Hydraulic power steering: the pump, the rack, and the noise on full lock

A hydraulic power steering system spends its whole life telling you how it is doing, and the noise it makes is the most specific thing it says.

A whine that rises as you turn and fades as you straighten is not a general complaint about the front of the car. It places the fault in the steering, because nothing else in the car is loaded by the steering wheel — the pump is the only thing whose workload changes when you move it. And the check that follows costs nothing, takes a minute, and is the one most often skipped: look at the fluid, and look at whether it is full of froth.

Froth means the pump is drawing air, and a pump drawing air destroys itself. That is the single most important sentence in this article, and it is the one the piece it is translated from does not contain.

What the system is, and the reassuring part

Power steering assists; it does not steer. The steering wheel is connected to the road wheels mechanically at all times, through a column, a pinion and a rack. The hydraulics push on that rack to reduce the effort you have to supply.

The consequence is worth stating plainly because it is the thing people fear: if the assistance fails, you have not lost your steering. You have lost the help. The wheel becomes very heavy, most noticeably at parking speed, and the change can arrive abruptly — but the car remains steerable and can be driven gently somewhere safe. That is true whether the failure is a snapped belt, an empty reservoir or, on a modern car, a module withdrawing assistance deliberately.

The parts, which the original lists correctly:

• A pump, driven by a belt from the engine.
• A reservoir, holding the fluid and letting air separate out of it.
• A high-pressure hose to the steering, and a low-pressure return back to the reservoir.
• A control valve, usually built into the rack around the pinion, which decides which side of the ram gets the pressure.
• A double-acting ram — a piston on the rack itself, which can be pushed either way.
• A pressure sensor or switch in the high-pressure line, whose job is not what you would guess. It is covered further down.

How it works, and the number the original gets wrong

The original says the pump delivers fluid "at a pressure of 70 bar" to the valve. That figure is roughly right and the description around it is not, and the difference is worth getting straight because it explains the single most damaging thing a driver can do.

A power steering pump is a flow source, not a pressure source. It moves a more or less fixed volume of fluid per revolution. What the pressure does is whatever the rest of the system makes it do:

• Driving straight, the control valve is open. Fluid circulates from pump to valve and back to the reservoir against almost no resistance, so the pressure is low and the pump is doing very little work.
• Turning, the valve closes off one path and opens another, so flow is directed to one side of the ram. Now the fluid has somewhere to push, and pressure rises to whatever that push requires.
• At full lock, the rack hits its stop. The fluid has nowhere left to go, so pressure rises until the pump's relief valve opens and dumps flow back to the reservoir. That relief setting is where a figure like 70 bar comes from: it is the system's ceiling, not its normal running pressure.

This is why holding full lock is the classic way to kill a pump. With the wheel against the stop, the entire output of the pump is being forced across a relief valve, and every bit of that energy becomes heat in the fluid. Seals can be damaged in well under a minute. Turn to full lock if you need to, then back it off a fraction and hold it there — the geometry is nearly identical and the system is no longer relieving.

The original, which describes the pressure as a constant 70 bar delivered all the time, has no way to explain any of that. A model that says the system is always at maximum cannot tell you that pressing it against maximum is the harmful bit.

The noise, and what each kind means

Three different noises get reported as "the power steering is making a noise" and they have different causes. Separating them is most of the diagnosis.

The whine that follows the steering wheel

Check the fluid first, and check two things about it rather than one.

The level. These are sealed systems. There is no normal consumption, no evaporation and nothing that uses fluid up — so a low level always means a leak, and topping it up without finding the leak means meeting the same fault again later with a pump that has been wearing in the meantime.

Whether it is foaming. This is the half the original leaves out entirely, and it is the more informative of the two. A level low enough to let the pump draw air gives a system that whines, feels lumpy rather than smooth, is worse on lock and worse when cold, and — the giveaway — changes with cornering and gradient, because the fluid is moving around in the reservoir and uncovering the pickup at different moments. Foam in the reservoir is the confirming evidence, and it means the pump is being run on a compressible mixture it was never designed to pump.

Air also gets in after any repair that opened the system. If a car whines after a hose or pump replacement, it usually needs bleeding rather than another part.

The squeal, which is a different fault

The original attributes a squeal on turning to a defective crankshaft pulley first, and that is an odd place to start. A belt squeals because it is slipping, and the moment you turn the wheel is exactly when the pump's load spikes — so a belt that is glazed, perished, contaminated with oil, or simply loose will announce itself then and at no other time. Check the belt's condition and tension, and the tensioner if the car has an automatic one, before considering the pulley.

Note that on most engines the same belt drives the alternator and often the water pump, so a belt that eventually snaps takes considerably more with it than the steering. Engine noises covers telling belt noise from the noises that come from inside the engine.

Heavy steering, in the order worth checking

Steering that has become hard work is the other common complaint, and the original's sequence for it is good and worth keeping: belt, then fluid level, then fluid condition. Those three are free, and between them they explain most cases. What follows is what to do when all three are right, and here the original stops at "have a technician look at the pump, the valve or the ram", which is where the useful part begins rather than ends.

Compare one lock against the other. This is the single most useful test in the whole article and it separates the two expensive possibilities without removing anything:

• Heavy at one lock and better at the other, or assistance that fades as the fluid warms up, points at a restricted high-pressure hose. These degrade from the inside, where the liner can delaminate and act as a flap valve, so the hose can look perfect from the outside and still be strangling flow one way.
• Uniformly weak in both directions points at a worn pump.

And before condemning the pump, understand what wore it. A pump runs immersed in the fluid it pumps and depends on that fluid for cooling and lubrication as much as for pressure. Its wear is nearly always a consequence rather than a cause — a leak, a low level, or fluid that was never changed. Fit a new pump into a system that is still low and it wears out the same way, which is the same argument as replacing a catalytic converter without finding what killed it.

If the fluid is right and both locks are equally heavy and the pump has been proved, what is left is the control valve, the ram, or something that is not the steering system at all — seized suspension or steering joints, or tyres and geometry, which is where the original's third cause belongs.

The cause the original names that is genuinely mechanical

The original blames non-standard wheels and tyres, and alignment angles being out, for extra load on the system. That is real: a wider tyre or more aggressive geometry increases the effort needed to turn the wheels, particularly stationary, and the pump has to supply it. The point survives, with two clarifications.

It is not usually the cause of a power steering failure so much as a load the system was not designed for. And the fix is at the other end — wheel alignment and fitting the size the manufacturer specified, rather than anything hydraulic. Check tyre pressures before anything else, since soft tyres make the steering heavy at parking speed for free.

The part the original lists and never explains

Among the leak points the original names is the "electrical switch on the high-pressure line" — and it never says what the thing is for, which is a shame, because its purpose is not steering at all.

Turning the wheel at parking speed loads the pump heavily, and that load is enough to pull an idling engine down and stall it. The power steering pressure sensor exists to tell the engine module that a large load is arriving, so it can raise idle speed to meet it.

So the characteristic symptom of losing it is an engine problem, not a steering one: an idle that dips or an engine that stalls when you turn the wheel while parking, often worst with the air conditioning also running. A car that stalls while manoeuvring is very often this, and it gets looked for in the idle control system rather than in the steering. The codes are P0550 through P0554, and they are rated low severity precisely because the steering itself is unaffected.

The original is right that it leaks, though. It threads into the high-pressure line in a hot part of the engine bay, so leaks at its fitting are as common as electrical failure, and a weeping one is usually found by the fluid loss rather than by a code.

The fluid, where the original's advice does not travel

The original recommends a specific industrial hydraulic grade — ISO VG46 — for two national fleets' worth of cars, and a change interval of 40,000 km. Both need care, and the first is the one that could cost somebody a rack.

Power steering fluids are not interchangeable, and that is the rule that survives translation. There are three broad families: automatic transmission fluid, which many older Ford, Chrysler and Toyota systems genuinely specify; dedicated power steering fluid; and manufacturer-specific synthetics. They differ in their additive packages, and that is exactly what attacks the wrong seals — ATF carries detergents and friction modifiers that can swell or harden the rubber in a system designed for PSF, and it foams more readily, which brings back the aeration problem above. PSA's own specification for the platform most of the original's examples sit on is a named fluid rather than a generic grade.

So: use what the handbook names, and do not mix types. If you cannot find out what is in there, the answer is to flush and fill with the specified fluid rather than to top up with a guess.

On the interval, condition is a better test than mileage. Many handbooks specify no change at all; others give a figure. What matters more is what the fluid looks and smells like: dark and burnt-smelling fluid is a system that has been running hot, which is itself a finding — it points at a restriction, or at a pump that has been relieving more than it should. Check the level with the engine off and against the correct cold or hot mark, since the two differ by more than people expect.

The fire risk, which the original gets right

One thing in the original deserves more emphasis rather than less, and it is the only fault here that is not merely expensive.

Power steering fluid is flammable. Typical formulations have a flash point around 175–200 °C, and an exhaust manifold comfortably exceeds that. The original notes that on the Peugeot 206 the high-pressure hose is routed near the exhaust manifold, that the hoses perish with age, and that engine fires have been reported from exactly this.

That is a correct and serious observation, and the mechanism is worse than a drip suggests: a leak in a pressurised line atomises. A mist of fluid ignites far more readily than a puddle, and the high-pressure side is under real pressure whenever the engine is running.

Treat a power steering leak near the exhaust as urgent rather than as a job for the next service. Most fluid leaks on a car are a mess and an inconvenience. This one is the exception, and the reason is the routing rather than anything about the fluid.

Where it actually leaks

In rough order of how often it turns out to be each, and every one of them is visible as wetness once you know to look:

• Hoses and their crimped fittings, especially where a hose has been rubbing or has aged near heat.
• The reservoir and its low-pressure hoses — often just a clip.
• The pressure sensor's fitting, as above.
• The pump's shaft seal, which leaves fluid running down the front of the pump and along the belt.
• The rack's own seals. This is the one the original misses entirely, and it has a distinctive tell: the fluid is caught inside the rubber gaiter at the end of the rack rather than dripping on the ground, so a heavy gaiter with fluid inside it means a rack seal. Worth inspecting them anyway — a split gaiter lets in water and road salt, and that ends a rack faster than the hydraulics ever will.

The original's model-specific list is more useful than it first looks, and one entry is worth generalising. On the Peugeot 405, Pars and Samand it reports worn ram bushes producing a knock when the wheel is turned quickly — but only with the engine running, never with it off. That contrast is a real test rather than a description: with the engine off there is no hydraulic force being applied to the ram, so a worn mounting is not being driven against its slack. Any fault that appears only with assistance active is in the assisted part of the system, which is a way of thinking worth carrying beyond that one car.

If the car has electric power steering instead

Most cars sold now do, and a reader looking up "power steering" may well have one, so it is worth saying which of the above applies: almost none of it.

Electric power steering replaces the pump, belt, hoses, fluid and ram with a motor on the column or the rack. There is no fluid to check, no belt to slip and nothing to leak, which removes most of this article's failure modes at a stroke. It also takes no engine power when you are not steering, which is why it replaced the hydraulic system.

What replaces those failures is electrical, and it presents differently:

• Assistance is commanded rather than fixed, so it can be generous when parking and reduced at speed. P0635 to P0637 cover that control circuit, and a module that loses communication (U0131) will generally default to a safe fixed level or withdraw assistance entirely.
• Heavy steering is often the system protecting itself. Sustained full-lock manoeuvring heats the motor, and the controller reduces assistance until it cools. That is designed behaviour and it returns to normal. Low battery voltage does the same thing.
• A warning light with heavy steering usually means assistance has been withdrawn deliberately because a module stopped trusting an input — sometimes an input that has nothing to do with steering, since assistance level is set from road speed.

That last point is the one worth carrying: on any speed-sensitive system, hydraulic or electric, steering that feels wrong may be reporting a speed signal problem rather than a steering problem, and over-light at speed is as much a symptom as heavy at parking speed.

The short version

1. Listen to when the noise happens. Following the steering wheel means the steering; at idle regardless means the pump; only at the moment you turn means the belt.
2. Look at the fluid — level and froth. Froth means air, air means the pump is destroying itself, and a sealed system that is low has a leak somewhere.
3. Compare one lock against the other. Asymmetry is a hose; symmetric weakness is a pump.
4. Do not hold full lock, and back off a fraction when you have to use it.
5. Use the fluid the handbook names, and never mix types.
6. Treat a leak near the exhaust as urgent.
7. Losing assistance is not losing steering — the car is heavy, not uncontrollable.

Video guides

Video, for the parts of this that are easier watched than read.