering a step stencil leaves the offset exactly where it was, with the step's costs added on top.
2. Exhaust these three routes before changing local thickness
A step is one way to resolve a relative conflict, but it should not be the first. The reason is in the next section: it turns a geometrically flat stencil into a non-flat one, and the cost lands on the squeegee and on cleaning — both of which govern how consistent every subsequent print is. So the order of work is to look for room without touching geometry first.
Route one: change aperture size and shape. The fine-pitch problem is low area ratio, and area ratio has aperture area on top with aperture perimeter feeding the bottom. For the same open area, moving from a square to a rounded square or an oblong changes the perimeter and therefore the area ratio; modestly enlarging apertures within what the pad allows also raises it. Large pads work the other way — they rarely lack area ratio, they lack volume, and volume can be bought with aperture area rather than thickness. Plenty of boards declared to "need a step" simply inherited library-default apertures that were never checked group by group against area ratio. The aperture design reference table on /en/smt-stencil/ gives practical starting combinations by component type, and the same page states that released stencil data after engineering review governs the actual apertures.
Route two: change release conditions instead of thickness. For the same aperture at the same thickness, wall condition and surface state visibly change release behaviour. Two optional processes on the stencil page address exactly this: electrolytic polish (option, improves paste release) and nano-coating (optional, cuts wipe frequency and extends service life). Build type is a variable too — electroformed and laser-cut walls are not the same thing. If the fine-pitch symptom is paste staying in the aperture, these cost far less to try than thickness does. How much a given aperture group can improve has to be confirmed in engineering review against the incoming data and the equipment.
Route three: go back to the design. In some mixed-assembly boards the paste conflict originates in the pad definition itself. A connector's large pads sit so close to neighbouring fine-pitch parts that apertures can neither grow nor thicken; or packages with extreme size spread were placed together when the layout still had room to move. The cheapest fix there is not on the stencil, it is on the board. The test: if apertures and release conditions are both exhausted and both ends still sit in unacceptable territory, the question has stopped being "how do we print this" and become "does the board have to be laid out this way".
All three tried and still unbalanced — that is where a step belongs.
3. What a step changes, and where the cost lands
The principle is simple. Build zones of differing thickness into one foil so each zone serves its own aperture group. The stencil page puts it as different thickness zones on one stencil letting large connector pads take more paste while fine-pitch areas take less. It converts "one global parameter against two opposing demands" into "two local parameters, each with one job".
The cost comes from the same fact: the stencil surface is no longer flat.
Cost one: squeegee sealing. Contact printing depends on the squeegee riding sealed against the foil, pressing paste across it through continuous, even contact. At a step boundary there is a height change, and the seal is interrupted as the blade crosses it. Paste behaviour near the boundary therefore differs from behaviour mid-zone: it may bleed, it may not wipe clean, it may leave material on the lee side of the step. That produces one important constraint — boundaries cannot be drawn casually. A boundary should run through clear space between components, should never cross an aperture group, and should keep total length and the number of turns to a minimum. The more finely you subdivide, the more height changes the blade has to cross, and the harder consistency is to hold.
Cost two: cleaning and maintenance. Height changes and the roots of steps are where paste accumulates, especially on the board-contact side. That affects how effective wiping is and raises what the cleaning routine has to do. The stencil page describes nano-coating as cutting wipe frequency and extending service life — which is to say wipe frequency is already something you manage, and a step pushes it in the wrong direction. The wiping strategy for a specific step stencil has to be set in engineering review against the actual zone layout and the equipment.
Cost three: heavier data and release. The paste-layer data for a step stencil is not just apertures; it also carries the extent of each thickness zone. That has to be expressed in the incoming data or agreed explicitly at review. Accepted input on the stencil page is Gerber RS-274X (paste plus solder mask plus outline layers), ODB++ and DXF, with order quantity and quotation confirmed after review of the Gerber data, quantity and process requirements. Review of a step stencil takes one pass more than a flat one: not just apertures, but whether the zone boundaries are compatible with the component layout.
What these costs share is that they act on repeatability, not on whether the first article prints. First articles off a step stencil often look excellent. The question is whether print 50 and print 200 still match print 1. That is why a step should not be the default for mixed assembly.
4. The conditions, collected
Compressed into a check you can run before ordering. A step is worth building when all four hold:
- Deviations point in opposite directions. Large-pad and fine-pitch paste deviations are not on the same side, which rules out a whole-board systematic offset.
- Apertures and release conditions are exhausted. Every group has been checked against area ratio (fine-pitch groups pushed above the IPC rule-of-thumb 0.66 as far as the pads allow), shapes optimised, electrolytic polish and nano-coating evaluated.
- A boundary can actually be drawn. The two conflicting component groups are spatially separated on the board, and a boundary can run through clear space without crossing an aperture, without becoming fragmented. This condition is frequently the veto — interleaved large pads and fine-pitch parts leave a step physically nowhere to go.
- The production pattern can carry the maintenance load. Print count, changeover frequency and cleaning cadence for this stencil are confirmed in review as compatible with what a step requires.
When condition 3 fails, the answer is usually route three from section 2: change the layout, not the stencil. When condition 4 fails, what is worth discussing is splitting the two conflicting sides onto separate stencils — the stencil page supports Top, Bottom, or Top+Bottom builds, with the split strategy confirmed per board in process review — rather than forcing a step.
5. What to state before you order
Whether or not a step ends up being built, the ordering side should express the problem, not the conclusion. Writing down which component groups have conflicting paste demands, their packages and positions, and the direction and grouping of measured deviations is far more useful than ticking "Multilayer / Step" — because whether a boundary can be drawn, and where, is settled at review, and settling it needs information that lives on the design side.
Prepare alongside it: complete paste-layer data (with solder mask and outline layers), board outline and fiducials, and the intended use pattern for this stencil (prototype or volume, print frequency, whether long runs with less wiping matter). Choose frame sizes and material from the ranges published on /en/smt-stencil/. How to pick the foil thickness tier is a different question, covered separately on this site. Any capability figure that carries conditions is governed by engineering review.
Wider manufacturability boundaries — minimum trace and space, hole sizes, outline tolerances — are on manufacturing capabilities and tolerance standards. If the stencil is ordered together with placement, it goes through the PCB assembly service.
FAQ
Q1: Do all mixed-assembly boards need a step stencil?
No. A step addresses only the case where paste demands inside one board conflict in opposite directions, and only after apertures and release conditions are exhausted. Many mixed boards balance out once each group is checked against area ratio and aperture shapes are optimised.
Q2: How do I tell whether to change thickness or apertures?
Look at the direction. Everything off the same way → a global problem in the thickness tier or aperture compensation, unrelated to local thickness. Large pads and fine-pitch off in opposite directions → a relative conflict. Within a relative conflict, change apertures and release conditions first, local thickness last.
Q3: What limits a step boundary?
It should not cross any aperture group, it should run through clear space between components, and total length and turns should be minimised. The seal between squeegee and foil is interrupted at the height change, so the more fragmented the zoning, the harder consistency is to hold. Whether a boundary can be drawn, and where, is confirmed in engineering review against the actual layout.
Q4: Will a step stencil hold more paste residue?
Height changes and step roots are where paste accumulates, which raises what wiping has to achieve. /en/smt-stencil/ describes nano-coating as cutting wipe frequency and extending service life, so it is worth evaluating alongside. The specific wiping strategy follows from the zone layout and the equipment.
Q5: What if no boundary can be drawn?
That means the conflicting components are interleaved across the board and a step has physically nowhere to go. What is worth discussing then is changing the layout, or splitting the two sides onto separate stencils. Forcing fragmented zones rarely yields stable print consistency.
Q6: What data does a step stencil order need?
Per /en/smt-stencil/, accepted input is Gerber RS-274X (paste, solder mask and outline layers), ODB++ or DXF. A step stencil additionally needs the grouping, packages and positions of the conflicting components so boundaries can be settled at review. Order quantity and quotation are confirmed after review of data, quantity and process requirements.