HVAC Delta T: Why 20°F Is the Wrong Target

The short version

Delta T is the difference between the temperature of the air going into the evaporator coil and the temperature of the air coming out. The rule of thumb the trade teaches is 20°F.

The 20°F target is not universal. The correct split depends on the entering air, the airflow, and the equipment, and it comes from the manufacturer’s published performance data or an applicable diagnostic procedure. The published target tables show how wide the range of correct splits runs. On a published temperature split table adapted from Carrier and from California’s Title 24 procedures, a 75°F return dry bulb temperature gives a target split running from about 23.6°F when the return air is dry down to about 12.7°F when it is humid, at the same entering dry bulb temperature; only the moisture changed. Read a 17°F split on a system whose target is 22°F and you have found a condition worth diagnosing. Read a 17°F split on a system whose target is 16°F and the split has identified nothing.

The customer pays for that confusion. A tech who reads 17°F, calls it low, and adds refrigerant to a system that was never low has now overcharged a system that still does not cool, because the actual fault was airflow. The house stays warm, the head pressure goes up, and a filter and a coil cleaning turn into a compressor conversation.

What delta T actually measures

Delta T, also called the temperature split, is the dry bulb temperature of the return air minus the dry bulb temperature of the supply air. You take one temperature in the return, one in the supply, and subtract.

What it tells you is the sensible temperature drop of the air across the equipment, which stands in for what the coil is doing. That is not the same as how much heat the coil is removing in total, because some of the coil’s work goes into pulling moisture out of the air and never shows up as a temperature change. Pair it with airflow and you can estimate sensible capacity. Two of the biggest movers of that split are capacity and airflow, and the difficulty of the measurement is that they are not independent of each other:

How much heat the coil can absorb. A system low on refrigerant, or with a metering device that is underfeeding, cannot absorb as much, so the air comes out warmer and the split narrows.
How much air is moving over it. Cut the airflow and each pound of air stays longer against the coil, comes out colder, and the split widens even though the system is doing less total work.

Change the airflow and you do not just change the split; you change what the coil is doing. Sensible and latent capacity shift with it. So you cannot reason about the two causes one at a time, which is why the split has to be read against known airflow rather than on its own.

That second one is also what makes the split dangerous when you treat it as a charge indicator. A wide split does not mean the system is producing more cooling. Low airflow can make the air leaving the coil colder while the system moves less total heat out of the house.

And a normal split does not prove the system is healthy. This is the part that gets lost. It tells you the air-side temperature drop is consistent with the target; by itself it establishes neither correct charge nor a healthy system. In its Solving Delta T piece, HVAC School calls the split an unreliable test of charge: you are as likely to read a 20°F split on an improperly charged system as a 12°F split on a properly charged one. Read that as a way of making the point rather than as a measured rate. The DOE’s own airflow and charge guidance treats the split the same way, as the preliminary reading that determines whether a full diagnostic is warranted, with charge, liquid line restrictions, and non-condensables all checked separately afterward.

So the split is a screening measurement. It tells you when to look harder. It never tells you the system is healthy.

Where 20°F came from

The 20°F rule of thumb is real, a fixture of trade teaching, and not wrong so much as incomplete. HVAC School has an article whose title is a fair summary of the problem: it calls 20 delta T a lazy rule of thumb.

The honest version of the rule is a band rather than a single target. Resideo, the Honeywell Home manufacturer, publishes 16°F to 22°F as a reasonable cooling-side range, and treats readings below roughly 16°F as pointing at the narrow-split causes covered below and readings above roughly 22°F as pointing at low airflow from a dirty filter, a dirty evaporator, or a blower wheel that needs cleaning. Those are Resideo’s boundaries for a typical system, not a line where a system becomes faulty. A unit whose published target is 16.5°F is not broken at 17°F.

Even that band is a starting point rather than a target. The Air Conditioning Contractors of America (ACCA) has published a worked example on its blog where the manufacturer’s own data called for a required drop of 16.5°F, and the article says it plainly: the split is not one size fits all.

My take

The reason 20 persists is that it is easy to remember and it is close enough to right often enough to feel true. It is the 3,000 mile oil change of this trade. Everybody knows the interval and nobody looks it up, and the interval the manufacturer actually publishes for the car in the driveway is something else entirely. A round number that is easy to repeat gets used ahead of a correct number that has to be looked up, every time.

That is exactly what makes it expensive. A rule that is right most of the time trains you to stop checking, and the calls where it is wrong are the calls where somebody adds refrigerant to a system that did not need it.

How to find the right number for the system in front of you

The split you should expect depends on the condition of the air entering the coil, specifically its dry bulb temperature and its humidity. Warmer, wetter return air produces a narrower split, because more of the coil’s capacity goes into pulling moisture out of the air rather than lowering its temperature. Cooler, drier return air produces a wider one.

ACHR News describes the working method: take the return air dry bulb temperature and the relative humidity, then look up the required temperature drop on a chart. For the equipment-specific number, the chart to use is the manufacturer’s performance data for the model in front of you, which is exactly where ACCA’s worked example gets its target. That published number, not a rule of thumb, is what you compare your measurement against.

Here are two practical notes on using the chart.

The chart is equipment-specific. Expanded ratings are the manufacturer tables that give performance across a range of conditions. The chart is in the model’s product data, expanded ratings, or service literature, which can mean a manufacturer portal that asks which company you work for before it shows the chart. If you cannot get to the chart in the field, the 16°F to 22°F band is a fallback, not an equivalent. Remember that you are working from a fallback.

Much of the difference here is humidity. On a humid day a system in perfect condition can produce a split in the mid-teens and be operating exactly as designed. Reading that as a fault is one way a good system gets refrigerant it does not need.

How to take the reading

The arithmetic is simple. The errors live in the setup, the instruments, and the waiting.

Let the system run fifteen minutes. This one is not a rule of thumb. California’s Title 24 charge procedure states it directly: run in cooling for at least fifteen minutes to let temperatures and pressures stabilize before taking the prescribed readings. Proctor Engineering’s field protocol uses the same fifteen minutes. ACHR News’s Joe Marchese asks for about 10 to 15 minutes after startup so the system can reach steady state, and warns that readings taken earlier lead to misdiagnosis. Work to fifteen minutes. Readings taken before the system settles are unreliable. Wait it out.

Take the return reading in the return plenum. California’s Title 24 charge procedure, which is the most prescriptive published version of this measurement, specifies a dry bulb sensor inserted into the return plenum, and the same placement appears in the field protocols built on it. If you have to make a probe hole to get there, kill the power first and know what is behind the metal; the DOE’s guide specifically warns against drilling between the blower and the coil, because a bit there can puncture the coil and release refrigerant. Do not take the temperatures in the room, and do not take them at the return grille if you can reach the plenum instead. The two can be close on a short insulated return and far apart when the return runs through an attic or leaks, and that difference is exactly what a grille reading leaves out.

Take the supply reading in the supply plenum, not at a register. HVAC School is blunt about this being where techs go wrong. Air leaving the coil is turbulent and poorly mixed, and a probe that can see the cold coil surface reads low from radiant effect on some instruments. Get far enough downstream to be in settled, mixed air. There is no universal distance that works; plenum geometry determines it.

Use two calibrated probes at the same time where you can. The DOE’s tool list prefers exactly that: a thermometer with two probes reading supply and return simultaneously. It is the better method on a system holding steady. Two thermometers with different calibrations give you a split that includes their disagreement, so compare paired probes in the same stable air before you start. That is a sanity check for gross mismatch, not a calibration; calibration is its own procedure. If you only have one instrument, moving it between the two points does cancel a fixed calibration offset. What it does not cancel is probe response, thermal lag, or the system changing between the two readings, so take them close together.

Record the return humidity at the same time. Without it you cannot look up the required drop on a humidity-indexed chart, and the fallback band is all you have left.

Give yourself a tolerance. Do not treat a 1°F miss as a fault. The DOE’s diagnostic guideline uses plus or minus 3°F against the calculated target, applied after airflow has been verified by direct measurement, and California’s older split procedure used the same band before the state retired the split test from its current charge verification method. The number is specific to those procedures rather than a universal HVAC tolerance, and a split inside the band does not verify airflow; only measuring airflow does that. Above the band, go look at airflow and blower settings.

Below it, move to the refrigerant side the way the equipment requires, which starts with identifying the refrigerant. On new equipment that increasingly means an A2L refrigerant such as R-454B, and the manufacturer’s service literature applies: Trane’s R-454B manual requires you to use only R-454B rated service equipment and components, never to puncture refrigerant tubing, and to disconnect all power and follow lockout procedures before servicing. Hooking up an unrated gauge set is not a shortcut on that equipment; it is a hazard. Work the full A2L procedure in the manual, not just the tool rating.

What a wide split means

A split wider than the system should be producing is a reason to go verify airflow, and it is the more useful of the two directions. It does not prove low airflow. Resideo lists abnormally low return humidity as a separate cause of a wide split, which is the same latent effect running the other way: dry air gives the coil less moisture to condense, so more of its capacity shows up as temperature drop.

With that ruled out, the usual causes are on the airflow side, listed roughly in the order worth checking:

A loaded filter. It is the first check, and the one that costs nothing but a look.
A dirty evaporator coil. It is worth its own inspection, because the filter and the coil are separate causes. A clean filter does not settle the question, and coils get dirty from the back side where you cannot see them from the front.
A blower wheel packed with dust. The wheel loses its ability to move air long before it stops turning.
A blower speed set wrong. It is common after an equipment change where the tap was never adjusted for the new coil or duct.
Undersized or restricted ductwork. It is potentially the most invasive fix on the list, and the one that takes component by component pressure tests to confirm rather than the split.

The split tells you to go look at airflow. It does not tell you where the restriction is. An excessive total external static pressure is a sign that a hidden airflow problem exists, and it does not identify the restriction; pressure tests across each component track that down. Static pressure is also not the same thing as airflow: it tells you what the blower is fighting, and an airflow measurement tells you how much air is actually moving.

What a narrow split means

A narrow split is the harder direction, because the list of causes is longer and some of them look alike.

Low refrigerant charge. The coil cannot absorb as much heat, so the air leaves warmer.
Too much airflow. It is real, and worth checking for after a blower tap was set high to fix a comfort complaint.
High humidity in the return air. It is not a refrigeration fault, which is the point, though what is keeping the return air that humid can be its own question. Check the required drop before you diagnose.
A restriction in the liquid line. It starves the evaporator the same way a low charge does.
A malfunctioning metering device. Resideo’s low split list has it alongside the charge and airflow causes.
A leaking return duct.
A leaking reversing valve on a heat pump. Both are on Resideo’s low split list, alongside the metering device.
A compressor that is not pumping. It is the last item on the list, and one the split cannot confirm; the manufacturer’s compressor diagnostics determine it.

With airflow verified and the entering air recorded, suction pressure is a useful reading, though it is a clue rather than a verdict. A low charge and a restriction both drop it. A high latent load tends not to, because the coil is absorbing more heat and the evaporator does not run as cold. How cleanly any of that shows up depends on airflow, outdoor conditions, coil condition, and how the equipment controls itself, so read suction pressure alongside the entering air condition and verified airflow rather than on its own.

What suction pressure will not do is tell the fault cases apart. A low charge and a liquid line restriction produce similar symptoms, and the DOE is explicit that separating them takes airflow verification plus superheat and subcooling, not a single reading.

Common mistakes

Look up the required drop for the equipment and the conditions before deciding a split is wrong
Let the system run until it stabilizes, which is at least fifteen minutes under the procedures cited here
Measure return humidity along with return temperature
Use two probes at the same time where you can, and compare them in the same stable air first
Treat a wide split as an airflow finding and go measure static pressure

Do not

Compare any reading to 20°F without checking what the equipment calls for
Use room temperature as your return reading
Add refrigerant on a narrow split alone
Take the supply reading where the probe can see the coil
Assume a clean filter means the coil behind it is clear

Frequently asked questions

What should delta T be on an air conditioner?

There is no single correct number. The required temperature drop depends on the dry bulb temperature and humidity of the air entering the coil, and the manufacturer publishes it for that equipment. As a fallback, 16°F to 22°F is a commonly cited range, and ACCA has published a worked example where the manufacturer’s data called for 16.5°F. Compare your split to what the equipment calls for under the conditions you measured, not to a rule of thumb.

Is 20°F delta T correct?

Sometimes it is. Twenty falls inside the range many residential systems produce under common conditions, which is why the rule persists. It is not a target, and treating it as one means a system that should be producing 16°F looks low and a system that should be producing 23°F looks fine.

Does a low delta T mean low refrigerant?

A low delta T does not mean low refrigerant by itself. A narrow split can come from low charge, too much airflow, humid return air, a liquid line restriction, a malfunctioning metering device, a leaking return duct, a leaking reversing valve on a heat pump, or a compressor that is not pumping. The split alone does not say which. Humid return air can narrow the split with nothing wrong in the refrigerant circuit, though humidity that stays abnormally high can be its own problem to run down. Check the required drop for the conditions before you assume the system is low.

What does a high delta T mean?

A high delta T is a reason to investigate airflow: the filter, the evaporator coil, and the blower wheel and its speed setting, in whatever order access allows. When low airflow is the cause, the cooling is being spread over fewer pounds of air, so each of them leaves colder. It is not proof of an airflow problem, though. Abnormally dry return air widens the split too, because the coil has less moisture to condense and more of its capacity shows up as temperature drop.

How long should the system run before I measure delta T?

Run the system for fifteen minutes at minimum. California’s Title 24 charge procedure runs the system at least that long. ACHR News warns that readings taken before steady state lead to misdiagnosis. Readings taken before the system settles are unreliable.

Glossary

A2L: the ASHRAE class for mildly flammable refrigerants such as R-454B.
ACCA: Air Conditioning Contractors of America, whose worked example is cited here.
ACHR News: a trade publication cited here on run time before measuring.
Delta T: the temperature drop from return air to supply air across the equipment.
Dry bulb: the air temperature an ordinary thermometer reads.
HVAC School: a practitioner training and reference organization cited here.
Latent capacity: the share of cooling that removes moisture rather than lowering air temperature.
Plenum: the duct box directly at the equipment, where a probe reads cleaner than a grille.
Sensible capacity: the share of cooling that lowers air temperature rather than removing moisture.
Steady state: the condition after pressures and temperatures stop moving, when a reading means something.
Temperature split: another name for delta T, the return-to-supply temperature drop.
Title 24: California’s building energy code, one source of the split table cited here.

Drafted with AI assistance and reviewed by the author.

← All posts