Embedded Resistor and Buried Component Design
An embedded resistor is formed from a resistive layer inside the board rather than from a component placed on the surface. The technique removes parts from the surface, shortens the connection to the layer it terminates, and improves the behaviour of high speed nets by eliminating the stub that a surface component would create. It also changes the fabrication and the test flow, so it is a decision that has to be made early.
What Embedded Components Are
Several kinds of component can be formed inside a board. Resistors are the most common, made from a thin resistive foil laminated into the stackup, and capacitors can be formed between closely spaced power and ground planes using a high permittivity material. Inductors are possible but rare, because the values that can be formed are small.
buried resistor layers are patterned before lamination, in the same way that copper is patterned, and connected to the outer layers with vias. The value is set by the sheet resistance of the material and by the aspect ratio of the patterned element, so a wide short element gives a low value and a narrow long element gives a high value.
The main motivations are area, performance and reliability. Removing surface components frees space for routing, a terminator placed directly on an inner layer has no stub, and fewer solder joints means fewer opportunities for a joint to fail. The cost is a longer and more specialised fabrication flow with fewer suppliers.
Embedded Resistor Materials and Values
Resistive foils are characterised by their sheet resistance, quoted in ohms per square. A material with 25 ohms per square produces a 50 ohm resistor from a pattern two squares long, while a 1000 ohm per square material produces the same value from a pattern that is much shorter, which is easier to fit in a dense layout.
The practical range of values from a single material spans about two orders of magnitude, so a design that needs both very low and very high values requires either several materials or a compromise. Values below about ten ohms and above about a hundred kilohms are usually better served by a discrete component.
The material also has a temperature coefficient and a tolerance that are wider than those of a good discrete resistor. Typical tolerances are five to twenty percent before trimming, which is acceptable for a terminator and not acceptable for a precision divider. Where accuracy is needed, the resistor is either trimmed after fabrication or the design is arranged so that the ratio matters rather than the absolute value.
Tolerance, Trimming and Temperature
resistor tolerance has to be considered together with the matching between resistors on the same layer. Absolute tolerance may be twenty percent while the matching between two adjacent elements is much better, because they are formed from the same material on the same panel. Designs that rely on ratios rather than absolute values therefore benefit most from the technology.
Trimming is done by laser, either after fabrication or after assembly, and it can bring a value to within a fraction of a percent. The process adds cost and time and is usually reserved for the critical resistors in a design rather than applied across the board. Trimming also removes material, which changes the power handling of the element slightly.
Temperature behaviour is dominated by the material rather than by the geometry. The temperature coefficient is quoted in parts per million per degree, and a value of a few hundred is typical, which is worse than a good thin film discrete part. Where a design needs stability over temperature, the embedded resistor is used for the parts of the circuit where that does not matter.
Design Rules and Layer Stackup
The resistive layer adds thickness and a process step, so the layer stackup has to be planned before the design starts. The resistive layer is usually placed between two dielectric layers with copper above and below, and the whole arrangement is repeated for each buried resistor layer. The general sequencing of layers, described in our guide to layer stackup design, applies with the additional constraint of where the resistive material can sit.
The minimum feature size for the resistive elements is larger than for copper, because the material behaves differently during etching. Design rules from the fabricator should be obtained before the layout is started, and the element geometry should be checked against them rather than drawn to the copper minimum.
The elements also need a keep out area. Copper above or below a resistive element changes its effective value slightly through capacitive loading, and a plane placed directly over the element can affect both the value and the power handling. The keep out is defined by the fabricator and should be respected on every layer.
Via Connection and Routing
A buried resistor is connected to the outer layers by vias that land on the terminals of the resistive element. The via connection adds a small resistance of its own, which matters for low value elements, and it also adds inductance that matters at high frequency. Where the element is used as a terminator on a high speed net, the connection should be as short as the stackup allows.
Landing the via directly on the element is preferred over connecting through a short trace, because the trace adds both resistance and inductance in series with a component whose value is deliberately small. The design rules of the fabricator define the allowed overlap and the annular ring, and the rules in our guide to via design rules govern the drill and pad geometry.
Where the vertical connection is made through several layers, the cumulative tolerance of the stack becomes part of the resistor value. The fabricator can provide a nominal and a worst case figure, and for a low value resistor that figure should be included in the tolerance budget rather than treated as negligible.
Cost, Test and Rework
Cost is driven by the extra process steps and by the panel utilisation, not by the material alone. The resistive layer is laminated across the whole panel, so the cost is incurred whether or not the design uses much of it, and small volume runs carry most of that cost per unit.
Test access changes as well. An embedded resistor cannot be probed directly, so the value is verified through the nets that connect to it, and a fault inside the board cannot be reworked. That makes a design error more expensive, which is the main argument for prototyping the layer separately before committing to a full design.
Repair is limited to the outer layers. If an embedded resistor is found to be out of specification after assembly, the board is scrap, so the yield of the fabrication process matters as much as the design. A fabricator with experience of the material will quote a realistic tolerance and a realistic yield, and both should be part of the decision.
When Embedded Passives Make Sense
The technology pays off where the same small resistor is repeated many times, where a terminator must be placed without a stub, or where surface area is genuinely scarce. A design with a handful of resistors that could equally well be placed on the surface gains little and takes on a great deal of process complexity.
A good candidate is a high speed bus with series terminators on every line, or a dense array of pull up resistors where the surface area saved allows a smaller board. A poor candidate is a one off design with a handful of precision resistors, where the accuracy requirement outweighs the area saving.
If the decision is close, a prototype run on a small panel gives the answer. Building a test coupon with the intended element geometry alongside the first articles shows the actual tolerance, the temperature behaviour and the yield. Specialised fabrication processes are best validated with a coupon rather than with the first production order, and the documentation practices described in HDI CAM methods are a good model for what to include. The wider set of fabrication decisions that the technology interacts with is covered in our guide to PCB design and fabrication.
FAQ
How accurate can an embedded resistor be? Untrimmed, typically five to twenty percent. Laser trimming after fabrication brings the value to within a fraction of a percent, at additional cost and processing time.
Can embedded resistors replace all surface resistors? No. The available range, the tolerance and the power handling are all more limited, and rework is impossible. They are best used where the design benefits from the removal of a specific component.
Do embedded resistors improve high speed performance? They remove the stub that a surface terminator would create and shorten the connection to the layer being terminated, which matters most at the highest frequencies.
