Aircon diffuser: what the fitting in the ceiling decides

A diffuser gives the air a shape on the way out. That is its whole job. Price Industries defines a grille as an outlet with a square or rectangular face and neck, faced with fixed or adjustable louvres that deflect the air. A register, in the same guide, is a grille carrying a damper for volume control.

The shaping goes past deflection. Concentric cones, a plaque, or a row of pattern controllers turn the discharge into a form the maker has measured and published. That form is the property being bought. Two fittings can pass identical air into an identical room and put it in different places.

None of that gets worked out on site. Titus states that its performance data comes from tests run to ANSI/ASHRAE Standard 70. Price puts the pressure side the same way. An outlet's static pressure loss depends on its geometry and free area, and has to be derived by test.

Most Singapore flats never carry one. On a wall unit the machine and the outlet are the same object, since air leaves through the body's own face. These fittings turn up once that body moves into a ceiling. A ducted system ends every branch at one.

Ceiling fittings fall into a handful of families, and the family settles the shape of the discharge. Round and square cone diffusers stack concentric cones over a round neck. Price describes the square version as holding a uniform 360 degree horizontal pattern even at extremely low flows, which is why it suits a system whose airflow varies.

Louvre face and perforated face diffusers work from a modular core. Price says that core allows a one, two, three or four way pattern to be selected. On a perforated face the aiming is done by deflection vanes at the face or in the neck. The perforations themselves aim nothing.

That distinction matters the moment a household reports a warm corner. The visible face gives no clue which way the fitting throws. A four way and a one way perforated diffuser look identical from the floor, and only the vanes behind the sheet differ.

Plaque diffusers hang a flat disc in front of a round inlet. Price lists three plaque positions that can be set on site, taking one fitting from fully horizontal throw through to vertical. Slot diffusers carry a pattern controller in each slot, set to throw left, right, or straight down.

Radial and twist diffusers put vanes across a circular face to spin the discharge, which Price describes as a twisting pattern for rapid mixing of room air. Industrial ranges run the other way. Drum louvres and nozzles carry air a long way across a large volume, and neither belongs over a bed.

Three properties come off a single selection, and no one of them gets set on its own. How far the air reaches, how much pressure the fitting costs, and how loud it is in the room all move together. Push any one and the other two answer.

One page of one catalogue shows the shape of it. Titus publishes performance for its PAS flush face adjustable supply diffuser, and a 24 inch module with a 6 by 6 inch neck runs across nine tabulated conditions. Airflow along that row climbs from 75 to 350 cfm. Reach to a terminal velocity of 50 fpm, on the four way setting, goes from 4 feet to 15.

What that reach costs is nowhere near proportional. Total pressure along the same row rises from 0.017 to 0.360 inches of water, roughly twenty one times over. Noise starts unlisted, which the catalogue notes means below NC 10, and finishes at NC 45. Under five times the air buys under four times the reach, at about twenty one times the pressure.

Pattern is the one lever that moves reach without disturbing the other two. Hold that neck at 150 cfm and the table gives a single total pressure of 0.066 inches of water and a single noise figure of NC 19. Reach changes anyway. It reads 17 feet on one way, 13 on two way, 11 on three way, and 8 on four way.

Nothing was paid for that difference, and nothing was gained either. A one way pattern spends its whole allowance in one direction. A four way splits the same air and covers a square instead of a line. The choice decides where the air goes, and one fitting cannot do both.

Neck size is the third lever, and it trades the opposite way. That same page puts a 6 by 6 neck at 200 cfm on 0.118 inches of water and NC 28. An 8 by 8 neck passing more air, at 222 cfm, sits on 0.052 and NC 19, reaching around two feet less. More air, under half the pressure, nine points quieter, slightly shorter.

So a fitting sized down to suit a ceiling grid pays for the neatness twice. It takes more of the branch's pressure allowance and it makes more noise, buying reach a bedroom never needed. Where a quotation names a diffuser size before anybody says how far the air must travel, that size came off the ceiling tile.

A diffuser has no single performance. Its catalogue page is a grid, and each number on it carries the condition it was taken at.

Reach gets published against three terminal velocities. Titus tabulates it to 150, 100 and 50 fpm, and states the tests were isothermal, meaning supply air at room temperature. A cooled room is nothing like that, and the catalogue makes no claim otherwise. The tabulated distances also assume jet attachment at a surface. It warns as well that actual performance with a flexible duct inlet may vary in the field.

Noise carries assumptions of its own. Titus derives NC from octave band sound power minus a room absorption of 10 dB, and a dash on the table means a figure below 10. An unlisted reading is a floor, and never a promise about a particular room.

Speed through the neck settles nothing about noise on its own, and the makers say so. Price compares four linear diffuser designs at one 24 inch module, 380 cfm and a 700 fpm neck velocity. NC across the four runs from 31 to 46. Its conclusion is blunt. Selecting outlets by neck velocity is a poor indication of acoustic performance.

The design point is also not the only condition worth checking. Price asks that minimum and maximum air quantities both be considered for reach, and warns that outlets holding a good horizontal pattern at very low flow can still fall short. Titus asks the same, at maximum flow and at the lowest flow expected while a space is occupied.

Published performance describes the article as it left the factory, tested in clear air. A flat can move it out of that condition without anybody touching the machine.

Obstruction is the ordinary version. Price names ceiling layout, walls, partitions and other boundaries as things that may obstruct the pattern and produce unacceptable velocities in the occupied zone. Reach also counts along the wall, not only across the room. Air that reaches a wall with speed left carries on down it.

Patterns can run into each other as well. Price sets a rule for that case. Where two patterns will meet, each outlet should be chosen so its reach equals half the distance between them. A partition built later moves that midpoint, and nobody reselects anything when a wardrobe goes up.

Controllers get left in the wrong position. On the families that adjust, a previous fit out or tenant can leave a slot throwing at a wall, or a plaque dropped to vertical. The ceiling looks finished either way, and no marking on the face says what position it is in.

Paint over the face is worth reporting for a reason the data does not quantify. No maker publishes a penalty for it, because a fitting is tested as it is made. The narrower point still holds. A painted face is no longer the article those figures describe, and on a perforated diffuser the coating sits across the free area they were measured with.

Swapped faces belong on the same list. A fitting changed during a renovation to match a new ceiling was picked against that ceiling, and not against the branch feeding it. It may pass different air, at a different pressure, in a different pattern.