1. The reason for hybrid is not sophistication, it is spending performance where it is needed
In a 12-layer board, maybe four layers carry the high-speed differential pairs. The rest are low-speed control, power and ground. Switching the whole board to low-loss laminate for the sake of those four layers multiplies the cost by the full board area, and most of the performance you paid for lands on power planes that cannot use it.
That is the problem a hybrid stackup solves: put the expensive laminate only on the layers that need it.
This is not a theoretical proposition here. The FAQ on our high-frequency PCB page states it directly for one Isola pair: I-Speed and I-Tera MT40 "can also share one stackup, with I-Tera MT40 on the critical signal layers and I-Speed on low-speed and power layers to balance cost." Layered material assignment is something already in practice, not a feasibility question needing proof.
Note the scope of that statement, though. It describes I-Speed and I-Tera MT40 — one pair from one supplier, in the same material family. Same supplier and same family means their processing systems are close, and that closeness is the precondition for sharing a stackup. Reading it as "any two laminates can be mixed" is the first misconception this article takes apart.
> Division of labour with the product page: the multi-layer PCB page owns the board-type overview and the engineering-review entry point, answering "what range can you build." This article answers "how do I write a hybrid stackup, and what does writing it lock in." Each links to the other.
2. Why a hybrid stackup is not "pressing two laminates together"
Open the parameter table on our materials library and the span alone shows where the problem sits:
| Material | Family | Tg (°C) | Td (°C) | Dk | Df |
|---|---|---|---|---|---|
| Ventec VT-42 | FR-4 | 140 | 310 | 4.3 | 0.015 |
| ITEQ IT180A | FR-4 | 175 | 345 | 4.4 | 0.015 |
| Isola I-Tera MT40 | FR-4 | 200 | 360 | 3.45 | 0.0031 |
| Panasonic Megtron 6 | FR-4 | 185 | 410 | 3.71 | 0.002 |
| Rogers RO4350B | Hydrocarbon/Ceramic | 280 | 425 | 3.48 | 0.0037 |
| Rogers RO3003 | Ceramic/PTFE | N/A | 500 | 3.0 | 0.0013 |
| Isola P95/P25 | Polyimide | 260 | 416 | 3.76 | 0.0179 |
(Values quoted as published on the materials page, which carries its own qualifier: these are typical values, and the manufacturer datasheet governs for complete and accurate data.)
Tg spans 140 to 280, Td spans 310 to 500, and the families run from FR-4 through to PTFE composite. The point of these numbers here is not to calculate anything. It is to establish one fact: these are not the same kind of thing. Different families have different processing systems, so "which two laminates can coexist in one press cycle" is not a table lookup. Every combination needs engineering review.
That is the fundamental difference between a hybrid and a single-material stackup. A single-material stackup asks "is it enough." A hybrid stackup asks "are they compatible." The second question has no general answer.
Among our existing lamination content, multilayer lamination registration control covers how offset accumulates in a single-material system, and lamination void and delamination prevention covers how voids and delamination arise. Both still hold in a hybrid scenario. They simply sit behind one additional precondition: confirm the material combination is pressable first, then discuss registration and voids.
3. A hybrid stackup moves three things at once
Hybrid does not only change the material column. The moment a second laminate enters the stackup, the following three things shift together, and each of them sits on a different page's published scope.
First: finished thickness and its tolerance. Available thickness options differ by material, so the thickness combinations a hybrid build can produce are fewer than a single-material build. The multi-layer PCB page lists 0.4–6.0 mm; the capabilities overview lists 0.2–6.0 mm as the site-wide process range. Tolerance per our manufacturing tolerance page is ±10% or ±0.005" (≈0.13 mm), whichever is greater. With hybrid, confirm the target thickness can actually be composed from the available material thicknesses before discussing tolerance. Reversing that order produces "tolerance agreed, thickness unbuildable."
Second: impedance. Different layers in a hybrid stackup have different Dk — the table above ranges from 3.0 to 4.4 — which means the same trace width computes to a different impedance depending on the layer. The multilayer product page states ±5% (±10% standard), and the high-frequency product line is likewise ±5%. The point is not the number but that the reference plane has to be explicit: which layer a differential pair references, and which dielectric sits between that plane and the signal layer, determines the inputs to the impedance calculation. A hybrid stackup without stated reference planes turns the impedance requirement into an unexecutable sentence. For what each tolerance class actually requires, see impedance tolerance classes.
Third: layer-level traceability. This is the most underestimated item in hybrid work. Among the ten in-process checkpoints on our quality assurance page, checkpoint 05, multilayer lamination, is specific: prepreg and core assignments are re-verified by data-matrix before pressing, and after pressing every production panel is measured for finished thickness and logged to the job file for registration and dielectric traceability.
On a single-material board, that re-verification confirms thickness and count. On a hybrid board, it also has to confirm that the laminate on a given layer is the one that belongs there. With two base materials in the stackup, putting the wrong material on a layer is a different failure from putting the wrong count — and checkpoint 04, inner layer AOI, is immediately followed by lamination, after which inner-layer defects cannot be reworked. Wrong material on a layer belongs in that same irreversible class.
So a hybrid stackup drawing has to give that re-verification something to check against. Naming the material and part number layer by layer is not a formality. It is the precondition for checkpoint 05 to run at all.
4. The four inputs an RFQ has to carry
The high-frequency PCB page states the RFQ requirement in one actionable sentence: send the stackup, target impedance and the exact Isola material callout with your enquiry, and engineering will confirm manufacturability before quoting. For a hybrid stackup that expands into four items:
- The full stackup, with material and part number on every layer. Not "high-frequency laminate plus FR-4," but the specific grade on each layer. Section 3 explains why: the pre-press re-verification depends on it.
- Target impedance and reference plane. For each controlled group, give the target value, the tolerance class (±5% or ±10%) and the reference plane. Without the reference plane, a hybrid impedance requirement cannot be acted on.
- Finished thickness and how tolerance is judged. Give the target, and state that tolerance follows the manufacturing tolerance page rule — ±10% or ±0.005", whichever is greater. Which of those two is greater changes with thickness, so cite the rule rather than a single number.
- The source for each material's Dk/Df. The materials page already notes its figures are typical and that the manufacturer datasheet governs. If your design is sensitive to a particular Dk, state which datasheet it was calculated against, so engineering and design are not working from two different sources.
With those four present, engineering review can return "buildable / needs adjustment / recommend a different approach." Missing any one of them, what comes back is a list of questions rather than a quotation.
5. Three signals to drop the hybrid approach
Hybrid has boundaries. In these three situations it is the long way round:
Signal one: the whole board needs low loss. If high-speed links are spread across most layers and few low-speed layers remain, hybrid saves limited area while buying a material combination that has to be reviewed item by item. Going fully high-frequency is cleaner. High-frequency PCB material selection and Isola material selection by frequency handle selection in that direction.
Signal two: something has to bend. If part of the motivation for hybrid is "one section needs to flex," that is a construction problem, not a material-mixing problem, and it belongs in rigid-flex. See the rigid-flex PCB page. Dropping polyimide into a rigid stackup to buy bend capability is not the same thing as rigid-flex.
Signal three: the high-frequency region is a small corner of a large board. Hybrid is done at panel level — every press cycle for that large board is scheduled around the most demanding material combination in it. If the high-frequency area is one corner, splitting it into a separate board with an interconnect is often more appropriate than making the whole board go through the hybrid flow. Whether that works depends on the mechanical and interconnect scheme, and it is an item for engineering review.
One case often mistaken for hybrid is mixed copper weight. 1oz and 6oz coexisting in one stackup mixes copper, not base material, and the constraints are entirely different — see heavy copper multilayer mixed-weight stackup. The two can occur together, but they get reviewed separately.
6. What goes on the drawing
Compressed into drawing language, a hybrid stackup needs at least these lines:
- Stackup table: layer number / layer type (signal, power, ground) / copper weight / material grade / nominal dielectric thickness
- Finished thickness: target value plus "tolerance per SKPCB manufacturing tolerance page (±10% or ±0.005", whichever is greater)"
- Controlled impedance table: net group / target impedance / tolerance class / reference plane
- Material data source: datasheet revision behind each grade's Dk/Df
- Acceptance class: IPC-A-600 Class 2 or Class 3 (both listed on the capabilities overview)
- One note: this is a mixed-material stackup — engineering review to confirm pressability of the combination and achievability of the thickness build
That last line earns its place. It hands over the fact that this is a hybrid explicitly, instead of leaving engineering to discover it from the material column. A hybrid stackup processed as a single-material board is the most common source of rework on this kind of project.
For overall stackup planning method see the multilayer PCB stackup guide; for how layer count gets decided see the 4/6/8 layer comparison. To submit a hybrid stackup for review, use the quote entry.
7. FAQ
Q1: Can any two base materials be mixed in one stackup?
No. The "share one stackup" statement on our high-frequency PCB page applies to I-Speed and I-Tera MT40 — one pair from one supplier in the same family. Whether a cross-family combination (FR-4 with PTFE composite, for instance) is pressable has to be confirmed by engineering review for that specific combination. There is no general answer.
Q2: Does hybrid make ±5% impedance tolerance unachievable?
Both the multilayer and high-frequency product pages publish ±5% (±10% standard). Hybrid does not remove either class, but it increases input complexity: Dk differs by layer, so reference planes must be stated per group. Whether ±5% is achievable depends on the specific stackup and is an item for engineering review.
Q3: Is finished thickness as free on a hybrid board as on a single-material one?
Not entirely. Available thicknesses depend on each material's stock options, so a hybrid build can compose fewer thickness points. The 0.4–6.0 mm on the multilayer page and 0.2–6.0 mm on the capabilities overview remain the published ranges, but whether a specific target thickness can be composed from your chosen materials needs confirming first.
Q4: Why must the part number be on every layer instead of just "high-frequency laminate"?
Because checkpoint 05 on the quality assurance page requires prepreg and core assignments to be re-verified by data-matrix before pressing. That check compares against a specific material; "high-frequency laminate" cannot be matched to an incoming material record. And after pressing, inner layers cannot be reworked, so wrong material on a layer is an irreversible error.
Q5: Is hybrid the same as mixed copper weight?
No. Mixed copper weight means different copper thicknesses on different layers of one stackup (1oz alongside 6oz), and its constraints sit in etching and fill. Hybrid means different base materials, and its constraints sit in material compatibility and lamination. Both can appear on one board, but they are reviewed separately.
Q6: Can you quote before the stackup is fixed?
We can discuss it, but a hybrid quotation depends on the stackup. The high-frequency PCB page explicitly asks for the stackup, target impedance and exact material callout together, with manufacturability confirmed before quoting. If the stackup is open, send the design constraints instead — frequency range, link length, cost target — for a material and stackup recommendation first, then move to quotation.

