The Patch That Showed in Raking Light
Why this matters
A drywall patch that looks perfect at noon and shows at six in the evening is not a mystery and it is not bad luck. It is a patch that was checked in the wrong light. This case is worth walking through because the tech did nothing obviously wrong, the patch was sound, and the first two explanations anyone reached for were both wrong. The real causes were two small things that only exist at a low angle, and once the shop understood them it changed one habit and stopped getting this call.
The call
An HVAC technician replaced a failed duct takeoff above a second-floor hallway. Access was through the ceiling: a 12 in by 22 in opening cut between joists, near the hallway's end window, which faces west. After running the system and feeling around the new takeoff's seams for leakage with the opening still open, he closed it the same visit with a screwed-in patch piece over backing, taped it, and came back two days later for the finish coats and paint. He used the flat white he carries on the truck. At noon on the second visit he stood under it, looked up, and could not find it. Neither could the customer.
At about 6:30 that evening the customer called. The sun was coming straight down the hallway through the end window and there was a rectangle on the ceiling, outlined by a faint raised ridge, with the inside of it looking slightly shinier than the rest. The estimate had said: "ceiling patched and painted, grade blended: not noticeable in normal room light; may show in low light from the hallway window." The customer's point was fair. This was not "may show." It was a box drawn on the ceiling.
First suspect: the patch moved
A patch that shows a day or two later often means it moved: a seam cracking because the backing flexed, a screw head popping, the tape lifting at an edge. That would be a structural failure, and it would mean opening the patch.
The tech went back the next afternoon and checked it by hand and eye before the sun came round. He pressed along every edge of the patch with a flat palm: nothing gave. He looked for a hairline crack along the tape line with a light held at an angle: none. He tapped the field of the patch: solid, the same sound as the ceiling beside it. No screw heads showed. The patch had not moved. Eliminated.
Second suspect: something coming through the paint
The other common late-appearing mark is a stain bleeding through: water from the attic side, or the brown of raw paper facing bleeding through paint that was applied straight over compound without a primer. That would show as a color difference, not a ridge, and would be visible in any light.
It was not visible in any light. Under the hallway's ceiling fixture and in the diffuse afternoon light the ceiling read as one even white. The outline only appeared when the light came in nearly flat. A stain does not depend on the angle of the light. Eliminated.
The work light test
That left the surface itself, and the way to look at a surface is to make raking light on purpose. From a stepladder set fully open and level in the hallway, with the customer kept out from under it, the tech held an LED work light flat against the ceiling at one end of the patch, aiming across it, so the light grazed the surface the way the evening sun did. Within a few seconds he could see the whole thing, and it was two separate problems, not one.
A ridge at the edge of the compound. The tape sat on the ceiling, and the compound over it had been feathered (tapered out to nothing) only about 2.5 in past the tape on each side. On a ceiling that is not perfectly flat to begin with, a short feather leaves a small step where the compound ends. Running a fingertip across the edge found it: barely there, but a definite rise. At noon nothing grazes a ceiling; it is lit from below by light bouncing off the floor and walls, and a step that small casts no shadow worth seeing. With the sun nearly level with the ceiling it casts a long one, which is the whole of the raking-light effect, owned by Why a Repair Shows in Raking Light. As an illustration of the scale: a rise of 1/32 in lit from 5 degrees above the surface throws a shadow about 0.36 in long (1/32 in divided by tan 5 degrees, about 0.0875), while light from 60 degrees above throws one about 0.018 in long (1/32 in divided by about 1.73), roughly a twentieth of the length. Same ridge, twenty times the shadow.
A sheen difference inside the ridge. The inside of the rectangle looked shinier because it was. The truck paint was a flat wall paint. The ceiling had been painted with a ceiling paint, a flat formulated for an even lower sheen. At a normal viewing angle two flats look identical. At a grazing angle, low-gloss finishes separate: this is why paint makers rate the sheen of flat and matte paints at a very shallow measuring angle, because that is where they differ. The wall flat on the patch was reflecting more of the low sun back down the hallway than the ceiling flat around it. And the patch was painted on the patch only, so the edge of the new paint ran exactly along the ridge, putting the sheen boundary and the shadow in the same place. See related: Why Paint Touch-Ups Flash.
Either problem alone might have survived as "may show in low light." Together they drew a frame.
SECTION ACROSS ONE EDGE OF THE PATCH (not to scale)
AS LEFT
|<-tape->|<-- 2.5 in -->|
patch ==[########]=============^========== ceiling
|
step where the compound stops;
wall-flat paint stops here too
AS FIXED
|<-tape->|<---------- 12 in ---------->|
patch ==[########]============================= ceiling
compound tapered to nothing 12 in
past the tape; ceiling paint wall
to wall across the whole hallway
Why noon was the wrong test
The tech's check was honest and it was the check most people do: stand under it in daylight and look. The problem is that the estimate's own test named the condition that mattered, low light from the hallway window, and the check did not reproduce it. A blended grade is defined in normal room light, but it carries a warning for raking light, and a patch that will meet a low-angle sun every clear evening has to be finished for that light even if the customer accepted that it "may show." "May show" means a faint irregularity you can find if you look for it. It does not mean a ridge and a sheen boundary that outline the repair.
The fix
The repair went in two visits:
- Sand the paint off the patch area and flatten the step. Sanded back with a vacuum-attached pole sander, eye protection on and the hallway closed off, until the ridge could not be felt. The house was built in 2004, so the old paint and compound were not a lead or asbestos concern; on an older house, The Legacy Material Check Before You Disturb a Finish comes before the sander does.
- Feather much wider. Two coats of compound with a 12 in knife, then a third with a wider trowel, taking the taper out about 12 in past the tape on every side instead of 2.5 in. A wider feather spreads any remaining height difference over a longer distance, so the slope is too gentle to throw a shadow.
- Check with the work light flat to the ceiling after each coat. From all four sides, before the compound was sanded and again after. Anything that cast a visible shadow line got another pass.
- Prime the patch. A drywall primer so the compound and the old paint film take the finish coat the same way.
- Paint the whole hallway ceiling with ceiling paint, wall to wall, which puts the only paint edges at the wall lines, a natural edge where a slight difference reads as intended.
The rework took about 3.5 hours over the two visits against roughly 1.0 hour for the original finish coats and paint. Because the estimate had sold blended and the result was not blended in any honest sense, the shop did it at its own cost. At 6:45 on the evening of the second visit, with the sun down the hallway, the tech and the customer looked together: the patch was findable only by someone who knew where to look, which is what the estimate had promised.
Where the same symptom would have meant something else
The outline was a finishing problem here because both suspects that would have made it a structural or moisture problem were ruled out on evidence first. Three facts about this ceiling decided how the fix went, and each one changes it:
- A textured ceiling. Orange peel or knockdown hides a small step far better than a smooth ceiling does, because the texture already throws its own shadows. The ridge might never have shown. The sheen boundary would have, and the fix would have been matching the texture as well as the paint, which is handwork with a lower ceiling on the grade.
- A leftover can of the ceiling's own paint. With the original ceiling paint on hand, the sheen half of the problem disappears, and a wider feather plus a repaint of the patch area might have been enough without taking the paint to the walls. Even so, the original paint has aged on the ceiling, so a patch-only touch-up can still read slightly different.
- A ceiling that moves. If the palm test had found the patch giving, or a hairline crack along the tape, no amount of feathering would have held. The fix would have started with more backing and screws, and the finish work would only come after the patch was rigid.
What the shop changed
The work light check is now part of closing every patch. Before the paint goes on and again after, the tech holds a work light flat to the surface from at least two directions and looks for shadow lines and sheen edges. It takes under two minutes. It reproduces the worst light the room will ever get, at any time of day, which is the one thing standing under the patch at noon cannot do.
Paint is matched to the surface it is going on. The truck now carries a ceiling flat as well as a wall flat, and the tech asks or checks which was used before painting. Leftover paint from the customer, with its label, beats either.
A patch in a raking-light location is priced to the natural edges. If the estimate names a raking-light source and sells blended, the paint line goes to the walls, not to the edge of the patch. That turned an argument about whether the patch "may show" into a line item the customer could choose. See related: How to Touch Up Paint So the Patch Does Not Flash.
References
- See related: Why a Repair Shows in Raking Light
- See related: Why Paint Touch-Ups Flash
- See related: How to Patch Drywall After an Access Cut
- See related: Invisible, Blended or Acceptable: The Three Grades of Finish Repair
- See related: The Legacy Material Check Before You Disturb a Finish