Aircon EEV: why inverter systems need active metering
Something has to stand between the high-pressure side and the indoor coil, deciding how much refrigerant gets through. Older equipment settled that at the factory and never revisited it. An inverter cannot work that way, and the part that replaced it fails looking like low gas.
By Team Snowflake | Updated 5 Aug 2026
What the metering stage is there to do
Every cooling circuit has one point where high-pressure liquid is let down to low pressure, and that point is the metering stage. Refrigerant leaves the outdoor coil warm and under pressure. It has to reach the indoor coil cold enough to boil against room air. Nothing else in the loop makes that change happen.
Pressure sets the temperature at which refrigerant boils, so the drop is the whole trick. Hold a liquid under pressure and it boils hot. Release the pressure and it boils cold. Refrigerant pushed through a narrow gap arrives on the far side at a much lower pressure. On that side it evaporates far colder than the room, and that gap is what pulls heat out of passing air.
The same gap also settles quantity, and that half gets far less attention. Too little through, and the coil runs out of liquid before the air has finished crossing it. The remaining fin area does nothing. Too much through, and liquid is still present at the outlet, travelling back toward a compressor built to handle gas.
Two jobs therefore sit on one restriction. Pressure drop and flow rate are set by the same gap, so neither can be adjusted without moving the other. Every metering design is an answer to a single question, which is how wide that gap ought to be at this moment.
Where the metering stage sits
On a household split system the metering usually sits at the indoor unit, on the thin liquid line arriving from outside. Some designs place it in the outdoor unit instead. Either way it marks the border between the high side of the circuit and the low side.
Systems with more than one indoor unit meter each branch on its own. A single outdoor unit feeds several rooms, and each room needs its own answer, because each carries its own heat load and its own demand. That detail matters later, since a metering fault on one branch leaves every other branch untouched.
Three ways to set the gap, in order of control
The oldest answer is a fixed length of very narrow tubing with nothing moving inside it. Bore and length are chosen for one model at one design condition, and that choice is made at the factory. Nothing about it changes afterwards. The restriction offered on a mild night is the restriction offered at the peak of a hot afternoon.
A thermostatic valve adds feedback without adding electronics. A sensing bulb clamped to the pipe leaving the coil is filled with a small charge of its own. Pipe temperature raises or lowers the pressure inside that bulb, and the bulb pushes on a diaphragm linked to the valve. Warm gas leaving the coil opens the valve wider. Cold gas closes it down. The correction happens with no wiring and no instruction from anywhere.
The electronic version replaces that mechanical link with an instruction from the control board. A small stepper motor turns a threaded shaft in counted increments, and each increment sets the size of the opening. Thermistors on the pipework report what the refrigerant is doing. The board weighs those readings against a target and sends a new step count.
The three differ in authority rather than in accuracy. A fixed tube holds no information and can act on none. A thermostatic valve senses one condition and answers it mechanically. An electronic valve is directed by a board that can weigh coil readings, compressor speed and room demand together, and that can hold a chosen position deliberately instead of settling into one.
| How the gap is set | What it can respond to | Where it runs out of answers |
|---|---|---|
| How the gap is setA fixed bore sized once at the factory | What it can respond toNothing. The restriction it offers never changes | Where it runs out of answersCorrect at the condition it was sized for, drifting wrong on either side of it |
| How the gap is setA mechanical valve driven by a sensing bulb | What it can respond toTemperature at the coil outlet, and nothing beyond that | Where it runs out of answersAnswers after the fact, and only inside a narrow band of movement |
| How the gap is setA stepper-driven opening commanded by the board | What it can respond toCoil readings, compressor speed and room demand at once | Where it runs out of answersDepends on its sensors and its firmware being right about the circuit |
Why fixed metering lasted as long as it did
A capillary tube is close to unkillable. No motor, no wiring, no sensor and no firmware, so the list of things that can fail on it is very short. It costs almost nothing to fit. On a single-speed compressor it also had little to answer for, because that compressor only ever ran at one output.
Its weakness only turns expensive once the rest of the system starts moving. Fixed metering with fixed output is a matched pair. Fixed metering with variable output is a mismatch, and the mismatch shows up as capacity quietly not delivered rather than as a breakdown.
A compressor that changes speed changes the sum
An inverter compressor has a range of outputs rather than one. Speed climbs while a flat is still hot and drops back once the room has settled. Each speed sends a different amount of refrigerant round the loop. The metering stage has to size its gap for whatever amount is arriving at that moment.
A fixed restriction can only be correct once. Size the bore for full output and it throttles the circuit at low speed. Size it for low speed and it floods the coil as soon as the compressor winds up. Wherever the sizing lands, it is right at a single operating point and steadily wrong on either side of it.
The two technologies arrived together for exactly that reason. Variable output is worth little if the metering cannot follow it. The saving an inverter is sold on comes from long stretches at part load, and part load is where a fixed bore sits furthest from correct.
So the valve is not an accessory bolted onto an inverter system. It is what makes the modulation usable at all. Compressor drive, control board and valve form one control loop with three parts, and the loop was designed as a whole. Losing sight of the third part is why a metering fault so often gets attributed to the compressor or to the charge.
- Compressor speed, which the board changes continuously as the room approaches its setpoint
- Heat load in the room, which rises with afternoon sun on the wall and with people in it
- Conditions around the outdoor unit, which decide how much heat the circuit can shed
- Indoor fan speed, which sets how much warm air actually crosses the coil
- On a multi-split, how many indoor units happen to be calling at the same time
What active metering buys across the load range
The figure the board is holding steady is superheat, and superheat measures how completely the coil was used. Liquid enters the indoor coil, boils along its length, and should finish boiling a little before the outlet. How far the gas warms past that finishing point is the reading. What the figure rules in and out on a service report is set out elsewhere.
Holding that figure steady across the load range is precisely what a fixed bore cannot do. Compressor speed, room temperature and outdoor conditions all shift where boiling finishes inside the coil. A valve that can reopen or close down keeps the finishing point roughly where it belongs whatever else moves. The coil then stays fully used at low output and at high.
The penalty at each end of the range is worth naming plainly. Starve the coil and its last stretch carries gas that has already boiled, warming up and doing nothing useful. Capacity the equipment was paid for is simply not delivered. Flood the coil and liquid reaches the outlet, then travels down the return line into a machine built to compress vapour.
Neither condition announces itself at the vent. A coil running short still blows cold air, only less of it than the model was rated to deliver. A coil running flooded blows cold air as well, while damage collects in a part nobody can see. Metering is judged against readings for that reason, and never against how the room feels.
Constant movement is the design, not a symptom
An electronic valve rarely sits still on a running system, because the conditions it answers to keep shifting. Small corrections, made continuously, are what a steady superheat looks like from the inside. An owner who hears that the valve is always adjusting has been told the machine is working.
The abnormal state is a valve that has stopped moving. A shaft seized by debris or corrosion, a stepper that has lost its drive signal, and a board that has stopped issuing steps all end in the same place. The opening freezes at whatever size it last held, and the system reverts to fixed metering without anybody choosing that.
A valve stuck part closed reads as low gas
A partly closed valve and a partly empty circuit make the same complaint, because both starve the coil. Less refrigerant reaches the evaporator than the design assumed. Low-side pressure falls. Cooling weakens, the flat takes longer to settle, and in the worse cases part of the coil frosts over. Nothing in that description separates the two.
The gap then gets filled by the likelier cause, and the likelier cause is refrigerant. Loss through a joint is common and valve failure is not, so a weak-cooling call gets read as a leak before anything is measured. The hose comes out and the charge goes up. The flat may feel marginally better afterwards, and it may not, but the restriction is untouched either way.
The billing pattern that follows is the reason to know any of this. A top-up billed against a jammed valve gets paid for and leaves nothing behind. The complaint returns, and the second visit reads that return as proof of a larger leak. A leak search is then quoted for an opening that never existed. Two invoices have gone by and the metering stage has not been examined once.
Three things separate the two before the circuit is opened. Charging history comes first, since a system that has taken gas more than once with no leak point ever named is describing a flow problem rather than a loss. Room selectivity comes second, because branch metering means a fault at one valve leaves the other rooms cooling normally, while a circuit short of refrigerant pulls every room down together. Readings settle it, and the pair that settles it is superheat weighed against subcooling.
| What the household reports | What a refrigerant shortage would need | What a jammed valve explains directly |
|---|---|---|
| What the household reportsGas has gone in more than once and no leak point was ever named | What a refrigerant shortage would needAn opening nobody has managed to locate across repeat visits | What a jammed valve explains directlyNothing left the circuit, so a search was always going to come back clean |
| What the household reportsOne room underperforming while every other room holds its setting | What a refrigerant shortage would needAwkward, since a shared circuit loses its charge for everybody | What a jammed valve explains directlyDirect, because that branch carries a valve of its own |
| What the household reportsCooling that fell away with no run-up to it | What a refrigerant shortage would needPossible only on a fast loss through a sizeable opening | What a jammed valve explains directlyOrdinary. A shaft that seizes gives no notice beforehand |
| What the household reportsA charge added against a gauge rather than weighed in | What a refrigerant shortage would needUnproven. The circuit was never confirmed short of anything | What a jammed valve explains directlyConsistent, since a restriction produces gauge readings of the same shape |
The sensor gets blamed last and should be cleared first
A valve does what it is told. If the thermistor reporting coil temperature has drifted, the board issues a wrong instruction and the valve carries it out faithfully. The circuit then behaves as though the metering has failed, while the metering is doing exactly what it was asked to do.
That matters because the two repairs sit in different categories of job entirely. A sensor sits outside the sealed circuit and comes off with a screwdriver. A valve sits inside it, and reaching one means recovering the refrigerant first. Clearing the outside-the-loop suspect before committing to the inside-the-loop one is the order any sound sequence follows.
Metering is diagnosed by reading, not by looking
Nothing about the metering stage can be settled by looking at it. It sits inside the sealed circuit, the refrigerant around it is invisible, and the valve body is opaque metal that gives away nothing about the shaft position within. Opening the circuit to see is not a casual step, because everything inside has to come out first.
What can be observed is the effect. Pressure and pipe temperature taken either side of the metering stage describe what is passing through it. A restriction shows as too large a drop across that stage while the circuit still holds a full charge. Both halves are needed. Pressure on its own cannot tell a starved coil from an empty circuit.
The electronic version carries one advantage the fixed kinds lack, which is that it can be commanded. Service software can drive the valve open and closed while the readings are watched. A valve that will not move gives a flat response that no shortage of refrigerant would ever produce. Some boards also report the step position they believe the valve holds, and that claim can be set against what the pressures say is happening.
The order of the work is what protects the customer here. Charge gets confirmed by weight before the valve is blamed, because a circuit genuinely short of refrigerant imitates a restriction in every reading. Airflow gets confirmed before either, because a coil starved of air and a coil starved of liquid look alike from the room. The valve is what remains standing once those are cleared, not the first place to reach.
- Whether the charge was established on a scale rather than judged off a gauge
- Which readings were taken either side of the metering stage, and under what running condition
- Whether the valve was driven open and closed, and what it did when it was
- Whether the sensors feeding the board were checked before the valve was condemned
- Whether the branch showing the fault is the only branch affected
What an owner can settle without tools
Two facts decide whether any of this applies, and both are available in the flat. The first is whether the system modulates at all, which the model number on the outdoor unit will confirm. Fixed metering on a single-speed unit fails in its own way and is a separate conversation.
The second is the charging record. How many gas visits this system has had, and whether any ended with a leak point actually named, is worth more to a technician than anything else you can supply. Writing it down takes nothing, and it narrows the search before anyone has arrived.
Ready to get started?
Tell us what’s going on. Symptoms, setup, photos, anything we should know. We’ll assess and come back with the right next step.