Seamless flexible waveguide is the non-twistable option in the ranges compared here; seamed, twistable waveguide allows one flange to rotate relative to the other. Choose between them by the required movement, power, bend radius and pressure. Do not assume that a seam automatically increases insertion loss: Dolph Microwave lists identical loss per metre for both families at all 17 catalogue sizes, while its seamless average-power ratings are about 1.7 to 3 times as high.
For example, the published WR-90 ratings are 3 kW average for seamless and 1 kW for twistable, both at 0.40 dB/m. Those figures describe this manufacturer’s range, not a universal rule for every flexible waveguide. A short twistable section still has to withstand the power reaching it.
Table of Contents
Construction and movement
The practical distinction is bending versus twisting. A non-twistable assembly can accommodate a bend but should not be used to rotate a flange about the guide axis. A twistable assembly supports axial rotation within its specified limits. MIL-DTL-287J distinguishes class 1 twistable, class 2 non-twistable and class 3 extra-flexible twistable assemblies; its stress definitions include torsion for twistable assemblies, but not for non-twistable ones. [1]
Seamed flexible guides commonly use wound metal strip with interlocking contacts. “Seamless” describes a construction, whereas “non-twistable” describes a movement capability; neither term alone proves a military class or qualification. There is also a naming conflict in the linked seamless product page: its material description includes “Helically wound silver-plated copper alloy.” Ask for the actual core construction and permitted movement on the quoted part’s drawing rather than resolving that conflict from the page title.
| Requirement | Seamless non-twistable family | Seamed twistable family |
|---|---|---|
| Flange rotation | Do not impose axial twist | Use the specified twist angle and active length |
| Average power | Higher published ratings in this range | Check the rating even for a very short section |
| Insertion loss | Same catalogue dB/m values as the twistable family | Same catalogue values do not prove identical installed performance |
| Tight routing | Class 2 has smaller minimum radii than class 1 in the compared bands | Allow more space for class 1; consider class 3 if rotation and tight bending are both needed |
| Pressurisation | Catalogue lists 30 PSIG; confirm the assembly rating | No pressure column in the compared catalogue; obtain a rating |
| Repeated motion | Requires an appropriate flexure-life rating | Requires repeated-twist and flexure ratings, not just maximum torsion |
Catalogue ratings by size
The twistable model table and seamless product table supply the values below, checked on 26 October 2026. Bands are reproduced as manufacturer-listed operating ranges, not as a separate EIA band standard. Power means average power; it is not a peak-power rating. The ratio is seamless power divided by twistable power, rounded to one decimal place.
| Waveguide size | Band (GHz) | Insertion loss (dB/m), both types | Seamless average power (kW) | Twistable average power (kW) | Power ratio | Twistable max torsion (deg/m) |
|---|---|---|---|---|---|---|
| WR-650 | 1.13–1.73 | 0.12 | 20 | 10 | ×2.0 | 5 |
| WR-430 | 1.72–2.61 | 0.15 | 20 | 10 | ×2.0 | 5 |
| WR-340 | 2.17–3.30 | 0.15 | 20 | 8 | ×2.5 | 15 |
| WR-284 | 2.60–3.95 | 0.15 | 10 | 6 | ×1.7 | 46 |
| WR-229 | 3.22–4.90 | 0.17 | 8 | 4 | ×2.0 | 85 |
| WR-187 | 3.94–5.99 | 0.24 | 6.5 | 3 | ×2.2 | 129 |
| WR-159 | 4.64–7.05 | 0.22 | 6 | 2.5 | ×2.4 | 170 |
| WR-137 | 5.38–8.17 | 0.25 | 5 | 2 | ×2.5 | 205 |
| WR-112 | 6.57–9.99 | 0.30 | 4 | 1.5 | ×2.7 | 262 |
| WR-90 | 8.20–12.50 | 0.40 | 3 | 1 | ×3.0 | 308 |
| WR-75 | 9.84–15.00 | 0.50 | 1.5 | 0.75 | ×2.0 | 360 |
| WR-62 | 11.90–18.00 | 0.80 | 1 | 0.4 | ×2.5 | 436 |
| WR-51 | 14.50–22.00 | 1.10 | 0.5 | 0.2 | ×2.5 | 445 |
| WR-42 | 17.60–26.70 | 1.20 | 0.3 | 0.1 | ×3.0 | 510 |
| WR-34 | 21.70–33.00 | 1.50 | 0.2 | 0.1 | ×2.0 | 510 |
| WR-28 | 26.30–40.00 | 2.00 | 0.15 | 0.075 | ×2.0 | 510 |
| WR-22 | 32.90–50.10 | 2.70 | 0.05 | 0.025 | ×2.0 | 510 |
The twistable page’s headline gives 3.94–50 GHz and 4 kW, whereas its model rows extend from WR-650 to WR-22 and include higher power values. The table above preserves the model-level information for comparison. Obtain written confirmation of the exact model, length, temperature, pressure and motion conditions before treating a row as an assembly guarantee.
Bend radius and pressure requirements
The following values are requirements from MIL-DTL-287/3F for class 1 and MIL-DTL-287/4G for class 2, both dated 14 April 2009. They are not measured properties of the catalogue products. [2] [3]
| Band (GHz) | Twistable class minimum radius E / H (in) | Non-twistable class minimum radius E / H (in) | Twistable class pressure (PSIG) | Non-twistable class pressure (PSIG) |
|---|---|---|---|---|
| 1.12 to 1.70 | 24.12 / 48.08 | 13.02 / 27.07 | 2 | 15 |
| 1.70 to 2.60 | 12.39 / 25.95 | 9.51 / 14.00 | 4 | 20 |
| 2.20 to 3.30 | 11.14 / 19.64 | 5.00 / 9.75 | 5 | 25 |
| 2.60 to 3.95 | 6.88 / 14.38 | 4.50 / 6.38 | 15 | 30 |
| 3.95 to 5.85 | 4.94 / 12.04 | 3.00 / 5.50 | 30 | 30 |
| 5.85 to 8.20 | 3.88 / 9.25 | 2.50 / 5.00 | 30 | 30 |
| 7.05 to 10.0 | 3.57 / 8.38 | 1.75 / 3.49 | 30 | 45 |
| 8.20 to 12.40 | 3.38 / 7.51 | 1.38 / 3.00 | 30 | 60 |
| 12.4 to 18.00 | 2.75 / 5.75 | 1.25 / 2.00 | 30 | 60 |
| 18.0 to 26.5 | 2.33 / 4.50 | 1.13 / 1.25 | 30 | 60 |
At 8.2–12.4 GHz, class 1 needs an E-plane centreline radius of 3.38 in, or 85.9 mm, against 1.38 in, or 35.1 mm, for class 2. Their H-plane minima are 7.51 in and 3.00 in, or 190.8 mm and 76.2 mm. These conversions use exactly 25.4 mm per inch. A 50 mm radius therefore clears the class 2 E-plane minimum, but not its H-plane minimum or either class 1 minimum.
The pressure figures also depend on the band. Class 2 specifies 15 PSIG against class 1’s 2 PSIG at 1.12–1.70 GHz; at 8.2–12.4 GHz the figures are 60 and 30 PSIG. PSIG is gauge pressure. The class 2 sheet also has a 7.0–11.0 GHz row with radii of 1.44/2.50 in and 50 PSIG. These overlapping bands must be matched to the applicable part designation, not selected solely because the operating frequency falls inside them. [2] [3]
A specification requirement does not increase a product rating. MIL-DTL-287J §3.1 says that its detail sheets govern a conflict with the general specification. It does not override a commercial catalogue or establish that an ordered product meets the sheet. A catalogue assembly listed at 30 PSIG cannot be treated as a 60 PSIG assembly without supporting product documentation. [1]
When extra-flexible twistable helps
Class 3 provides another option when the route needs both rotation and a small bend radius. At 8.2–12.4 GHz, MIL-DTL-287/5G specifies 1.38 in E-plane and 2.25 in H-plane radii, 240 deg/ft axial twist, 0.13 dB/ft insertion loss and 60 PSIG pressurisation. These are class 3 requirements, not an upgrade that can be assigned to an ordinary class 1 part. [4]
For comparison at that same band, class 1 specifies 3.38/7.51 in radii, 95 deg/ft axial twist and 0.10 dB/ft insertion loss. Class 3 permits a tighter route and more axial twist but has a higher loss limit. This is why “twistable” alone is not a complete purchase specification. [2] [4]
Does a seam increase loss
The catalogue comparison does not show a loss penalty for its twistable family. It also does not isolate the effect of a seam in a controlled test. Frequency, conductor finish, length, deformation and flange connections all matter when comparing complete assemblies.
The military sheets do not specify equal loss at every band. Class 1 and class 2 both specify 0.10 dB/ft at 8.2–12.4 GHz, but their values are 0.005 versus 0.010 dB/ft at 1.12–1.70 GHz and 0.35 versus 0.50 dB/ft at 18.0–26.5 GHz. Neither class has a universal loss advantage based on the seam label. [2] [3]
At X-band, 0.10 dB/ft converts to 0.328 dB/m using 1 ft = 0.3048 m. The catalogue’s 0.40 dB/m and the sheet’s 0.328 dB/m describe the same physical quantity under different specification contexts. The catalogue number is higher, so it cannot establish compliance with the sheet’s baseline loss limit.
For a procurement invoking MIL-DTL-287J, §3.5.2 limits insertion loss during the specified deformation test to 125% of the specified value. Section 4.7.3 requires the maximum observed loss to be recorded while the assembly is deformed. This is a limit relative to the specified baseline, not a claim that only a 25% increase from an arbitrary straight measurement is allowed, and not permission to add 25% to any catalogue value. [1]
A preliminary length calculation is loss in dB/m multiplied by flexible length. At 0.40 dB/m, 1.5 m of WR-90 gives 0.60 dB and 3 m gives 1.20 dB. These estimates exclude separately introduced joints and configuration-specific changes; request a loss-versus-frequency record for the ordered assembly.
Rigid guide is worth considering for portions that need no movement. University of Texas at Austin course notes give 0.11 dB/m for copper WR-90 at 10 GHz. Compared with the catalogue’s 0.40 dB/m flexible figure, the numerical ratio is about 3.6, but the materials and test conditions are not matched. Use that as context for considering rigid straight sections, not as a universal flexible-to-rigid loss ratio. [5]
Twist angle and service life
The WR-90 catalogue torsion figure of 308 deg/m converts to 93.8784 deg/ft. That is close to, but below, the class 1 axial-twist requirement of 95 deg/ft. Conversion and rounding do not prove compliance, particularly without matching the test conditions. [2]
Maximum axial twist is also not a repeated-motion rating. In the same class 1 band, Table I gives repeated-twist amplitudes of 47.5 deg/ft for 400 cycles and 14.25 deg/ft for 100,000 cycles, distinct from the 95 deg/ft axial-twist figure. State the movement amplitude and required life together. [2]
A known, fixed flange rotation may be handled by a rigid waveguide twist. Continuous rotation calls for a suitably rated rotary joint, rather than accumulating torsion in a flexible section. Confirm bandwidth, average and peak power, loss and service life for that separate component.
Power and pressure in mixed runs
A seamless section and a twistable section connected in series do not divide the RF power between them. Each must handle the local forward power, with losses and reflected-power conditions accounted for. Making the lower-rated section shorter reduces its estimated insertion loss; it does not automatically raise its power rating.
The same principle applies to a shared pressure volume: each exposed component must meet the operating pressure. A higher-rated seamless section does not protect a lower-rated twistable section. Check flanges, seals and pressure inlet hardware as well as the flexible core.
Average and peak power need separate checks. For an ideal rectangular pulse train, average power equals peak power multiplied by duty cycle; this arithmetic does not supply a missing peak-power or breakdown rating. The catalogue table alone is therefore insufficient to approve a pulsed system. Where the MIL-DTL-287J power test is specified, §4.7.13 requires the specified RF power for one hour after thermal equilibrium, under the specified frequency, altitude and temperature conditions. [1]
Three worked selection checks
These are calculations from the published tables, not installation reports or measured test results. They show how to reject an unsuitable option and what information is still needed. Length-based losses assume the listed dB/m value applies throughout the stated length and exclude additional interface losses.
A stationary X-band feed
Assume 2 kW average input, 8.2–12.4 GHz, 2 m of flexible guide, 20 PSIG and no axial rotation. The WR-90 catalogue values screen out the 1 kW twistable model and leave the 3 kW seamless model as a candidate. Estimated loss is 0.40 × 2 = 0.80 dB. Under a matched-line assumption, transmitted power is 2 × 10−0.80/10 ≈ 1.66 kW; about 0.34 kW is the corresponding input-to-output difference.
That estimate makes cooling and the exact assembly rating important. The listed 30 PSIG seamless pressure exceeds the assumed 20 PSIG demand, but the supplier must confirm the ordered configuration. If a class 2 assembly is required, specify that requirement separately, including its 1.38/3.00 in E/H radii. Do not replace the catalogue’s loss or pressure rating with military values.
A rotating feed above the twistable rating
Assume 100 W average reaches a WR-28 rotation section at 26.3–40.0 GHz. The catalogue twistable rating is 0.075 kW, or 75 W. It fails this power screen, even if the twistable section is short and the preceding seamless section is rated at 150 W. A different rotation component with adequate power and bandwidth, or a lower operating power, is needed.
For scale, reducing 100 W to 75 W through attenuation alone would require 10 log10(100/75) ≈ 1.25 dB before that section. That would deliberately discard 25 W and leave no power margin at the 75 W rating; it is not a recommendation to solve the mismatch with a lossy flexible run. Continuous rotation also needs a rotary-joint solution.
A pressurised lower-frequency run
Assume a 5 kW average link over 10 m of flexible guide at 1.13–1.70 GHz, with 10 PSIG pressure and no rotation. Both WR-650 catalogue power values, 20 kW seamless and 10 kW twistable, exceed the demand. Estimated loss is 0.12 × 10 = 1.20 dB. With matched terminations, output is about 5 × 10−1.20/10 = 3.79 kW, leaving approximately 1.21 kW as the input-to-output difference.
The class 1 sheet’s 2 PSIG requirement does not substantiate 10 PSIG service; the class 2 sheet specifies 15 PSIG for this band. A documented class 2 assembly is therefore a candidate, subject to its complete power, length and thermal specification. The example stops at 1.70 GHz deliberately: the catalogue extends to 1.73 GHz, but these sheet values do not establish performance from 1.70 to 1.73 GHz. Ten metres is a budget example, not evidence that a single flexible assembly of that length is available or thermally suitable.
Connections and acceptance records
A seam is a reason to examine contact construction, not proof that one assembly has worse passive intermodulation. ITU-T K.149 identifies dirty, damaged and corroded interfaces and improper connections as potential contributors. Its waveguide guidance calls for clean interfaces and the manufacturer’s flange torque. Both constructions still have flange joints, so a seamless core does not make the complete run immune. [6]
Specify an actual intermodulation test if it matters to the system; no comparative PIM measurements are supplied here. Match the flange interface, surface finish and gasket arrangement to the assembly drawing.
For a usable quotation and acceptance record, provide:
- Operating band, waveguide size, flange interfaces and relaxed assembly length.
- Average power, peak power, pulse duty cycle where applicable, temperature, altitude and cooling conditions.
- Minimum centreline radius in each plane, active flexible length and any fixed flange rotation.
- Motion amplitude, cycling rate and required service life for a moving run.
- Working pressure and the ratings of every component in the shared pressure volume.
- Loss and VSWR limits across the full band, including the installed deformation conditions.
- The exact specification revision, detail sheet and part designation if military conformity is required, plus the supporting inspection records.
MIL-DTL-287J §4.7.2 requires at least ten equally spaced points if insertion loss is tested at fixed frequencies. The class 1 and class 2 sheets also set VSWR limits by length at 12, 24, 36 and over 36 in; intermediate lengths take the next higher value. A short-sample VSWR value should not be assigned unchanged to a longer assembly. [1] [2] [3]
Which construction should you choose
Start with movement: use a documented twistable design when axial alignment or limited torsional motion is required. When no twist is needed, the seamless non-twistable family offers higher catalogue average-power ratings, and class 2 requirements allow tighter bending than class 1 in the compared bands. Then check loss, pressure and thermal conditions for the actual assembly. Neither a matching dB/m value nor a military class name replaces that check.
Specify the movement and the RF load together
Send your operating band, average and peak power, flange details, routing sketch and motion requirements. Include the working pressure and any required specification so the quotation can address the complete assembly.
Sources
Manufacturer catalogue values were checked on 26 October 2026. The military documents below are original specification texts accessed through EverySpec mirrors. Their stated revisions define the comparisons here; they are not a claim of current procurement status or certification of the linked products.
- MIL-DTL-287J, 10 April 2012. General requirements, classification, test methods and definitions.
- MIL-DTL-287/3F, 14 April 2009. Class 1 flexible twistable assemblies, Table I.
- MIL-DTL-287/4G, 14 April 2009. Class 2 flexible non-twistable assemblies, Table I.
- MIL-DTL-287/5G, 14 April 2009. Class 3 extra-flexible twistable assemblies, Table I.
- University of Texas at Austin waveguide course notes, 2022. Copper WR-90 attenuation at 10 GHz.
- ITU-T Recommendation K.149, December 2020. Passive intermodulation test methods and interface precautions.
