Loss Tangent: Design Rules and Process Limits

Two numbers on a laminate datasheet decide how much a high speed channel loses: the dielectric constant and the loss tangent. The first sets the impedance and the propagation velocity, the second sets how much energy the material absorbs. Choosing a laminate is therefore a matter of reading those two figures at the frequency the channel actually uses, and of understanding that they change with frequency.

What the Two Numbers Describe

The dielectric constant, often written as Dk, describes how the material stores energy in an electric field. It sets the capacitance per unit length, and therefore the impedance of a trace of a given width and the speed at which a signal travels. A higher value means a narrower trace for the same impedance.

The loss tangent, written as Df, describes how much energy the material dissipates as heat. It is the material contribution to insertion loss, and it grows in importance as frequency rises. At low frequencies conductor loss dominates, while at high frequencies the dielectric loss becomes the larger term.

Frequency Dependence

Both values vary with frequency, and a datasheet that quotes a single figure is quoting it at one measurement frequency. A material specified at one megahertz may behave noticeably differently at ten gigahertz, and the difference is not a small correction for a wideband channel.

Selection should therefore be based on the value at the frequency of interest, and the supplier should be asked for a curve rather than a point. Where the channel spans a wide band, both the low and high ends matter, and our article on high frequency laminates describes how the curves are compared.

Insertion Loss and Reach

Insertion loss is the total attenuation of the channel, and it combines conductor loss with dielectric loss. The two scale differently: conductor loss rises with the square root of frequency, while dielectric loss rises almost linearly. That difference is why a material change helps most on long channels at high data rates.

The practical consequence is a change in reach. A link that is limited by loss can be extended by lowering the loss tangent, or it can be run at a lower rate. The trade is cost, because low loss materials are more expensive and often harder to process. Our article on high speed design rules sets out how the budget is assembled.

Impedance and Geometry Coupling

A change of dielectric constant changes the geometry needed to hold a target impedance. A material with a lower Dk requires a wider trace for the same impedance on the same dielectric thickness, or a thinner dielectric for the same trace width. Both affect routing density and layer count.

This coupling is why the laminate and the stackup have to be chosen together rather than in sequence. A laminate change late in the design invalidates the stackup, the impedance calculation and the test coupon, and the resulting delay is usually longer than the time saved by choosing the material quickly.

Processing Differences

Low loss materials are often filled with ceramic or use a different resin system, and both change the fabrication process. They can be more brittle, they may need different drilling parameters, and they may require a different desmear chemistry. Lamination cycles also differ, which affects how many times the material can be pressed.

Those differences matter for a multilayer board with sequential lamination. A material that works well in a two layer test vehicle may behave very differently when it has been through three lamination cycles. The processing consequences are set out alongside the material data in our article on laminate material properties.

Cost and Where It Is Worth Paying

Low loss laminates cost more per square metre and often carry a longer lead time. The money is well spent on the layers that carry the highest speed channels and on long routes, and it is wasted on layers that carry only power and slow signals.

A mixed stackup using a low loss material on the critical layers and a standard FR4 equivalent elsewhere is common, and it can capture most of the benefit at a fraction of the cost. The key is to know which layers carry the channels, which is a signal integrity question rather than a purchasing one.

Verification

Material properties should be verified on a coupon rather than assumed from the datasheet. A test coupon with known geometry measured with a time domain or frequency domain instrument gives the effective dielectric constant and the loss of the finished board, including the effect of the lamination process.

Measuring the finished board also captures the variation across a panel, which can be larger than expected for filled materials. Where the channel is marginal, that variation belongs in the design margin rather than in a hope that the nominal value holds.

Checks Before Release

A change that is not recorded is a change that cannot be explained when the result moves, which is why the record is part of the process. The checks that matter are the ones performed on the product rather than on a sample kept for the purpose, because a coupon that travels with the panel is the only evidence about that panel.

Where an operation cannot be verified afterwards, it has to be controlled during the operation, and that control has to be visible in the record. Consumables have a life measured in cycles, and the replacement point should come from the measurement rather than from a failure.

The sequence of operations is part of the specification, because a different order produces a different result from the same steps.

Verification and Records

A result that cannot be reproduced is not a result, and reproducibility should be demonstrated rather than assumed.

FAQ

Can FR4 be used for high speed? Yes, at moderate rates or on short channels. Its loss tangent is higher, so the reach is shorter before equalisation becomes necessary.

Does a lower loss tangent always help? It reduces dielectric loss, but conductor loss is unaffected. Where the channel is dominated by conductor loss, a material change delivers little.

How much does dielectric constant vary between suppliers? More than most designers expect, particularly for filled materials. A qualification sample from the actual supplier is the only reliable figure.