Pump Loses Pressure Mid-Job

Why this matters

Full pressure at the start of a house wash and a steady, weaker stream twenty minutes in is one of the two failures that actually stop a job partway through, not just annoy you at startup. Before diagnosing anything, shut the trigger and let the rig sit a moment rather than working a weak stream harder by opening the nozzle wider or leaning on the trigger longer; forcing a struggling pump to keep making pressure it cannot sustain is how a marginal problem becomes a seized one. This case follows a real steady drop, not a pulsing or surging pressure, which is a different failure covered in its own article; the distinction matters because a steady drop and a fluctuating one point at different parts of the rig.

The symptom

Full working pressure at the start of the job. Over roughly twenty minutes of continuous use, the stream visibly weakens, settling at a lower but stable pressure rather than fluctuating. The gun still fires a solid, single stream, just noticeably weaker, with no surging, no sputtering, and no unusual engine sound. That last detail matters: the engine is not laboring or changing pitch, which rules out an engine-side power problem before you even open the pump.

First hypothesis: worn or wrong nozzle tip

The cheapest, fastest check, so it goes first. A worn nozzle orifice enlarges over time and drops pressure the same way a bigger hole in a garden hose nozzle does, and it is indistinguishable from a real pump problem by feel alone.

Swap to a known-good tip of the correct size and fire the gun. If pressure returns to full spec, the nozzle was the whole problem and the diagnosis stops here. In this case it does not: pressure with the fresh tip installed comes back only marginally higher, still well below the rig's rated output. Nozzle wear ruled out; it was contributing a small amount but is not the primary cause.

Second hypothesis: restricted water supply

A restricted inlet starves the pump of the volume it needs, and the pump responds by cavitating, pulling air into water it cannot fully fill the plunger chambers with. Check the spigot for full flow with the supply hose disconnected, check the supply hose itself for a kink or a crushed section, and pull the pump's inlet screen to check for debris loading.

All three come back clean here: the spigot delivers full flow, the supply hose runs straight with no kink, and the inlet screen is clear. This also matches the symptom description rather than contradicting it: a genuine inlet restriction usually announces itself as a rough, noisy, often fluctuating pump note from cavitation, not the smooth, steady, quieter drop described above. A steady drop with no unusual pump noise argues against a supply-side problem even before the physical checks confirm it. Water supply ruled out.

Third hypothesis: chemical injector check valve

This rig runs a sodium hypochlorite house-wash mix regularly through a downstream injector, and a gummed or corroded check valve on that injector's suction line has caused inlet restriction on other rigs in this fleet before, so it is worth ruling out explicitly rather than skipping it. Before opening the injector body, relieve system pressure and flush the injector line with clean water if a source is available, since residual mix in the line is still full-strength chemical. If a clean-water flush is not available on this site, treat the line as carrying full-strength mix and wear eye protection and gloves before cracking the fitting rather than assuming a short exposure is fine.

On inspection, the check valve seats cleanly with no gumming or corrosion, and the metering valve is confirmed fully closed, matching how it should look with the injector not in use on this job. Chemical injector ruled out.

Fourth hypothesis: unloader valve bleeding pressure internally

A worn unloader seat can bleed part of the pump's output to bypass even while the trigger is held open, which would produce exactly this kind of steady, stable pressure loss rather than a pulsing one. The unloader service and cycle-test procedure is covered in full in its own article rather than repeated here; running that test on this rig shows a clean, prompt snap to bypass on trigger release and a full return to rated pressure on re-pull, with no creep during an extended hold. That test result rules the unloader out as the cause of this particular symptom, even though a worn unloader remains a real possibility worth checking on the next rig that shows this pattern.

What was actually happening: a slipping drive belt

With the first four hypotheses ruled out, the belt connecting the engine to the pump is the remaining link between an engine running at normal, unlabored RPM and a pump producing less than its rated output. With the engine off and the belt guard removed, a hand-deflection check on the belt is run against the manufacturer's spec: press at the belt's midpoint with a consistent thumb load and measure how far it gives. The spec for this belt and pulley span calls for roughly a third of an inch of deflection under that push. This belt gives nearly twice that before it firms up, and its underside has a glazed, shiny appearance rather than the matte texture of a belt gripping properly. Glazing is the physical signature of a belt that has been slipping under load: friction heat polishes the contact surface, which then grips even less, a cycle that gets worse the longer it runs, and a belt already loose enough to deflect at roughly double its spec has been slipping for a while before this job, not starting fresh today.

A slipping belt lets the engine spin at its normal, unlabored RPM (explaining why the engine sounded fine) while the pump shaft turns slower than it should, which reduces pump output volume proportionally and produces a smooth, steady pressure drop rather than the rough, fluctuating symptom a cavitation or valve problem would cause. That is also why it took about twenty minutes to become noticeable: the belt was gripping well enough at the cold start of the job, and only lost enough grip to matter once friction heat built up and the glazing worsened under sustained load.

The fix and why the order mattered

The belt is retensioned to the pump manufacturer's spec, or replaced outright given the visible glazing, since a glazed belt tends to keep slipping even after retensioning because the surface itself has been polished smooth. After replacement, the rig is run back up to working pressure and holds full rated output through a five-minute continuous test, confirming the fix.

The reason to work through the cheaper checks first, rather than jumping straight to the belt: a tech who assumed a pressure problem on a belt-drive rig must be the pump and pulled the wet end for inspection would have found nothing wrong internally, spent real shop time on an unnecessary teardown, reassembled it, and still had the same symptom waiting on the next job, because the pump itself was never the fault. The order here, cheapest and fastest first, is also roughly the order of how commonly each cause actually occurs on rigs in this duty class, which is not a coincidence.

Log the belt's condition and the deflection reading in the rig's service record alongside the pump's own oil and hour tracking, not as a separate note. A belt failure that shows up as a repeat pattern on the same rig every few months is a scheduling problem, not a bad-luck problem, and it only becomes visible as a pattern if someone is writing the readings down each time rather than fixing the belt and moving on.

What changes the answer: duty cycle and season

A rig in daily commercial rotation reaches belt wear at this stage far sooner than a weekend-only residential rig, simply from accumulated run hours under load, and a belt already near the end of its service life is far more likely to be the answer on the daily rig than on one that has logged a fraction of the hours. Heat and humidity change the timeline too: a belt running in full midday sun in hot weather glazes faster under the same slip than the same belt would on a cool, overcast day, which is part of why this exact symptom, fine at the start, weaker after twenty minutes, shows up more often on the hottest jobs of the season rather than spread evenly across the calendar.

What would point somewhere else

A fluctuating or pulsing pressure instead of a steady drop points toward the surging-and-pulsing article, not this one. A pump that is noisy or vibrating unusually alongside the pressure loss points back toward the water-supply hypothesis even if the spigot and hose checked out, since a restricted return path or a partially closed valve somewhere in the supply line can still starve the pump intermittently in a way a single static check might miss. And a pressure loss that shows up immediately at startup rather than gradually over the job points toward the unloader or nozzle hypotheses first, since a belt that is going to slip under sustained heat has not yet had time to heat up and glaze at the very start of a cold run.

References

• Pump and engine manufacturer belt tension specifications (model-specific deflection and torque values)
• See related: Unloader Valve and Pressure Regulation Service
• See related: Rig Surges or Pulses Instead of Running Steady
• See related: Belt-Drive Pump Service and Oil Change