110-145GHz PCB Calibration Boards — TRL/LRM Coupons and Connector Launch Design
Test and measurement equipment covering 110-145GHz and beyond is moving from lab waveguide setups toward repeatable coaxial test links using 0.8mm and 1.0mm connectors. That shift puts the PCB calibration board itself under scrutiny: at these frequencies, via stubs, panel-to-panel repeatability, and connector launch tuning determine whether a measurement is trustworthy at all. This guide covers what a TRL/LRM calibration coupon set needs, and how connector launch design changes above 110GHz.
Table of Contents
Wideband bias tees now specified to 145GHz and VNA test-cable platforms extending to 120GHz are a sign that sub-terahertz measurement is becoming a repeatable, coaxial-connector activity rather than something confined to waveguide test benches. The same trend shows up in the growing number of E-band and sub-100GHz MMICs entering satellite, backhaul, and test-equipment applications, each of which needs a chip evaluation board, a probe transition board, or a calibration standard before a single meaningful S-parameter can be reported.
For a board fabricator, this shifts the value proposition. At 110GHz and above, “impedance is within spec” is not a useful claim on its own — what a customer actually needs is a calibration and de-embedding structure built with enough repeatability that the measurement reflects the device under test, not the fixture around it.
Why Calibration Boards Are a Different Manufacturing Problem
A calibration or de-embedding coupon isn’t a functional RF circuit — it’s a measurement reference. Its entire value comes from being predictable and repeatable: the same coupon design, fabricated again six months later, needs to behave close enough to the original that a calibration performed with one panel stays valid when applied to data taken with another. That requirement pushes tolerances on substrate thickness, copper roughness, etch geometry, and connector launch tighter than most functional RF boards ever need, because any panel-to-panel drift becomes measurement uncertainty rather than a performance margin.
TRL and LRM Calibration Coupons
Thru-Reflect-Line (TRL) and Line-Reflect-Match (LRM) are the two calibration methods most commonly used to remove fixture and connector effects from a measurement at millimeter-wave and sub-terahertz frequencies, because they rely on physical transmission-line and reflect standards rather than idealized lumped-element models that break down at these frequencies. A TRL coupon set typically needs a Thru standard, one or more Reflect standards (open or short), and at least one Line standard of a different, precisely known length — sometimes several Line standards to cover a wider frequency range. LRM substitutes a known-impedance Match standard for one of the Line lengths, which can shrink the panel footprint at the cost of needing an accurate resistive standard.
What all of these standards have in common is that their calibration accuracy depends entirely on how precisely the fabricated line lengths, launch geometry, and impedance match the design intent — a calibration coupon with an off-spec line length doesn’t just perform slightly worse, it silently corrupts every measurement calibrated against it.
| Parameter | Requirement | Typical Value | Notes |
|---|---|---|---|
| Line length accuracy (TRL) | Precisely known, repeatable line length per standard | ±5μm or tighter | Off-spec length silently corrupts every calibrated measurement |
| Substrate thickness | Ultra-thin, low-loss laminate | ≤127μm (5 mil) class, per design | Thickness is a significant wavelength fraction near 145GHz |
| Copper roughness | Low-profile / reverse-treated foil | Ra ≤ 1μm class | Roughness loss scales unfavorably with frequency |
| Via stub length | Eliminated via back-drill or blind/buried via | Near-zero residual stub | Standard practice on calibration boards, not optional |
| Connector interface | 1.0mm (to ~110-120GHz) or 0.8mm (extended range) | Per target measurement band | Far less tolerant of launch error than 2.92mm/1.85mm |
| Board thickness/flatness | Tight tolerance matched to connector pin depth | ±1mil class or per connector spec | Variation shows up directly as launch mismatch |
| Solder mask keepout | Clear of RF line and connector footprint | Full keepout, no mask creep | Proportionally larger effect on narrow, thin-substrate lines |
| Panel-to-panel repeatability | Documented across production lots | Consistent S-parameters lot-to-lot | Not just a single qualification build |
The same discipline used for lower-frequency de-embedding structures, covered in our RF and microwave test fixture design guide, still applies at 110-145GHz — the difference is how much tighter every tolerance in that guide needs to be held.
Via Stub Control and Ground Return
Residual via stub length — the unused portion of a plated through-hole left below the signal layer it’s serving — is a well-known source of resonance and loss at lower millimeter-wave frequencies, and the problem only gets worse approaching 145GHz, where even a short stub can fall inside the resonant range of interest. Back-drilling or using blind/buried vias to eliminate the stub is standard practice on a calibration board, not an optional refinement, since an uncontrolled stub effectively adds an unintended element to every standard on the panel.
The ground via fencing discipline used to isolate RF sections on mixed-signal boards, described in our guide to RF and digital mixed-signal PCB stackup design, scales down directly to a calibration coupon: tight via pitch around the launch and along the transmission line keeps the ground return low-inductance and consistent from coupon to coupon.
Connector Launch Design for 0.8mm and 1.0mm Interfaces
Precision 1.0mm connectors typically cover measurements to around 110-120GHz, while 0.8mm interfaces extend usable range further, into the range these newer sub-terahertz bias tees and test cables are specified to. Both interfaces are considerably less forgiving of launch geometry error than the 2.92mm or 1.85mm connectors used at lower frequencies: pin depth, board thickness, and the transition taper from connector pin to microstrip or grounded coplanar waveguide all have to be tuned as a system, not adjusted independently.
Board thickness and flatness tolerance matter here in a way they don’t for a typical RF board — the connector’s mechanical interface assumes a specific board thickness to align the pin with the trace, and any board-to-board thickness variation shows up directly as launch mismatch at these frequencies.
Substrate and Surface Considerations
Calibration boards at these frequencies are almost always built on very thin, low-loss laminate, since dielectric thickness itself becomes a significant fraction of a wavelength as frequency climbs toward 145GHz. Copper roughness and etch sidewall control matter more here than the raw dielectric loss tangent in many cases, because surface roughness loss scales unfavorably with frequency and etch geometry directly sets the line’s characteristic impedance at these narrow trace widths. Solder mask keepout around the RF line and connector footprint also needs explicit control — a small amount of mask creeping onto the trace has a proportionally larger effect on a narrow, thin-substrate microstrip than it would on a wider lower-frequency line.
A calibration board that isn’t repeatable doesn’t fail obviously — it produces a calibration that looks clean but doesn’t transfer between panels, connector re-mates, or labs. That kind of error is expensive precisely because it’s invisible until two parties compare S-parameter data and can’t reconcile it. Panel-to-panel repeatability and documented connector launch tuning are worth specifying up front, not discovering after a calibration dispute.
Where This Fits Alongside Functional RF Boards
A calibration board rarely ships alone — it usually accompanies a chip evaluation board, a probe transition board, or a functional module built with the same high-frequency practices covered in our radar and high-frequency PCB manufacturing guide, just pushed to a tighter tolerance band. Delivering both together, with matched material lots and documented process parameters, is what actually lets a customer trust that the calibration applies to the board it was made to test.
What to Look for in a Manufacturing Partner
Building a usable 110-145GHz calibration board takes a narrower set of capabilities than general mmWave fabrication:
- Panel-to-panel repeatability documented across production lots, not just a single qualification build.
- Back-drilling or blind/buried via capability to eliminate residual stub at these frequencies.
- Connector launch tuning validated with measured S-parameter data for the specific 0.8mm or 1.0mm interface, not a generic launch template.
- Tight board thickness and flatness tolerance control matched to the connector’s mechanical requirements.
New on this subject
- antenna OTA and RCS test coupon PCB — far-field distance, mounting repeatability and thermal drift at 77 GHz
110-145GHz PCB Calibration Boards — Q&A
Common questions on TRL/LRM coupon design and connector launch for sub-terahertz test boards.
What is a TRL or LRM calibration coupon and why does a PCB fabricator need to make one?
TRL and LRM are calibration methods built from physical transmission-line and reflect standards fabricated as PCB coupons. The whole calibration's accuracy depends on how precisely those coupons match the intended line lengths, impedance, and launch geometry.
Why do 0.8mm and 1.0mm connectors matter above 110GHz?
1.0mm connectors generally cover measurements to about 110-120GHz, and 0.8mm extends further into the sub-terahertz range. Both are much less forgiving of launch geometry and board thickness variation than lower-frequency connectors like 2.92mm or 1.85mm.
What causes a calibration board to give inconsistent measurements between labs?
Panel-to-panel differences in line length, substrate thickness, copper roughness, or connector launch geometry can prevent a calibration from transferring cleanly between boards — and since each board can look individually valid, the mismatch usually only shows up when two data sets are compared.
Can standard PCB tolerances support measurements above 110GHz?
Not usually. Calibration boards need tighter control on thickness, copper roughness, via stub length, and flatness than typical functional RF boards, since any drift here becomes measurement uncertainty rather than just a performance margin.
Need a 110-145GHz Calibration or Test Board?
We build TRL/LRM calibration coupons and de-embedding boards for 110-145GHz test programs, with back-drilled vias, tuned 0.8mm/1.0mm connector launches, and documented panel-to-panel repeatability.
Get a Quote → WhatsAppRequest a PCB Quote
Upload your Gerber ZIP file and project requirements. Our engineering team will review your PCB material, stackup, impedance needs, surface finish, and production quantity before quoting.
Please prepare:
- Gerber files in ZIP format
- PCB material or stackup requirements
- Controlled impedance notes if available
- Prototype or batch production quantity