HVAC temperature split and delta-T diagnostics field guide

The temperature split, and what it actually tells you

The air-side temperature split, also called delta-T or the air temperature drop, is the difference between the return-air temperature going into the cooling coil and the supply-air temperature coming off it. Return air at 76°F and supply air at 56°F is a 20°F split. That single number tells you something real in about a minute: whether the system is moving roughly the right amount of air across the coil and pulling roughly the right amount of heat out of it.

It works because the air gives up heat to the coil. How far the air temperature falls depends on two things fighting each other. How much heat the refrigerant side is removing, and how fast the air is moving past the coil. Slow the air down and each pound of air sits longer against the cold coil and leaves colder, so the split goes up. Speed the air up, or lose cooling capacity, and the split goes down.

That is the whole value and also the whole trap. The split moves when airflow changes and it moves when capacity changes, so a number alone does not tell you which one moved. It points you at a problem and tells you which direction to look. It is a screening tool, and the rest of this guide is about reading it without fooling yourself. For setting the actual charge, see the companion guide on charging by superheat and subcooling. For setting the airflow itself, see the air balancing guide.

How do you take a temperature split?

Measure the dry-bulb return-air temperature and the dry-bulb supply-air temperature, then subtract. The difference is the split. Read the return at the filter or the return grille, and read the supply at the first supply register or in the supply plenum a foot or two downstream of the coil. Two thermometers reading at the same time beats one probe moved back and forth, because the conditions drift while you walk.

Where you put the probe decides whether the number means anything. Read the supply far enough off the coil that you are in mixed air, not staring at a cold spot on one circuit. Stay away from any duct seam or boot that leaks, because return-temperature air bleeding into the supply will flatten the reading. On the heating side, never take a supply reading where you can see the heat strips or the burner section, because radiant heat off the elements lies to the probe.

Use a thermometer you trust and let it settle. A cheap probe that reads 2°F off on each end can swing the split by 4°F, which is the difference between a pass and a callback. Match the two instruments against each other in the same air before you start so you know they agree. The split is only as good as the two numbers you subtract.

The normal cooling split, and what it is not

A working cooling system usually lands somewhere around 16 to 22°F dry bulb, with about 20°F being the number most techs carry in their head. That is a rough target for screening, not a spec. The honest version is that the range floats with the indoor humidity, the airflow, and the equipment, and it can sit a few degrees outside that window and still be correct for the conditions in the house.

Treat the range as a first look, not a verdict. A split inside it on a dry day with normal airflow is a reasonable sign the system is doing its job. A split well outside it tells you something is off and which way to start looking. What it does not do is prove the system is right or wrong on its own, because the same 20°F can sit over a system that is undercharged and starved for air at the same time, with two errors canceling.

The biggest reason the plain range misleads people is humidity, covered next. The second biggest is that they reach for the split as a charging method. It is not one. It screens the air side. The numbers that set the charge come from superheat and subcooling, not from the air temperature drop.

Why does humidity change the temperature split?

Indoor humidity lowers the dry-bulb split even when the system is doing everything right, and this is the single most misread part of the whole check. On a humid day the coil spends part of its capacity pulling moisture out of the air, which is latent heat. Latent work condenses water on the coil and shows up as no change in dry-bulb temperature, so the air leaves cold and wet rather than colder and dry. The thermometer only sees dry-bulb, so it reads a smaller drop, and a perfectly good system on a sticky afternoon can show a 14 to 16°F split.

Run the dry climate the other way. Low indoor humidity means almost all the coil's capacity goes into dropping air temperature, so the same equipment posts a 22 to 25°F split on a dry day and nothing is wrong. Same machine, same charge, same airflow, different split, because the moisture in the room changed.

So the split has to be read against the indoor moisture, which you capture with the return wet-bulb temperature or the return relative humidity. A 16°F split at 65°F return wet bulb can be exactly on target. The same 16°F at 55°F return wet bulb says you have lost capacity. Read the dry-bulb split alone on a humid day and you will condemn good systems and chase charge that is already correct.

Target split by return humidity

The right target split is the one for the air actually entering the coil, which means the return dry bulb and the return wet bulb or relative humidity together. Higher indoor humidity means a lower target split, because more of the coil's work is going into moisture instead of temperature. Equipment makers and ACCA publish charts that map return wet bulb and return dry bulb to a target split, and on a job that has one, the manufacturer chart wins over any rule of thumb.

The table below is a representative shape of that relationship for return air near 75°F dry bulb. Use it to see the trend and to sanity-check a reading. Pull the real target from the equipment data or the ACCA chart before you make a charge or airflow decision off it, because the exact numbers move with the coil and the return conditions.

What does a high temperature split mean?

A high split, up around 23 to 25°F or more, points at low airflow first. The air is moving too slowly across the coil, so each slug of air sits against the cold surface longer and leaves colder than it should. The capacity has not gone up. The air has slowed down, and the bigger temperature drop is the symptom.

The usual causes line up in order of how often they bite. A loaded filter is first and most common, and a filter that looks fine in the rack can still be choking the system at running static. A dirty evaporator coil is next, since a fouled coil throttles air the same way a dirty filter does. After that comes a blower problem, a slow belt or a low tap or a wheel packed with dirt, then duct restriction from undersized return, crushed flex, or closed dampers.

There is a trap on the high side. A coil that is freezing or already iced will also throw a high split for a moment and then fall apart as the ice blocks the airflow completely. If you find a high split with a cold, sweating or frosting suction line, you are likely looking at low airflow or low charge driving the coil below freezing, and the split is the early warning. Fix the airflow before you read anything else into the number.

What does a low temperature split mean?

A low split, down around 13 to 15°F when the humidity does not explain it, points at lost capacity or too much airflow. The coil is not pulling enough heat out of the air, or the air is blowing past so fast it never gives up much temperature. Either way the dry-bulb drop is small for the conditions.

On the capacity side the common cause is a charge problem, most often an undercharge, along with anything else that hurts heat transfer: a restricted metering device, a dirty condenser rejecting poorly, or a compressor that is not pumping. On the airflow side, an oversized blower or a duct system moving more than design air can also flatten the split, which is why a low split on a system with a fresh filter and a clean coil makes you look harder at the refrigerant side.

Before you call it low charge, rule out humidity. A low dry-bulb split on a muggy day with a high return wet bulb may be exactly right, because the coil is busy wringing water out of the air. Check the return wet bulb against the target first. Only after the humidity is accounted for does a genuinely low split send you to the refrigerant numbers, and even then you confirm with superheat and subcooling, not the split.

Diagnosing the airflow side first

When the split reads high, work the air side before you ever touch a gauge. Low airflow drives the split up, fakes symptoms that look like overcharge, and will make any charging attempt come out wrong, so it has to be cleared first. The order is filter, coil, blower, then duct.

Pull the filter and look at it under load conditions, not in the parking lot. Check the evaporator coil face for dirt and the fins for damage or a mat of pet hair. Read the blower: belt tension and tap on older units, wheel cleanliness, and motor speed setting on ECM units. Then measure total external static pressure across the air handler with a manometer and compare it to the equipment's rated static. A system running well over its rated static is choked, and the high split is telling you so.

Static pressure is the single most useful air-side number and the one most techs skip. A high split plus high static is low airflow, full stop, and you fix the restriction. For setting and proving airflow at the registers, the air balancing guide covers the traverse and the proportional method. Get the air right, then read the split again before you decide anything about the charge.

The 350 to 400 CFM per ton rule

Cooling airflow is commonly set near 350 to 400 CFM per ton, and that range is the other half of the split. The split and the airflow are two readings of the same thing, since the temperature drop across the coil is what falls out of how much air you push over a given capacity. Move the rated air and a properly charged coil lands in its target split. Starve the air and the split climbs.

Where you sit in that range depends on the climate and the load. Drier climates and high-sensible loads run toward 400 to 450 CFM per ton to put more capacity into temperature. Humid climates run toward 350 CFM per ton, sometimes a touch lower, to slow the air down and pull more moisture, which deliberately trades some dry-bulb split for dehumidification. So a lower split on a humidity-tuned system in a wet climate can be by design, not a fault.

Carry the number as a check, not a setpoint to chase blind. The equipment data and the design airflow control. If the split says low airflow and the static confirms it, the CFM-per-ton math tells you roughly how far off you are and whether a blower tap or a duct fix gets you back.

The split is a screen, not a charging method

You cannot set a refrigerant charge by the temperature split, and trying to is one of the most common ways good systems get butchered. The split moves with airflow, humidity, return temperature, and capacity all at once, so it cannot isolate the charge from everything else acting on it. Add or pull refrigerant to make the split hit 20°F and you can land on a 20°F split that is overcharged, undercharged, or starved for air, because too many variables feed that one number.

What the split does is screen. It tells you the air side looks reasonable or it does not, and it points you toward airflow or toward capacity. From there you charge by the method the metering device calls for: superheat on a fixed orifice or piston, subcooling on a TXV or EEV, read against the equipment data plate and charging chart. The refrigerant-charging guide walks that through in full.

Think of the split as the triage and superheat and subcooling as the diagnosis. The split tells you whether to open the gauges at all and which way to look. The gauge numbers tell you what to actually do. Skipping the screen wastes time; skipping the diagnosis wastes refrigerant and creates the next callback.

Airflow first, then charge, every time

The order is not optional: fix the airflow, then set the charge. You cannot charge a system correctly on bad airflow, because the airflow changes the coil's saturation temperature and therefore changes the superheat and subcooling you are reading. Charge to good numbers over a starved coil and the charge will be wrong the moment the airflow is corrected.

Picture the loaded filter. Low airflow drops the suction pressure and the coil temperature, which raises superheat on a fixed-orifice system and tempts you to add refrigerant. Add it, then the customer changes the filter, and now the system is overcharged and the head pressure climbs all summer. The whole error came from charging before the air was right.

So the sequence on any charge-related call is the same. Confirm and fix the air side first, filter and coil and blower and static. Let the system stabilize. Then take superheat and subcooling and adjust the charge to the data plate. The split sits at the front of that sequence as the screen that tells you whether the air side is even close before you commit to anything.

What is furnace temperature rise?

Furnace temperature rise is the heating-side version of the split: the supply-air temperature minus the return-air temperature across a gas or oil furnace heat exchanger. The difference from cooling is that the furnace tells you the right answer. Every furnace data plate lists a temperature rise range, something like 35 to 65°F or 45 to 75°F, and the rise must land inside that range. The span between the two numbers is typically 30°F.

Too high a rise means too little air over the heat exchanger. Heat piles up on the metal, the supply runs hot, and the furnace starts cycling on its high-limit switch. That short cycling and the sustained heat are hard on the heat exchanger and shorten its life, so a high rise is not just an efficiency note, it is a durability and safety problem you fix.

Too low a rise means too much air. The blower is pulling more than the burner can heat, the supply runs cool, and in the worst case the flue gases drop below their dew point and condense inside a heat exchanger that was never built to stay wet. That condensation corrodes non-condensing exchangers from the inside. Both ends of the range are real failure modes, which is why the nameplate sets a window and not a single target.

Setting the blower to the nameplate rise

Set the blower speed so the measured rise lands inside the nameplate range, ideally toward the middle. Take the rise at steady state with the burners running and the blower at its heating speed, reading return and supply in clean spots away from the heat exchanger's radiant view. Then move the blower: raise the speed to bring a high rise down, lower the speed to bring a low rise up.

On PSC blowers that is a speed tap change on the motor. On ECM and variable-speed units it is a dip-switch or control-board airflow setting, and the heating airflow is often set independently of the cooling airflow, so check both. A unit set perfectly for cooling can still be out of the rise range in heat.

If the rise will not come into range no matter the blower speed, stop adjusting and look upstream. An undersized return, a closed zone, or a firing rate that does not match the installed altitude or fuel can push the rise out the same way a wrong blower speed does. The nameplate range is the manufacturer's limit, and staying inside it protects the heat exchanger that is the most expensive part to replace.

Heat pump and electric strip heat

Heat pumps and electric heat do not have a furnace rise spec, and reading them like a furnace will mislead you. A heat pump in heating moves heat with the refrigerant cycle, so the supply temperature runs lower than a gas furnace, often in the 90s to low 100s°F at the register depending on outdoor temperature and stage. That feels cool to a customer used to gas heat, but it is normal for the machine, and the air-side split is not a nameplate-bounded check the way furnace rise is.

Electric strip heat does behave like a rise, since resistance elements add a fixed amount of heat to whatever air passes over them. The temperature rise across the strips depends on the kilowatt rating and the airflow, and the same logic applies: too little air over the elements drives the rise and the element temperature up toward the limit. For a heat pump with auxiliary strips, you are reading two different heat sources, so know which one is running before you read the split.

Measurement errors that fake a bad split

Most bad split readings are bad measurements, not bad systems. The fixes are simple once you know where the number goes wrong.

Wrong location is the leader. A supply probe in the radiant view of the heat strips or burner reads high. A probe near a leaking duct seam or boot reads off because return air is mixing into the stream. A supply reading taken right at the coil on a single cold circuit reads colder than the mixed supply the building actually gets. Read return at the filter or grille and supply in mixed air a couple of feet downstream, away from leaks and away from any radiant source.

Bad or unmatched instruments are next. Two thermometers that disagree by a couple of degrees will throw the split by the sum of their errors, so verify them against each other in the same air first. And reading before the system is at steady state, covered next, is the third common miss. Get the location right, trust the instruments, and let it stabilize, and the split becomes a number you can act on.

Running to steady state before you read

Run the system about 10 to 15 minutes before you read the split, in cooling or in heat. A system that just started has not pulled down the coil, the building mass is still dumping heat into the return, and the refrigerant pressures are still moving, so an early reading is a snapshot of a transient and not the steady condition you are diagnosing.

The same applies after any change. Swap a filter, change a blower tap, or add refrigerant, and the readings need time to settle before they mean anything. Read too soon and you chase a moving target, adjust again, and end up worse than when you started. Let it run, let it stabilize, then read.

The wet-bulb (enthalpy) method for a real capacity check

When you need a real capacity check rather than a screen, measure across the coil with wet bulb, not just dry bulb. The dry-bulb split only sees sensible heat, the temperature change. It is blind to the latent heat, the moisture the coil removes, which on a humid day is a large share of the total work. Wet-bulb temperature, or the enthalpy you derive from it, captures both, so the total heat the coil pulls out shows up in the numbers.

The method is to read wet bulb entering and leaving the coil, convert to enthalpy off a psychrometric chart or a digital tool, and combine the enthalpy difference with the measured airflow to estimate the actual delivered capacity. That is closer to what the system is really doing than any dry-bulb split, because it counts the water as well as the temperature.

This is more work and you do not do it on every call. Reach for it when the dry-bulb split and the customer complaint disagree, when humidity is high enough that the dry-bulb number is suspect, or when you need to document delivered capacity at commissioning. The dry-bulb split screens; the enthalpy method measures.

The diagnostic workflow

Put the steps in order and the split does its job without leading you astray. Take the reading, read it against the humidity, decide airflow versus capacity, fix airflow first, then charge by the gauge numbers. Each step gates the next, which is what keeps you from charging on bad air or condemning a good system on a wet day.

• Run the system 10 to 15 minutes to steady state before reading anything.
• Take the dry-bulb return and supply across the coil in clean, mixed-air spots and subtract for the split.
• Measure the return wet bulb or RH and look up the target split for those return conditions.
• Compare the split to the humidity-corrected target, not to a flat 20°F.
• High split: go to the air side first, filter and coil and blower and static, before the gauges.
• Low split, humidity ruled out: suspect undercharge or lost capacity, but confirm on the refrigerant side.
• Fix any airflow restriction and re-read the split before touching the charge.
• Set the charge by superheat or subcooling to the data plate, not by the split.
• In heating, check furnace rise against the nameplate range and set the blower to land inside it.
• Record the split, the return wet bulb, the static, and the action so the next visit has a baseline.

Rooftop units and economizers

On a packaged rooftop unit, the split reads the same way, but the economizer can move it before refrigeration is even involved. When the economizer is bringing in outside air, the return entering the coil is a mix of building return and outdoor air, so the entering condition is not the space condition, and the split you read reflects that mix. Confirm the economizer position and the outside-air fraction before you trust an RTU split.

On a cool, dry morning with the economizer open wide, the unit may be cooling on outside air with the compressor off, and there is no refrigeration split to read at all. On a hot, humid afternoon with the economizer closed to minimum, you are reading a normal mechanical split against the mixed-air entering condition. Read the mixed-air temperature at the coil entrance on an RTU, not the space thermostat, or the split will not square with the equipment data.

Data-center CRAC and CRAH delta-T

In a data center the same return-minus-supply delta-T is a primary health number for a CRAC or CRAH unit, but the target is driven by the IT load and the airflow management, not a comfort-cooling range. A low return-to-supply delta-T on a computer-room unit usually means air is bypassing the load, cold supply short-cycling back to the return through gaps in the containment, open floor tiles, or unblanked rack spaces, so the unit moves air without picking up much heat.

Raising that delta-T is mostly an airflow-containment job: seal the bypass, blank the racks, and match unit airflow to the IT demand so return air comes back warm. The lesson carries straight from the field: a small delta-T on a cooling unit means the air is moving without doing work, and the fix is on the air side before the refrigeration side. Confirm the target against the equipment data and the facility design, since data-center setpoints sit well outside comfort-cooling assumptions.

Recording the split at startup and on PM

Record the split at startup and on every maintenance visit, and it stops being a one-off guess and becomes a trend. A baseline split taken at commissioning, with the return wet bulb and the static pressure alongside it, is the reference every later visit measures against. The same split a year later at the same conditions means the system is holding. A split that has crept up at the same conditions means airflow is degrading, usually a coil slowly loading up, long before the customer feels it.

The numbers only earn their keep if they are written where the next tech finds them. Capture them on the work order with the conditions they were taken under, because a split without its return wet bulb and outdoor temperature is half a reading. FieldOS keeps the split, the wet bulb, the static, and the action attached to the equipment so the next visit opens with the history instead of starting cold, which is how a slow airflow loss gets caught on a PM instead of on a 95°F no-cool call.

This is also the documentation that protects you. A recorded baseline that showed a clean system at startup is the evidence that a later airflow problem came from a filter nobody changed, not from the install.

Field checklist

What to document

A split with no conditions written next to it is barely worth recording, because the next tech cannot tell whether 16°F was a good humid-day reading or a bad dry-day one. Capture the split and the conditions it was taken under, every time, so the number can be compared later.

Record the return dry bulb and the return wet bulb or RH, the supply dry bulb, the calculated split, the total external static pressure, the outdoor temperature, the airflow setting, and what you did about it. On a heating visit, record the furnace rise and the nameplate range it was checked against. The table below is the short version of what makes a split reading reproducible.

Common mistakes

• Using the temperature split as a charging method instead of a screen.
• Reading the dry-bulb split without the indoor humidity and condemning a good system on a humid day.
• Charging refrigerant before fixing a known airflow restriction.
• Setting furnace temperature rise outside the nameplate range, high or low.
• Measuring near a duct leak or in the radiant view of strip or burner heat.
• Reading the split before the system has reached steady state.
• Reading an RTU split off the space temperature instead of the mixed-air entering condition.
• Trusting two thermometers that were never checked against each other.

Standards and references

The split ranges and the airflow rule are field practice anchored to a few sources, and the right move is to hedge them to the equipment and the humidity rather than treat any one number as law. ACCA addresses air-side diagnostics and airflow, and the 350 to 400 CFM per ton figure and the target-split-by-humidity idea trace to that body of work and to manufacturer charting. The target split for a given return wet bulb belongs to the equipment maker's chart or the ACCA reference, not to a memorized 20°F.

The furnace temperature rise range is the manufacturer's, printed on the data plate, and that nameplate range governs the heating setup. Do not substitute a generic number for the plate. ASHRAE provides the psychrometric framework behind the wet-bulb and enthalpy methods and the design conditions the equipment was selected against, which is why the enthalpy approach is the one to reach for when you need true delivered capacity.

Across all of it, three things hold. Humidity shifts the dry-bulb split, so read it against the return wet bulb. Airflow comes first, before the charge. And the split screens, it does not set the charge, which is the job of superheat and subcooling. Confirm the specific targets against the equipment data and the conditions on site.

Units, terms, and conversions

The split and its cousins go by several names across a service ticket, a manufacturer sheet, and a TAB report, so the same idea reads differently depending on where you see it.

The air-side temperature split is also called delta-T, the temperature drop, or the air temperature difference across the coil. Dry-bulb temperature is plain air temperature; wet-bulb temperature reflects the moisture and is what you read to account for humidity. Relative humidity is the percentage form of the same moisture information. Enthalpy, the total heat in the air, is read in Btu per pound of dry air. Airflow is in cubic feet per minute, often expressed per ton of cooling, where one ton equals 12,000 Btu per hour. Static pressure is in inches of water column, written in. w.c. or in. wg.

Temperature split / delta-T

Return-air temperature minus supply-air temperature across the coil, in °F

Dry bulb

Plain air temperature, what a standard thermometer reads, blind to moisture

Wet bulb

Temperature that reflects air moisture, used to read the split against humidity

Enthalpy

Total heat in the air, sensible plus latent, in Btu per pound of dry air

Furnace temperature rise

Supply minus return across a furnace, held inside the nameplate range

CFM per ton

Cooling airflow per 12,000 Btu/h of capacity, commonly 350 to 400

External static pressure

Air-side pressure the blower works against, in inches of water column

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• Do I charge by superheat or subcooling?
• What is air balancing?
• What is delta-T in a data center?
• What underfloor static pressure should a data center plenum hold?
• What is an air curtain?