The alarm you silenced in March was doing its job. It told you that the pressure in the WR-75 run had fallen below the dehydrator’s low-pressure threshold. What it did not tell you was where the gas went, how fast it was leaving, or whether the loss was a leak at all. A threshold alarm is a boundary condition, not a diagnosis. Silencing it converted a known, bounded pressure condition into an unknown one, and the run has been quietly consuming makeup gas ever since.
This is a procedure for turning that situation back into a measurement. The goal is not to find a gasket by intuition. The goal is to isolate the WR-75 run, stabilize it thermally, log pressure against time, and use the shape of the decay curve — cross-checked against the dehydrator’s own flowmeter — to localize the boundary that is passing gas. A gasket that breathes will show itself as a rate that changes with temperature or bolt load, not as a single number.
Start with the standard, not the soap bottle
IEC 60261, Sealing test for pressurized waveguide tubing and assemblies, specifies uniform measuring methods for sealing tests on pressurized waveguide components and assemblies, carried out both quantitatively and qualitatively. It defines a pressure-drop-during-elapsed-time method as Method A, a quantity test, and separates that from leak-rate measurement (Method B), bubble testing (Method C), halogen leakage (Method D), and helium leakage (Method E). The standard’s scope is explicit: these are measuring methods, and the choice of method belongs in the relevant specification before anyone climbs the tower.
That matters here because the methods answer different questions. Method A gives you a pressure change over a defined interval for a defined volume. Method C locates a leak at an accessible joint but, as ASTM E515 treats bubble emission, it is a detection and location technique, not a quantitative measurement of total system leakage. If your acceptance record says “no bubbles seen,” you have a qualitative result and no leak rate. If it says “pressure fell 0.4 psi in 30 minutes,” you have a number you can compare to next year’s number — provided you recorded the boundary, the gas, the pressure basis, and the temperature.
Instrument the run before you touch a flange
Before any isolation, establish what the installed equipment is already telling you. A dehydrator of the Gibson X/Z F4 class, for example, supplies dry air at a regulated pressure and carries a flowmeter that the manufacturer states allows leaks in the waveguides to be measured. The same manual documents a low-pressure alarm that triggers below 7.0 kPa and restores above 8.0 kPa in its standard configuration. Those are the factory figures for that model; your unit may be configured differently, and the setpoint on the drawing governs, not the catalog page.
Record, at minimum:
- The dehydrator model, regulated output pressure, and alarm thresholds as configured on site.
- The pressure basis of every gauge you will read — gauge pressure or absolute. Write it down. A 3.3% change in absolute pressure is not a 3.3% change on a gauge that reads zero at ambient.
- The flowmeter range and its current indicated flow with the run connected and stable.
- The internal volume you believe you are testing, and how you estimated it.
- A calibrated gauge at the far end of the run if one can be fitted, so the near-end and far-end readings can be compared.
The dehydrator’s catalog pressure rating is not automatically the operating setpoint for the line. One waveguide pressure-inlet vendor states this directly and warns against inferring a proof pressure by multiplying a catalog value by an unsupported safety factor. Use the approved drawing and the released acceptance specification.
Temperature stabilization is the first-order error
For a sealed rigid volume with approximately ideal gas behavior, absolute pressure divided by temperature remains constant when the quantity of gas is unchanged, with temperature in kelvin. A calculated cooling example from 303 K to 293 K reduces absolute pressure by about 3.3% without any gas loss at all. That is a 10 K overnight swing producing a pressure change that looks exactly like a slow leak on a gauge read at breakfast.
This is the most common misread in field pressure-decay testing: attributing thermal contraction to a gasket. The correction is cheap. Log ambient temperature and waveguide skin temperature alongside pressure for the full stabilization window, and do not start the timed decay until the rate of change of temperature is small compared with the rate of change of pressure you are trying to resolve. If the run is outdoors or in an unheated shelter, a night-to-day swing will dominate a small leak. Test in the most thermally stable window you have, and state the window in the record.
Pressure-decay testing, as described in the waveguide pressure-inlet guidance, means isolating the volume and measuring pressure over a defined period after temperature stabilization, with gauge resolution, test volume, and gas temperature included in the record. Skip the stabilization and you are measuring the weather.
Isolation strategy for a WR-75 run
Segment at accessible flanges and pressure windows, and test the smallest credible boundary first. The logic is simple: a small isolated volume reaches thermal equilibrium faster, and a leak in it produces a larger fractional pressure change per unit time than the same leak in the full run. Start with the section you suspect, then widen.
Keep the dehydrator’s regulated supply out of the isolated volume. You are testing the boundary, not the regulator. Close the isolation valve at the dehydrator outlet, or disconnect and cap per the approved procedure, and confirm the isolation itself does not leak before you trust any decay number. A leaking isolation valve produces a decay curve indistinguishable from a leaking flange, and you will spend the afternoon torquing the wrong joint.
Record the test boundary precisely — which flanges, which switches, which pressure windows are inside the isolated volume. A decay result without a boundary is not a measurement; it is an anecdote with units.
Reading the decay curve
Plot pressure against time after stabilization. Two features matter.
The initial settle. A fast drop in the first minutes that then flattens is usually the isolated volume equalizing with its fittings, a valve seat settling, or a small quantity of gas redistributing. It is not the leak signature you are chasing. Note it, exclude it from the rate calculation, and state the exclusion.
The steady slope. A constant linear decay over tens of minutes is a leak at a roughly constant rate. A gasket that breathes is different: its rate changes with temperature cycling or with bolt load. That is the signature — the change in rate, not the absolute rate. A single decay number cannot distinguish a breathing gasket from a pinhole in a pressure window. Two tests at two stabilized temperatures, or at two bolt-load conditions, can.
This is where the procedure earns its keep. Run the isolated segment at the ambient condition of the morning, then again after the run has warmed through the afternoon, or after a controlled re-torque. If the rate moves with temperature in a way that the ideal-gas correction does not explain, you have a boundary whose sealing depends on compression — a gasket, an O-ring, or a flange face that is not flat. If the rate is constant across both conditions, look elsewhere: a cracked window, a valve packing, a fitting thread.
Cross-check with the dehydrator’s flowmeter
The isolated decay test and the dehydrator’s makeup flow answer different questions. The decay test measures the leak rate of the boundary you isolated. The flowmeter measures the makeup flow the dehydrator is supplying to hold the connected system at pressure. A source that maintains pressure by replacing a large leak can hide the defect, which is exactly why consumption should be compared with an established baseline rather than read in isolation.
The diagnostic value is in the disagreement. If the isolated segment holds pressure within the acceptance limit but the flowmeter shows steady consumption when the run is reconnected, the fault is in the boundary you excluded from the isolated test — the isolation valve, the section beyond it, or the dehydrator plumbing itself. If the isolated segment decays and the flowmeter agrees, you have localized the leak to the tested boundary and can proceed to location. Either way, you have narrowed the search without opening a single flange on a hunch.
Confirm, repair, and re-test
Once the decay curve and the flowmeter point to a boundary, locate the leak at the suspect joint under the approved test pressure using a bubble test or an approved detection liquid. Bubble testing locates; it does not quantify. Record the conditions and the observation period, because “no visible bubbles” is only meaningful with both stated.
If the suspect is a flange joint, re-torque in a cross or star pattern — 50% of final torque on the first pass, 100% on the second — so the O-ring or gasket compresses evenly. The torque value must come from the flange or equipment manufacturer’s data for the specific flange and bolt size. One vendor application note gives 8 to 15 in-lbs for WR-112 to WR-62 with #6-32 or #8-32 bolts, but that table does not list WR-75, and a number borrowed from an adjacent size is not a specification. Get the WR-75 figure from the flange drawing or the manufacturer before the torque wrench comes out.
After locating and repairing a leak, repeat the test and remove any detection-liquid residue. The re-test is not a formality: it is the only evidence that the repair changed the rate, and it produces the number that goes into the acceptance record.
Close the loop so next March is comparable
The reason the March alarm restarted the investigation from zero is that the previous acceptance record probably did not carry the fields needed to compare like with like. The record should include the test boundary, the gas, the pressure basis, the stabilization time, the gauge resolution, the test volume, the ambient and skin temperatures, the acceptance limit and its source, and the measured decay rate with its uncertainty. State the uncertainty honestly — a gauge resolution of plus or minus 0.5 psi over a 30-minute test bounds what you can claim about a 0.2 psi change.
With that record in hand, next year’s alarm is a comparison, not an investigation. You will know whether the run is consuming more makeup gas than its baseline, whether the decay rate has moved, and whether the boundary that changed is the one you repaired. And you will have a defensible answer to the only question that matters when the alarm sounds again: is this the same gasket breathing, or is it something new?
Frequently asked questions
Can I just raise the dehydrator regulator and leave the alarm silenced? No. Raising the setpoint changes the pressure at which the alarm trips; it does not change the leak rate. The run continues to consume makeup gas, and the defect remains. The alarm threshold is a boundary condition, not a diagnosis, and suppressing it removes the only automatic indication that the boundary has been crossed.
How long should the decay test run? Long enough that the steady slope is distinguishable from the initial settle and from residual thermal drift, and short enough that the temperature stays within the window you recorded. There is no universal number; the test duration belongs in the acceptance specification along with the limit. What matters is that the duration, the stabilization criterion, and the temperature window are recorded so the next test is comparable.
Is a bubble test enough on its own? It is enough to locate a leak at an accessible joint. It is not a quantitative measurement of total system leakage, and it will not find a boundary that only passes gas under a particular temperature or bolt-load condition. Use it to confirm and locate after the decay test has pointed you at a boundary.
Do I need helium mass spectrometry for a WR-75 ground-station run? Not by default. Helium testing is reserved for hermetic or space-qualified hardware with very tight leak specifications. For a pressurized WR-75 feed run, the pressure-decay method plus a flowmeter cross-check is the proportionate tool. If the acceptance specification calls for a tighter limit than pressure decay can resolve, that requirement should be stated in the specification, not inferred in the field.
What acceptance limit should I use? The one in your operator’s specification or the applicable standard revision. A vendor application note’s figure is a vendor’s figure, not a standard, and should not be generalized to a run it was not written for. If no limit exists, that is the finding to escalate — not a number to invent.