Choose flexible waveguide when a connection must accommodate limited movement, flange misalignment or a route that rigid sections cannot fit conveniently. Choose rigid waveguide for fixed runs where low loss and stable geometry matter most. A practical compromise is a rigid main run with a short flexible section at the moving or misaligned interface.
The flexible section still has to meet the frequency, power, loss, bend-radius and motion requirements of the installation. MIL-DTL-287J describes flexible assemblies as a way to facilitate complex installations, accommodate misalignment and permit relative motion between connected components; it does not make every flexible assembly suitable for every moving connection. [1]
Table of Contents
Choose by the installation
| Installation need | Starting choice | What to confirm |
|---|---|---|
| Small offset between fixed flanges | Short flexible section | Installed bend radius in both planes, flange alignment and loss after forming |
| Repeated movement between components | Flexible assembly rated for that motion | Travel, bend angle, cycle life and electrical performance throughout the movement |
| Continuous rotation | Rotary joint | Rotation duty and RF ratings; a twistable jumper has finite travel |
| Long, fixed route | Rigid sections and designed bends | Total installed loss, supports, flange count and installation access |
| Phase-sensitive measurement path | Mechanically stable rigid route where practical | Phase change with temperature and movement, separately from VSWR |
| High-power or pressurized system | Compare qualified assemblies of either construction | Peak and average power, pressure, temperature, altitude and mismatch conditions |
When flexibility solves the problem
Flanges that do not line up
A flexible jumper can bridge a small offset without forcing two rigid components into alignment. Draw both flange faces, their relative rotation and the path between them. The assembly must fit that path without exceeding its E-plane or H-plane bend limit or imposing an unapproved twist.
Length tolerance is not an allowance for misalignment. For example, MIL-DTL-287J gives a relaxed-length tolerance of ±0.090 in for assemblies up to 36 in long. That describes the supplied length, not how far the flanges may be displaced during installation. Bend and twist requirements are specified separately. [1]
The WR size alone does not identify the flange. Check the full designation, bolt pattern, alignment features and sealing face against the mating drawing. Dolph Microwave’s EIA and IEC waveguide flange reference helps identify those interfaces.
Limited movement and repeated flexing
Separate installation bending from movement in service. Under MIL-DTL-287J, the bending test uses five complete cycles in each specified plane. The flexure test runs at a minimum of 30 cycles per minute, but its bend planes, angle and total cycle count must be specified. That test rate is not a product’s continuous-duty rating or a promise of service life. [1]
For a moving antenna feed, state the two endpoint positions, the motion rate and the required lifetime cycles. Ask for loss and VSWR acceptance limits in the installed configurations. For continuous rotation, use a suitably rated waveguide rotary joint; a flexible section may accommodate residual alignment, but it cannot accumulate unlimited twist.
Routes with little installation space
A flexible section may replace several rigid bends, but it still needs room for its minimum bend radius and supports. Compare the complete routes: straight sections, bends, flange joints and the space needed to assemble them. A shorter-looking route is not necessarily the lower-loss option.
For long field-routed feeds, elliptical waveguide and transitions are another option to evaluate. Their routing and transition requirements differ from those of a short rectangular flexible jumper.
When rigid waveguide is better
Favor rigid waveguide when the equipment positions are fixed and the route can be built without forcing the interfaces. It avoids repeated deformation of the RF path and is usually the first option to compare when the attenuation budget is tight.
MIL-HDBK-216 gives a historical comparison for one construction: interlocked flexible waveguide has about twice the attenuation of rigid silver tubing. Its interlocked-type discussion also identifies roughly 6 in to 4 ft as a useful length range. Neither statement is a universal limit or rating for modern flexible products. [2]
For phase-sensitive equipment, specify allowable phase change across temperature and motion. A low VSWR describes the impedance match; it does not demonstrate phase stability. If a flexible section is unavoidable, include it in the phase acceptance test rather than assuming a good match is sufficient.
Choose the flexible construction
MIL-DTL-287J distinguishes Class 1 twistable, Class 2 non-twistable and Class 3 extra-flexible twistable assemblies. Its axial-twist test applies only to twistable types, while repeated twist has a separate test with a specified cycle count. A five-cycle axial-twist result alone does not establish a repeated-motion lifetime. [1]
| Construction | Mechanical distinction | Selection consequence |
|---|---|---|
| Interlocked | Adjacent convolutions slide as the guide deforms | Confirm bend, twist and pressure limits for the complete assembly |
| Soldered convolute | More flexible than interlocked, but more fragile and not twistable | Do not use axial twist to correct flange rotation |
| Unsoldered convolute | Can twist, but cannot bend as sharply as the soldered construction | Check the bend radius as well as torsion |
| Seamless corrugated | The type described can bend and compress, but cannot withstand twisting | Resolve flange rotation elsewhere unless the specific product has a documented torsion rating |
These distinctions come from the handbook’s construction descriptions, not from tests of the catalog products below. Select a current assembly from its own drawing and ratings. [2]
What the published catalog shows
The following five selected rows come from Dolph Microwave’s twistable flexible waveguide catalog, checked on October 25, 2026. They are published model values, not measurements made for this article or proof of MIL-DTL-287J qualification.
| Size | Listed band (GHz) | Loss (dB/m) | Average power (kW) | Maximum torsion (deg/m) |
|---|---|---|---|---|
| WR650 | 1.13–1.73 | 0.12 | 10 | 5 |
| WR284 | 2.60–3.95 | 0.15 | 6 | 46 |
| WR137 | 5.38–8.17 | 0.25 | 2 | 205 |
| WR62 | 11.90–18.00 | 0.8 | 0.4 | 436 |
| WR22 | 32.90–50.10 | 2.7 | 0.025 | 510 |
Use these rows to screen candidates, then obtain limits for the ordered length, operating frequency and formed geometry. The listed bands are product-page values, not a universal operating-band definition for each WR size. MIL-DTL-287J ties a compliant flexible assembly’s operating range to its mating rigid waveguide; confirm that the complete assembly covers the required band. [1]
The separate seamless flexible waveguide catalog lists brass or beryllium-copper cores, silicone jackets, lengths from 300 to 1500 mm and a maximum pressure of 30 PSIG. Its WR650 row lists 20 kW average power, compared with 10 kW in the twistable table. Both list 0.12 dB/m. The comparison shows why loss alone cannot decide between constructions; it does not establish equal power capability under unspecified operating conditions.
A fixed offset with a loss budget
Consider a hypothetical 1 m WR284 jumper at a frequency within its listed band. Using the catalog value of 0.15 dB/m gives an estimated section loss of 0.15 dB. The transmitted-power fraction is 10(−0.15/10) = 0.966, or about 3.4% power loss through that section.
If the project allocates 0.20 dB to this connection, the estimate leaves 0.05 dB of margin. These are screening calculations: they assume the catalog rate scales with length and exclude additional installed-interface effects. Ask whether the quoted assembly loss includes its flanges and formed bends before assigning the remaining margin. No rigid loss can be inferred from this catalog row by dividing it by two.
A short jumper with flange rotation
For a hypothetical 300 mm WR284 section, multiplying 46 deg/m by 0.300 m gives 13.8 degrees. A 10-degree offset is below that simple length-scaled figure; a 20-degree offset is above it. This is an initial torsion screen, not installation approval: the supplier must confirm the usable flexible length, end restrictions, combined bend-and-twist limits and motion duty.
The same calculation for WR650 gives only 1.5 degrees over 300 mm. Shortening a jumper reduces its length-related loss, but may leave too little length to accommodate the required rotation. That trade-off is why the routing drawing must accompany the RF requirements.
A higher-frequency loss constraint
At the WR22 catalog rate of 2.7 dB/m, a hypothetical 300 mm section contributes an estimated 0.81 dB, leaving about 83.0% of the incident power. It already exceeds a hypothetical 0.50 dB allocation before additional installation effects. The loss-only length ceiling is 0.50 ÷ 2.7 = 0.185 m, about 185 mm; that calculation does not establish an available or mechanically workable product length.
In that case, compare a different route, a rigid bend or a hybrid assembly. Do not shorten the flexible section below its mechanical requirements simply to make the arithmetic pass. Dolph Microwave’s custom waveguide assemblies include flexible and rigid combinations for such routing discussions.
Check power and pressure separately
Power handling can change the selection. Specify peak power, average power, pulse conditions where relevant, frequency, maximum VSWR, ambient temperature and altitude. A catalog average-power value cannot substitute for a peak-power or breakdown rating.
MIL-DTL-287J’s power test, when specified, uses the required RF power, frequency, simulated altitude and temperature. It maintains power for one hour after thermal equilibrium, defined as no more than 5°C temperature change within 15 minutes. Those conditions illustrate what a meaningful power-rating comparison needs. [1]
Pressurization is not exclusive to rigid waveguide. The flexible specification includes an internal-pressure leak test and requires insertion-loss stability and VSWR measurements under pressure. The required test pressure is specified for the assembly; immersion depth is not its pressure rating. [1]
The handbook discusses pressurization for high-power breakdown control and heavier walls to limit pressure-induced deformation. That is design guidance, not permission to apply a generic pressure to an unqualified part. [2] Where a sealed RF boundary is needed, include the waveguide pressure window, flanges and seals in the assembly-level review.
Specify the assembly before ordering
| Requirement | Put on the drawing or specification |
|---|---|
| RF band and interfaces | Operating frequencies, WR size, full flange designations and mating drawings |
| Installed route | Flange positions, relative rotation, length convention, E-plane and H-plane bend radii, clearance and supports |
| Motion | Installation-only or repeated movement, endpoint positions, cycle rate and lifetime cycles |
| Electrical acceptance | Maximum installed insertion loss and VSWR across the band; phase-change limits where needed |
| Power and pressure | Peak and average power, pulse conditions, temperature, altitude, operating pressure and leak acceptance |
| Environment and evidence | Vibration, shock, moisture and temperature exposure; applicable specification revision, test conditions and reports |
MIL-DTL-3922E covers general-purpose flanges that couple waveguide sections mechanically and electrically. Naming the applicable flange specification supports interface control, but the mating drawing still has to match. [3]
Compare like-for-like acceptance limits. MIL-DTL-85H’s VSWR limit of 1.02 applies, unless otherwise specified, to fabricated waveguides tested under its stated procedure. It is not a blanket guarantee for every rigid installation or its bends and flange joints. [4]
For flexible assemblies, MIL-DTL-287J sets insertion loss and VSWR against specified values and separately tests insertion-loss stability under deformation. Even an installation-only section needs frequency, power, pressure where applicable, flange and installed-loss checks; bend radius alone is insufficient. [1]
Frequently asked questions
Is flexible waveguide always more lossy?
It commonly has higher attenuation than a comparable rigid section, but there is no universal multiplier for all constructions. Compare supplier data at the same frequency, length and installation conditions. Include the bends and joints that each complete route requires.
How long can a flexible section be?
Use both the mechanical drawing and the loss allocation. For preliminary screening, divide the section’s available loss budget in dB by its applicable attenuation in dB/m. Then check available lengths, bend radius, torsion and support requirements. The loss-only result is not a maximum service length.
Can flexible waveguide replace the entire rigid run?
Only if the replacement meets the complete RF and mechanical specification. A long fixed run usually offers little reason to accept additional flexibility. Localize flexibility where it solves an offset or movement requirement, and compare the installed loss of the hybrid route with the alternatives.
Does twistable mean suitable for continuous rotation?
No. A twistable assembly permits a specified angular displacement and, where documented, a specified number of motion cycles. Continuous rotation requires a rotary interface designed for that duty.
Make the final choice
Choose the shortest flexible section that comfortably meets the route’s bend, twist and movement requirements, then confirm its installed RF performance. Keep fixed portions rigid where that reduces loss and simplifies support. Approve the complete assembly from its drawing and operating conditions, rather than from the word “flexible” or a single catalog number.
Still not sure which construction fits the route? Send the operating frequency, the waveguide size, the power level, the flange type and a drawing of the installed path, and we will tell you whether a flexible section belongs in the design or rigid waveguide does the whole job, with the loss and interface consequences spelled out.
References
- MIL-DTL-287J, Waveguide Assemblies, Flexible, Twistable and Non-Twistable, US Department of Defense, April 10, 2012. Clauses 1.2, 3.4.2–3.4.4, 3.5.1–3.5.4, 4.7.3, 4.7.5, 4.7.13–4.7.17 and 6.1.
- MIL-HDBK-216, RF Transmission Lines and Fittings, US military handbook, Section 6 dated June 12, 1975; compilation with Notices 1–10, March 11, 1982. Sections 6.2.7 and 6.4.2.1–6.4.2.1.4. Historical construction guidance.
- MIL-DTL-3922E, Flanges, Waveguide, General Purpose, US Department of Defense, April 18, 2016. Clause 1.1.
- MIL-DTL-85H, Waveguides, Rigid, Rectangular, US Department of Defense, June 22, 2000; validation notice April 15, 2005. Clauses 3.9 and 4.5.7.
