What I Didn't Know About Roofs
A Correct Drawing and a Buildable Drawing Are Two Different Documents
A finished plan set for a Florida sound isolation project, a framer who asked how it was actually going to get built, and a roof we are still redesigning.
About three weeks ago I learned a way of framing a roof that I did not know existed. I have been designing sound isolated rooms for years, and I had never heard of it.
When I understood what it did, I realized it was the single best sound isolation detail available in a project I had already finished designing. The plans were done. We were headed to permit.
So we are changing them.
The strange part is that the technique has almost nothing to do with sound. It comes out of the framing world. A slightly different way to frame a roof turned into a solution to a problem I had quietly accepted as unsolvable.
This is a live project. We have not made the final call yet. I want to walk through what I missed, where I found it, and what it actually does, because this is what design work looks like before it gets tidied up into a case study.
The building
The project is a detached accessory dwelling unit behind a client’s house in Florida. I will call the client Dave. The room inside it has to be quiet enough for a full band to play without the neighborhood knowing about it.
The structure is a box inside a box. The outer shell is eight inch concrete block with the cores filled with sand. Inside that sits a completely separate wood structure that touches the concrete at no point.
The ceiling is where the real work is. There are two of them. A structural attic floor above, and below it, hanging free, an independently framed ceiling that bears only on the interior isolation walls. Nothing crosses between them.
Crammed into the attic above all of that are four baffle boxes roughly eleven inches tall and four and a half feet wide, plus an energy recovery ventilator for fresh air and a dedicated dehumidifier. The attic is the hardest part of this building, and it is the part everything else depends on.
The problem I had accepted as normal
In an assembly like this, the attic floor is the outer mass layer. Two layers of 23/32 inch OSB. Its entire job is to be heavy and to be airtight.
Here is the part I had stopped questioning. In a conventional roof, the rafters land on top of the wall and continue upward. That means the attic floor has to stop and be cut around every single rafter. Sixteen inches on center, all the way around the building.
On a building this size, that is dozens of individual notches. Every one hand fitted. Every one requiring sealant. Every one a place the assembly can leak.
An assembly is only as airtight as its worst detail, and it depends on a framer caring exactly as much at four o’clock on a Friday as he did on Monday morning.
I had drawn that condition many times. I had never questioned it, because that is simply how roofs get built.
Where the fix came from
A different project, a different state. The architect on that job specified a framing approach I did not recognize, so I went and worked out why it was there.
It is a framing technique where you move the sill plate from the top of the exterior wall, in this case our CMU wall, and transfer it up to the top of the attic floor OSB or plywood. This technique doesn’t have a name, but is a different way of framing a roof and it does something really important to our sound isolation.
Why a framing technique is a sound isolation detail
Three things happen when the roof gets built off the attic deck instead of off the wall.
First, and most importantly, the attic floor becomes continuous. No rafters penetrate it anywhere. The assembly goes from being airtight if thirty separate details are executed perfectly, to being airtight because of its shape. Those are two categorically different kinds of reliability, and only one of them survives a real job site.
Second, we gain roughly nine and a quarter inches of working height in the attic. That is the difference between the insulation installers getting trapped in our crowded attic and having some headroom to do their job.
Third, the structural relationship changes. In a conventional roof, the ceiling joists resist the outward thrust of the rafters. When the roof bears on a fully decked attic floor, that floor can take on the job instead. The architect and engineer confirm that for our wind loads, since this is hurricane country, but that principle is what allows the technique to work at all.
None of this was invented for sound, but it just so happens to solve a mass and airtightness problem that drastically improved our design for this ADU.
The sequencing problem that surfaced it
The reason any of this came up is a meeting. We got the general contractor, the project manager, the framer, and the client on a call and walked the plan set page by page.
Partway through, the framer started thinking out loud about the order of operations, and hit a collision.
The baffle boxes and their ductwork have to be in the attic before the deck closes. The spray foam goes on the underside of the roof deck, which happens after the roof closes. Which means a foam crew ends up working overhead in a sealed attic, in a few inches of clearance, moving around four large boxes they did not install.
The proposed solutions on that call were to cut a removable hatch in the finished attic deck, or to tent the entire structure and work under it through Florida rain and fifty mile an hour afternoon gusts.
A drawing shows you a finished building. It never shows you the building half built.
Nothing in that plan set was wrong. Every assembly met spec. The isolation performance would have been exactly what I designed. What was wrong was the order it would have to be built in, and a sequencing problem is invisible on a plan set by definition, because a drawing depicts the end state and not the twelve states the building passes through getting there.
The catch nobody was looking for
The roof change was not the most valuable thing that came out of that meeting. This was.
Once you insulate at the roof and insulate at the ceiling, the attic in between is thermally orphaned. It receives no heating or cooling from the room below. And the dehumidifier living up there has a rated operating range that tops out at ninety five degrees. A sealed, uncooled Florida attic will pass that in summer without much effort.
Nobody went looking for that. It fell out of a conversation about framing sequencing.
The fix is small. A balanced supply and return loop of roughly fifty cubic feet per minute, running continuously off the kitchen side of the building, penetrating only the kitchen ceiling. That side sits outside the isolation boundary, so it costs us nothing acoustically. It goes into the bid now rather than showing up as a change order later.
The value of that meeting was the second thing the roof change exposed. You cannot schedule that. You can only create the conditions for it, which means getting the people who will actually swing the hammers into a room before the permit goes in.
Where it stands
Still open, as of this writing.
- Rafter sizing is with the architect, since the change is structural and the load path is his.
- There is a code path we are chasing that would let us delete the spray foam entirely, and we do not have an answer yet.
- Go or no go is Friday. If we do not have it by then, we draw the original assembly and move on.
That last point matters more than it looks. Knowing when to stop chasing the better answer is part of the job. I am not going to hold a client’s permit timeline hostage to a detail I find interesting.
What I would take from this
I have spent years learning how sound moves through buildings. The best isolation detail in this design came from people who never think about sound at all. They were trying to keep a house from leaking air.
If you only read inside your own field, you will miss the thing that matters most. And if you draw a building without walking through how it actually gets built, in order, by real people, you eventually find out the hard way that a correct drawing and a buildable one are two different documents.
Planning a room that has to be quiet?
The problems worth finding are the ones you find in design, not in construction. A Sound Isolation Site Assessment is where we pressure test the idea before it becomes a set of drawings, and long before it becomes a change order.
We design sound isolated rooms all over North America.
SPYS Designs · soundproofyourstudio.com · Nashville, TN
