The primary sections of this guide include rigid, flexible, rigid-flex and HDI, as well as a limited number of material and performance constructions. The right type is the one that addresses a real constraint, a stable flat assembly, repeated bending, rigid-to-flex transition, very dense interconnects, heat, current, or high-frequency behavior. A more complex type does not necessarily make a better board.
The habit that wastes a first inquiry is stacking labels — “HDI rigid-flex high-frequency” — before the outline, bend, layer, and signal needs are known. Extra constructions add process steps, stack-up decisions, and assembly constraints. Once the family is chosen, PCB fabrication by board type starts from drawings and stack-up notes, not from a catalog slogan.
Rigid PCBs, the Default Construction Among Types of PCBs
The standard glass-reinforced laminate that remains flat during assembly and in the enclosure is a rigid board. Multi-layer, double-sided, and single-sided builds are all supported. In most cases, begin with most products unless the bending, moving joint or extreme density requires another family.
Importance of rigidity is simplicity of mechanism and process. These components rest on a levelled plane. Fixtures, stencils and test access are simple. The stack-up may be easy and simple, such as a single copper layer with a sparse control board or it can be complex with multiple layers of planes and impedance controlled traces for a denser product. Complexity of routing does not, by itself, move the design out of the rigid family.
The wrong first shot is to consider “rigid” as unsophisticated. Because of its small outline, it is not a reason not to leave this construction. Not a crowded connector field either until through vias and conventional multi-layer stack-up are proven to fail.
When rigid is enough, and when a multilayer rigid board is the real need
Differentiate between “rigid” and “multilayer.” The mechanical family is called rigid. Multilayer refers to the number of layers of copper to laminate and register by the fabricator. A product may require many layers for routing, return planes, impedance and/or be a rigid board. That is a stack up with no change in construction family.
If the circuit will not bend when used, and there is room for a flat enclosure, go with rigid. Then have students ask if the nets and planes can be carried by two layers without crowding. If they can’t, add layers. Avoid adding flex as the outline has a tight look to it. A smaller board with a denser rigid stack-up is typically less expensive, and easier to assemble than a flex circuit that does not require flexing.
The mistake people tend to make is that a request for “flex” is actually for a denser rigid stack-up. Next, the flex materials, coverlay and bend rules are set for a part to be housed in a specific structure. Additional process steps are passed down to the fabricator. The assembler is given rules for handling a product that the product itself didn’t need.
When a flat rigid construction is the right starting point, rigid PCB fabrication is the next conversation: layer count, material direction, finish, and what is still open. Leave HDI, metal-core, and high-frequency notes off that first package unless those constraints are already proven.
Flexible and Rigid-Flex PCBs
Rigid vs flex vs rigid-flex PCB is not one decision, it is two decisions. Flex is used when the circuit needs to bend, instead of a board-plus-cable set-up. Rigid flex: The area of components is rigid and flex is used for the transition only. Changes in materials, coverlay, stiffeners, bend design and assembly of the board. A compact version of FR-4 is not included.
Consider the comparison a mechanical issue first. Flex is typically not the appropriate family if there is NO movement. If it is moving, determine if the whole circuit moves or a joint between rigid islands moves. The second question is what makes the difference between a flex circuit and a rigid-flex board.
Flex when the circuit must bend
Specify flex when the copper itself has to move. There are two cases and they are not the same process problem.
A one-time bend is a bend that is folded into a housing. The circuit is created one time and remains. The bend radius, fold location and placement of copper on the outside or inside of the fold remain to be designed, but the number of cycles is low.
Repeated movements: a hinge, a sliding module, a door, a wearable joint – all of these are examples of dynamic bending. These are included in the product specification such as cycle life, copper type, coverlay instead of solder mask and the keep-out around the bend. Polyimide is the typical dielectrics material since it allows for bending without cracking as does a rigid laminate with glass reinforcement. Coverlay is a flexible alternative to solder mask and is a film that is laminated over the copper with holes at the pad area. Stiffeners (additional material under a connector or an add-on island) provide a local rigidity to the appropriate areas of the circuit which is not found elsewhere.
The process cost is an actual cost. Flex circuits do not handle like FR-4 on a standard SMT line. They require some type of carrier or stiffener during reflow. Copper cracks or lifts pads at tight bends at a component or connector area. The typical first assembly error if the flex can be panelized, printed and mounted in the same fixtures as a rigid board is when the flex is presumed to be capable of all of those.
In cases where the product requires only two rigid boards to be connected to each other, two connectors and a harness might be the more straightforward choice. These connectors become the reliability issue, the space is limited, and/or the motion needs to reside in the circuit, and Flex comes into play.
Rigid-flex when only the joint should flex
It’s not as broad as the marketing tagline suggests when it comes to when to use a rigid-flex. Use it to keep component areas rigid, and only allow movement between them — like in parts, connectors, assembly fixtures, etc. The rigid sections take the packages. The transition, and not a second product, is the flex section.
The transition zone is the design. Layers that are only in the rigid region must terminate properly. Continued layers in the flex area need to withstand the bend. The stack-up isn’t one number; it’s a map of where which layers exist. The rigid-to-flex shoulder is as important as the bend radius in the middle of the tail. The order of assembly is important, since the board is no more a flat coupon to the entire line. Access for soldering, inspection and test must be drawn, not assumed.
The alluring shortcut is to select rigid-flex to save two connectors. This might be a good recommendation. It is an incorrect call if the bend radius, the layer transition and the assembly access has not been drawn. A rigid-flex RFQ that contains no transition detail, and includes a product photo, requires the fabricator to develop the mechanical design.
Rigid-flex PCB designs are not the same as rigid PCB designs; only after the transition has been specified should the design move to rigid-flex PCB manufacturing. It is not recommended to combine labels “HDI” and “high frequency” on the same note unless both the conditions are independently satisfied.
Designs that combine rigid component areas with a flexible joint belong in rigid-flex PCB manufacturing only after that transition is specified. Do not add HDI or high-frequency labels to the same note unless those constraints are independently true.
HDI PCBs and Microvia Density
HDI is a density construction, that is, finer features, laser-drilled microvias, and sequential build-up to allow parts to sit closer together. It’s not a flair for prestige, it’s a choice. A multi-layered board is not necessarily a high density interconnect board.
Before the RFQ, it is best to compare the HDI PCB with the multilayer PCB. A conventional multi-layer board has electrically drilled through vias that pass through the entire board (or most of it). These vias take up routing space in each layer through which they travel. HDI includes laser-microvias to connect adjacent build-up layers and it is possible to remove or stack these via holes under a fine-pitch package so that holes do not form a forest. The additional process steps (laser drilling, sequential lamination, closer registration) come at that density cost.
If parts can be routed with through vias and a standard multilayer stack-up, then we have added cost and process risk to no benefit to the product. Honest start test is that.
When density justifies the process
Do not use a package name. Typically, fine-pitch BGAs, dense fields of connectors or parts that need to be on both sides with minimal space between them for fan-out. A conventional multilayer rigid board is the right choice if there is no need to change layout due to failure of the escape routes connecting through vias. When the only method of accessing the inner layers is a microvia that is only buried or stacked beneath or beside the package, then the fabrication note must state this. If an RFQ does not have a via structure, there is no build-up sequence, and there is no indication of what packages that the RFQ is choosing, the word “HDI” is not a specification.
Never use the number of layers as a measure for HDI. The 8 layers are still a “conventional multilayer board”. Sequential build-up of four layers containing microvias can be HDI. The process is not determined by how awesome the stack is on a slide, but on the other hand by the interconnect method.
The pitfall is using HDI as a quality claim. It does not make a sloppy stack-up more reliable. It makes a dense layout fabricable. When microvias are actually required, HDI PCB fabrication is the construction to quote — with the via structure and build-up intent written down, not implied.
The “And More” Group: Material and Performance Types
These are not second full taxonomy. They alter a solid board (and in some cases a more specialized structure) to adjust heat, current, temperature or signal characteristics. Once the mechanical family (rigid, flex or rigid-flex) is decided upon, pick one of them.
Metal-core and heavy copper
The key differences between Metal-core PCB vs FR-4 are thermal. When an LED array or a power device is attached to an aluminum or other metal-core board, a thin layer of dielectric is placed between the device and a metal plate attached to the back of the board, allowing heat from either the LED array or the power device to pass through the copper and into a heatsink or chassis. When heat isn’t particularly high and layout allows it to be spread out by the use of copper and vias, FR-4 can be the appropriate answer. Don’t have to check to see if a lighting board is a metal-core board. The metal-core also alters the way the board is used, and drilling and routing is different, so it’s not a drop-in replacement for a standard, rigid panel.
Heavy copper PCB uses are decorative, not current and thermal. Thicker Copper means more current and changes the etched, plated and warmth conduction across the board. Indicate it if it is necessary for the current or the thermal rise. By stating it, because “a power product should look robust”, you add an “etch and register” requirement to the circuit that it does not require. Copper weight should be entered into the fabrication note as a requirement of current, not as a factory slogan.
High-Tg, high-frequency, and high-speed
A high-Tg material has the ability to withstand a higher process/operating temperature. It is not a cure-all for a poor thermal design. High-Tg is a material conversation if the board experiences lead-free reflow, a hot enclosure or a power stage that causes the laminate to heat up. May not be necessary if board is cool and process window is ordinary.
For high frequency/high speed boards, appropriate low loss or controlled-impedance materials, a defined board stack-up and board geometry that can be controlled by the fabricator must be used. These are the fundamental requirements for high speed PCB stack-up: layer sequence, impedance targets, and frequency or edge-rate the circuit will run at. If a laminate family name is written “high-speed” on an RFQ without frequency, impedance target or layer order, the fabricator has to make an educated guess about the electrical design. A brand name is no stack-up. Capability MUST be verified on the project, not taken from a type label, if it is required for a given laminate and construction.
A Selection Check Before the RFQ
There are only a few options for choosing a type of PCB, not a second essay. To anyone who says “the best construction” you reply these questions:
- Are there any parts of the circuit that bend while in use? Yes, if so is it one-time or repeated?
- Does the bend serve just as a transition between stiff parts of components? If so, then rigid-flex is the family to look to. Flex is the family if the whole circuit flexes.
- When no bending occurs, will a conventional multilayer rigid board be able to route the parts with through vias? If yes, stay rigid. HDI is not a mechanical option; it is a process solution on a rigid board that requires the buried/stacked microvia fine-pitch packages.
- Is it heat, current, temperature or speed of signals? These refer to metal-core, heavy copper, high-Tg or a high-frequency / high-speed material and stack-up — on top of the family already selected.
- Will the deliverable be a bare board or assembly? Flex and stiff flex change carriers and inspection access. State this in the package.
Submit choice of construction, direction of material, finish, and a note as to what is still open. Before production, a partner from Guangdong PCB and PCBA can work based on the review of the fabrication data. NEVER request a supplier to select the type from a product photograph. The enclosure is seen in the picture. Does not display the bend, the via structure or the impedance target.