How to Reduce mmWave Insertion Loss — From Physics to Production Control

In the mmWave world (24–81 GHz and beyond), insertion loss is not a single number on a datasheet. It is the combined effect of materials, copper, geometry, and manufacturing tolerances – all stacked together along a real RF signal path.

For 77 GHz radar, 5G FR2, satellite links and high-density phased-array systems, every extra 0.1 dB of loss eats into detection range, link margin and beam accuracy. Reducing mmWave insertion loss is therefore not just “nice to have” – it is a core engineering requirement.

This guide summarizes how UltroNiu approaches mmWave insertion-loss control: starting from physics, and ending at production-grade process control and measurement.

1. What mmWave Insertion Loss Really Represents

Insertion loss is the attenuation a signal experiences as it travels through a real RF channel – not an ideal line in simulation. At mmWave, that channel includes:

• Dielectric material (Dk / Df, moisture, thickness tolerances)
• Copper conductors (roughness, plating profile, line geometry)
• Launch structures (connectors, probes, transitions)
• Vias, stubs and discontinuities along the way

An RF link budget only looks healthy if the total loss along this chain is under control. Once frequency climbs into 24–81 GHz, even millimeters of extra copper or small Dk shifts can visibly distort amplitude and phase.

2. Decomposing mmWave Insertion Loss into Engineering Terms

For practical engineering work, it is useful to think of total insertion loss as a sum of four contributors:

α_total = α_d + α_c + α_r + α_dis

• α_d – Dielectric loss 
  Dominated by the material’s loss tangent (Df) and operating frequency.
• α_c – Conductor loss 
  Driven by copper roughness, plating profile and skin effect.
• α_r – Radiation loss 
  Related to how well fields are confined or leak out of the PCB.
• α_dis – Discontinuity loss 
  Coming from bends, vias, connectors, layer transitions and any geometry changes.

At mmWave, none of these terms is negligible. “Optimizing” only one of them (for example, choosing an ultra-low-loss material) while ignoring the others leads to disappointing real-world results.

3. Dielectric Loss – Controlling Df and Its Stability

Dielectric loss tends to grow roughly in proportion to loss tangent (Df) and frequency. The basic engineering levers are:

• Choose materials for the actual operating band – Not just for a convenient 10 GHz datasheet line. For 77 GHz radar, Df @ 77 GHz and Dk stability across temperature and humidity matter more than a pretty 10 GHz number.
• Watch Dk / Df variation – Tight nominal values mean little if the material varies strongly lot-to-lot or across the panel. Phase error in phased arrays scales directly with Dk drift.
• Control water absorption – Moisture shifts Dk and Df. In outdoor systems (radar, telecom), a few tenths of a percent water uptake can noticeably change insertion loss and phase.
• Respect lamination windows – A material that looks ideal on paper can become unstable if pressed outside its recommended temperature or pressure profile.

In UltroNiu projects, material selection for mmWave always couples Df, Dk stability, moisture behavior and lamination process capability – not Df alone.

4. Conductor Loss – Copper Roughness vs Skin Depth

At mmWave, current flows in a very thin region near the copper surface. Skin depth in copper at tens of gigahertz is well below 1 µm. Typical etched copper roughness (R_a of 1.8–2.5 µm for standard ED copper) is several times that depth.

Engineering consequences:

• Rough copper “chokes” mmWave energy – The fields are forced to travel along a highly irregular surface, increasing effective path length and resistive loss.
• Standard ED copper becomes unacceptable – Loss can easily increase by 0.2–0.6 dB/cm compared with low-roughness copper at 77 GHz.
• Finish and plating matter – Nickel-based finishes, poorly controlled plating and aggressive etching can all worsen roughness seen by the RF field.

To reduce conductor loss, UltroNiu uses:

• HVLP / VLP or reverse-treated copper (RTF) for mmWave layers
• “Smooth-side-out” laminate orientation for RF signal layers
• Tight control of etching and surface-finish processes to preserve low roughness

5. Geometry – Microstrip, CPW, Stripline and SIW

The same material and copper can produce very different insertion loss depending on geometry. The main options in mmWave PCBs are:

• Microstrip 
  Simple, low cost and easy to probe. However, more fields leak into air, and radiation loss rises as frequency increases.
• Grounded coplanar waveguide (GCPW) 
  Better field confinement, excellent for 24–81 GHz. Gap width, ground spacing and solder-mask usage become critical design parameters.
• Stripline 
  Internally shielded routing with good EMI performance. Requires very tight dielectric-thickness control and stable stack-up for predictable impedance and loss.
• Substrate integrated waveguide (SIW) 
  Very low radiation loss and high Q, using via fences as a synthetic waveguide wall. Requires high via precision and consistent plating.

From a loss-reduction viewpoint, UltroNiu typically uses:

• Microstrip or GCPW for antenna feeds and external launches
• GCPW or stripline for critical internal RF distribution
• SIW for specific radar and SATCOM paths where maximum Q is required

6. Discontinuities – Vias, Launches and Transitions

Even if line segments look perfect, insertion loss can still be dominated by discontinuities:

• Via stubs – Long unused via tails act as resonant stubs. At 28 GHz and 77 GHz, even fractions of a millimeter matter.
• Layer transitions – Moving from microstrip to GCPW or stripline without controlled transitions introduces both reflections and extra loss.
• Connector launches – Poor launch design can easily add more loss than several centimeters of line.

Key control strategies:

• Use backdrilling or blind vias to remove stubs wherever possible
• Co-design RF footprints and launches with the connector vendor
• Simulate transitions as 3D EM structures, not as ideal ports
• Keep solder mask away from the highest-frequency sections of the RF launch

7. Manufacturing Levers – How the Factory Really Reduces Loss

mmWave insertion loss is not controlled only in the EDA tool – it is controlled on the factory floor. Critical production parameters include:

• Lamination consistency 
  Stable dielectric thickness, controlled resin flow and repeatable glass-weave effects.
• Copper and etch profile 
  Etch compensation and line-width control tuned for each copper thickness and material system.
• Surface finish 
  Choosing low-impact finishes (such as ENEPIG or carefully controlled ENIG) and avoiding unnecessary roughening of RF copper.
• Via formation and plating 
  Stable microvia drilling and plating to prevent voids, cracks and diameter drift, which all affect discontinuity loss.

UltroNiu treats these as engineering variables with defined process windows, not as “black-box” production steps.

8. Measurement – Δ-Loss Coupons and Correlation with Reality

At mmWave, measurement strategy is as important as design strategy. Without data, insertion-loss “optimization” is just opinion.

On production panels, UltroNiu uses:

• Impedance coupons – To confirm that geometry and stack-up hit the target impedance.
• Δ-loss (Delta-loss) coupons – To measure frequency-dependent loss over controlled line lengths, built with the same material, copper and process as the real RF path.
• Reference launches – Standardized probe or connector launches to remove launch uncertainty from coupon data.

These measurements are then correlated with EM simulation results. Once correlation is established, design teams can confidently explore trade-offs in geometry, material and stack-up with realistic expectations of actual insertion loss.

9. Engineering Checklist – How to Systematically Reduce mmWave Insertion Loss

A practical mmWave insertion-loss strategy should at least satisfy the following points:

• Specify Df and Dk at the actual operating frequency band, not only at 10 GHz.
• Specify copper roughness (HVLP / VLP / RTF) for RF layers – do not leave it implicit.
• Select line geometry (microstrip, GCPW, stripline, SIW) according to loss dominance and application.
• Eliminate or minimize via stubs and uncontrolled transitions with backdrilling or blind / buried vias.
• Control solder mask usage on RF lines, especially at mmWave launches.
• Model fabrication tolerances (thickness, width, roughness) in EM simulations, not ideal values only.
• Place impedance and Δ-loss coupons on the same panel as the real RF boards.

If any item in this checklist is missing, insertion-loss optimization is incomplete by definition.

10. Conclusion – Loss Must Be Engineered into Production, Not Just Simulated

mmWave insertion loss is where physics, design and manufacturing meet. Reducing it is not about a single “magic” material or a one-time simulation run – it is about consistent control of every element the signal touches.

At UltroNiu, mmWave RF boards are engineered with this full chain in mind: from Dk / Df and copper roughness, through geometry and transitions, to Δ-loss coupons and production SPC. The goal is simple:

The performance you see on day one must still be there after thousands of hours, in the real field, at full frequency.

Only then does a low-loss mmWave design move from theory to reliable, manufacturable reality.