Flare vs brazed aircon joints: which one leaks first
A split system is sealed at its joints, and there are two ways to make one. One can be undone and remade; the other has to be cut out. Nobody shows an owner which sits where, and that decides where a slow loss turns up years later.
By Team Snowflake | Updated 7 Aug 2026
Two ways to close a refrigerant circuit
A split system is a sealed copper loop, and every place two lengths of it meet has to keep holding pressure years after anyone last looked at it. Two methods do that job. One clamps metal against metal. The other fuses metal into metal.
A flare is a mechanical seal. The end of the tube is opened out into a cone, and a nut behind that cone pulls it hard against a matching taper on the fitting. Nothing is melted and nothing is added. The seal holds because two shaped copper faces are pressed together under force. Slacken the nut and the joint comes apart with both pieces intact.
A brazed joint is metallurgical. The two pieces are heated, and a filler alloy is drawn into the gap between them, where it sets into a continuous bridge of metal. There is no nut, no sealing face, and no way back. Undoing one means cutting the pipe.
Which method appears where is mostly settled by the equipment rather than by preference. Indoor and outdoor units arrive with flare fittings already on them, and with their own nuts supplied in the box. Daikin's R32 split installation manual tells the installer to use the nut fixed to the main unit, and treats forming the cone on site as an ordinary step of the job. The connections at the two units are flared because that is what the hardware is built to accept.
Brazing covers everywhere else. Copper is supplied in coils of a set length, so a route longer than a coil has to be joined somewhere in the middle. Extensions, reroutes during a renovation, and repairs to a damaged length are all made the same way. A typical flat therefore carries a small number of flares at fixed, known addresses, and any brazed joint sits out along the run instead.
| Where on the circuit | Which joint is usually there | Why it ends up that way |
|---|---|---|
| Where on the circuitAt the indoor unit connections | Which joint is usually thereFlared | Why it ends up that wayThe unit ships with flare fittings and supplies its own nuts |
| Where on the circuitAt the outdoor stop valves | Which joint is usually thereFlared | Why it ends up that waySame reason, and the joint has to be opened again to charge the system |
| Where on the circuitPart-way along a long run | Which joint is usually thereBrazed | Why it ends up that wayCopper comes in fixed lengths, so a longer route is joined mid-pipe |
| Where on the circuitWhere a run was extended or rerouted later | Which joint is usually thereBrazed | Why it ends up that wayNew copper meets old, with no equipment at that point to supply a fitting |
| Where on the circuitWhere a damaged length was cut out | Which joint is usually thereBrazed | Why it ends up that wayThe replacement section is spliced into the middle of the existing pipe |
The union that exists but rarely gets buried
Fittings that flare two lengths of pipe to each other do exist, so the two methods are not strictly divided by position. They stay uncommon in the middle of a residential run, and the reason is worth holding on to.
A clamped seal is a poor thing to bury inside a wall. It can slacken, and nobody will ever look at it again. Brazing wins mid-run partly because it is stronger, and partly because it asks nothing of a place nobody can get to.
Why the flare is the joint that usually leaks
The flare carries most of the leak risk on a residential circuit, and its own description explains why. It is the only joint held closed by clamping force instead of fused metal. Force can relax. Fused metal cannot.
It is also the joint made on site, in whatever conditions the site offers. A cone gets formed on the end of a tube at the top of a ladder or out on a ledge, and its quality rests on a few minutes of work. The manual is specific about what a sound one looks like. The pipe end has to be opened evenly into a true circle, and the inner face has to be free of flaws. A scratch across that face is a channel, and no amount of tightening closes a channel.
Tightening has a trap at both ends. Too loose and the faces never fully seat, so the joint passes on handover day and gives up quietly afterwards. Too tight is the less obvious one. Daikin's manual warns that an over-tightened nut can crack after prolonged use and release refrigerant, which puts the damage on installation day and the symptom years later. Every pipe size has a published tightening figure, and hitting it takes a tool rather than a practised hand.
Reused copper starts a flare at a disadvantage. A tube that has been flared before has to be cut back and formed again, on metal already stretched once and aged in a wall. The same manual takes the blunt line and tells installers never to use piping from a previous installation. On a particular job the decision to reuse existing pipes has real answers on both sides. The narrower point belongs here: every reused end is a fresh cone on tired metal.
Vibration is the slow cause. The outdoor unit runs a compressor and a fan, and both put a constant tremor into the pipework bolted to them. A joint held by clamping force sits in that tremor year after year. Anything that adds to it feeds the connections attached to it: a stand that is not level, a bracket slackening at its fixings, a run left unsupported near the unit.
The signature is a seep, not a burst
A failing flare rarely announces itself. The opening is a hairline gap between two clamped faces, so the charge leaves slowly and the room registers it before anyone suspects a joint. Cooling fades across weeks rather than stopping.
Refrigerant carries oil, so a long-running seep sometimes leaves a dark film on the nut or on the insulation beside it. That is one of the few things an owner can see without tools. What the loss feels like from the room belongs elsewhere, and the refrigerant leak signs are worth reading on their own terms.
The service port is the same story in miniature. It is a seal opened on purpose and asked to close again afterwards, and the schrader valve core inside it fails for reasons that belong to it alone.
What does a brazed joint bring with it?
A brazed joint is stronger than a flare and carries a different set of risks. Once made properly it is effectively part of the pipe, and no amount of vibration loosens it. What it can carry instead is whatever happened inside the tube while it was being made.
Heating copper in open air makes scale. Oxygen inside the tube reacts with the hot metal and leaves a black flaky oxide on the inner wall. Trade practice is to flow dry nitrogen through the pipe during the work, so there is no oxygen left in there to react. Where that step is skipped, the scale stays in the circuit.
Scale does not stay where it formed. It breaks loose later and travels with the oil, then settles wherever the circuit narrows. The filter drier, the strainers and the metering device are the usual destinations. The result is a refrigerant restriction with a cause nobody would connect to a torch lit years earlier.
The symptoms it produces look nothing like a joint problem. A restricted circuit cools poorly, ices in an odd place, or runs at pressures a technician can read and an owner cannot. A leak search on that system finds nothing, because nothing is leaking.
Brazing also puts an open flame on a circuit that holds pressure, so the charge has to be out of the pipe before a torch comes near it. That is the job refrigerant recovery does. It is also why a brazed repair is never as quick as a nut. The work runs in a sequence, and skipping the front of it is how a small repair contaminates a whole system.
None of this makes brazing the worse method. On a long run it is not optional, and a sound brazed joint will outlast several flares. Each method simply fails in its own way. A flare fails at its face, from how it was formed and what has been pulling at it since. A brazed joint fails from what was inside the pipe when it was made, or from heat that reached something sitting nearby.
What each joint means when something has to change
The difference stops being academic the day a unit has to move. A flare is designed to be undone, which is exactly why the connections at the stop valves are flared. The charge has to get in there somehow, and the unit has to come off the wall eventually.
Relocations and unit swaps run on that reversibility. Taking an indoor unit off its bracket means slackening two nuts. Putting a new one back means forming two fresh cones. The pipework survives the operation. On a circuit where every connection was brazed, the same job opens with a pipe cutter.
Changing a brazed joint means cutting it out and making a new one beside it. Each cut shortens the run a little, and each repair leaves a joint where there was not one before. A stretch that has been opened three times carries three brazed joints, all in the same area, all sited by whatever access made the work possible.
Any joint that has been remade has to be proven before the system goes back into service, whichever kind it is. A nitrogen pressure test is where that proof comes from, and a joint remade without one is an assumption. It matters more on a brazed repair, because the metal being tested was molten during the work.
| What has to happen | At a flared connection | At a brazed joint |
|---|---|---|
| What has to happenRemoving an indoor unit for work | At a flared connectionSlacken the nut, and the pipe stays whole | At a brazed jointCut the pipe, which shortens the run every time |
| What has to happenMaking good a joint found leaking | At a flared connectionCut the end back, form a fresh cone, remake it | At a brazed jointCut the joint out and braze a new section in |
| What has to happenExtending a run to a new position | At a flared connectionRare mid-run, since a clamped joint would be buried | At a brazed jointThe normal method, and the reason mid-run joints exist |
| What has to happenProving the repair before gas goes back | At a flared connectionPressure test, then vacuum, then charge | At a brazed jointThe same sequence, with more riding on the reading |
The joint you cannot reach
Access decides which of these turns expensive. A flare at an outdoor stop valve sits in the open and can be remade where it is. A brazed joint above a ceiling cannot be cut until somebody gets to it, and getting there is what does the damage.
Concealed piping settles that, and it gets settled at renovation rather than at the fault. A mid-run joint behind plaster is not a worse joint than one behind a removable cover. It is the same joint with a far larger bill attached to touching it.
What an owner can ask about the joints
None of this needs anything opened. The connections at the outdoor unit are already visible, and the rest is a question of what was done and whether anyone wrote it down.
Ask where the joints are before asking whether they are sound. A circuit with flares at the unit connections and no mid-run join has a short list of suspects at known addresses. A circuit extended during a renovation has one more somewhere behind a finish, and usually nobody remembers where. That single fact changes what the leak detection methods cost to run, and the person who ran the pipe is the only one who knows it.
Ask what happened at the joints during the last visit that touched them. A unit taken off the wall for a chemical overhaul comes off at its flares, and those flares get remade to put it back. A joint remade is a joint back on the list. The answer worth having is not that the work was careful. It is whether the system held pressure afterwards, and what the reading was.
Ask about the outdoor end specifically, because that is where clamping force and vibration meet. A condenser on a stand that rocks, or on brackets rusting at their fixings, works a load into two flared connections every hour the unit runs. Steadying the mounting costs less than chasing the joint it is slowly loosening.
Five things to put to whoever piped it
Every line below has a factual answer that somebody either holds or does not. No refrigeration knowledge is needed to put them, and a reassuring tone does not answer any of them.
- Does the pipe run join anywhere between the two units, and can that point be reached?
- Were the connections at each unit newly flared, or were existing ends reused as found?
- Was nitrogen flowed through the pipe during any brazing done on this system?
- After any joint was remade, did the circuit hold pressure before the refrigerant went back in?
- Do photographs exist of the run and its joints from before the casing went on?
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.