Subfloor question for tiling a bathroom floor with heat mat under the tiles

My bathroom's subfloor is an old tongue and groove sub-floor 7/8" thick. It seems to be in "OK" condition but I am little bit concerned of my plan.

Here are the layers:
1. Current tongue and groove 7/8" thick subfloor
2. A layer of modified thinset cement based mortar (using 1/4" notch trowel)
3. 1/4" Wedi building panel
4. SunTouch heat mat
5. at least 1/4" modified thinset cement based mortar (or maybe SLC) to cover the mats (let it dry)
6. 1/4" modified thinset cement based mortar
7. 3/8" ceramic tiles 12"x24"

Do you think I should strengthen the subfloor? The joists are spaced 16 OC.

Please advice,

Thank you.

Big hiccup...none of the tile substrates are rated for application directly over dimmensional lumber - you need at least a 1/2" of ply, C-faces or better (i.e., no D side) installed across the joists, and exposure I or exterior glue holding the ply together. Then, lots of screws to hold it down (screw any loose planks down before you start). Then, and only then, IF the joists are sufficient, can you tile things. You can use the Wedi, but are you sure it is rated for having heat applied over it? The amount of insulation 1/4" would provide is marginal, and probably not worth the extra cost. You have a bunch of choices on decoupling, you could install cbu-mat-tile, or tack the mats to the ply, put slc over it, then a decoupling membrane like Ditra, and then the tile. You can also check out www.johnbridge.com for some ideas on tiling. SLC over ply generally needs to be at least 1/2" thick over the highest point. If you install cbu first, you can put a thinner layer of slc if you wish to go that way. It's MUCH easier for a first-timer to get a good slc pour if it is thicker rather than thinner. Just like your pancake batter doesn't flow to fill the pan keeping a certain thickness, slc does too. Only if you get enough down, break the surface tension, does it actually flow, and then, only if it is thick enough. But, for a novice, getting a nice flat floor by trying to spread thinset out over mats without nicking wires is not the easiest thing, either.

Thanks for your advice. According to Wedi spokeperson, a heat mat can be applied on top of the Wedi board.

Is the Green Glue noiseproofing compound good enough to glue the ply to the subfloor?

You do NOT want to glue ply to your dimensional lumber! Or, generally, to a layer of ply underneath. You need 100% support between layers, and a thick glue you squeeze out of a tube will NOT give you a nice, consistent, complete layer - you'll end up with some voids, and tile hates voids. With stuff like cbu, the idea is decoupling, and glue just doesn't work. If you were trying to obtain maximum stiffness from ply, you can laminate (not simply glue) two layers of ply together if you use a full spread wood glue like Titebond II (it's easier and cheaper to just use thicker ply - exception, stone requires a two layer subfloor). You need a LOT of it, gallons on any sized room), a spreader to get a good coat everywhere, and speed so you can lay the sheet of ply into the wet glue before it starts to skin over, then the proper method of sinking the hundreds of screws needed. Do it wrong, and you'll end up with a much worse structure for tile than you start with. Generally, it's not for the feint of heart. The industry standards just call for screws. Plus, if you glue layers of ply together, you'll destroy the structure if you ever do decide to remodel!

If you go with Wedi, you MUST follow their instructions, which call for it to be set in a bed of thinset, sort of like a huge tile. If you use cbu, the same thing, except you also need to add a bunch of screws. In the first case with Wedi, the thinset actually holds it, but for the cbu, the purpose of the thinset is to actually fill in any minor imperfections and provide 100% support - the screws hold it in place. You'd likely use a different thinset to set the Wedi to the ply verses cbu to the ply, as you could use a dryset thinset for the cbu, but would likely need a modified thinset to embed the Wedi to plywood.

You need to read the installation instructions for the products you are contemplating. Another good item is the TCNA handbook (Tile Council of North America). This handbook describes the tested, known to work methods of doing nearly anything tiled. It is the basis of many of the building codes. There is often more than one way to do something right, but many more ways to mess up. It's good you're doing some research, but verify your resources, too.

The whole reason for ply over dimensional lumber is that that base moves too much. You don't want to glue to it. If you want to laminate sheets of ply together, you can and it is stronger, but if not done right, you'll have a mess. If you hit a void when screwing the ply down, move the screw! The TCNA handbook, from my last look, does not advocate gluing anything to planks. They do that for a reason. IF you do glue anything, other than say a construction adhesive on the top of the joists, you want a full-spread glue. This is almost impossible with anything you squeeze out of a tube. IF the layer you are putting down is stiff enough to bridge any gaps or voids, you may not notice, but the object is to get intimate contact, and a thick glue you squeeze out of a tube will NOT give you 100%, even coverage with intimate contact.

They do not advocate installation of Ditra over sawn lumber unless you add at least a nominal 1/2" of ply, just like I suggested. See page 9 http://www.schluter.com/media/DitraHandbook.pdf.

I just calculated the deflection and I got L/318. Looks like not enough for a ceramic tiles which I want to install.

I am attaching a couple of sketches of the bathroom floor right above the basement. The sketch is not in scale but the dimensions that are shown in text are right dimensions. Could you take a look at it and let me know if there is a way to have the tiles installed even though the deflection is below the recommendation?
The area in red will have tiles installed. The entrance to the bathroom (not shown) is between the cabinet and the tub.

Thank you,
Andy

Thanks John. I already have blocking in the center of joists.
Also, on John Bridge forum somebody mentioned that the measurement of the joists should be taken at the unsupported joists only, is that true? If it is, then in my case the joists would be about 14 '25". According to deflecto calculator (using 14' joists) I am getting a deflection of L / 380 (good for ceramic tiles).

If it is not true, my second option would be to add sister joists. In this case I could easily add at least half of additional sister joists.

And lastly (third option), I may add a supporting wall in the basement (basement is unfinished). There is an area right under the current bathroom that I plan to convert to a basement bathroom (there are already installed drainage / vent pipes).

johnfrwhipple said:

Andy (sic) blocking could help you with your final deflection numbers. I would consult a local structural engineer and ask for a new blocking layout. Perhaps you may need to double up all the joists.

JW

Click to expand...

All blocking does is transfer load to an adjacent joist and keep them from twisting, which maintains whatever vertical strength they have; blocking does NOT decrease the deflection - only shortening or strengthening the joist vertically can do that. Preventing the joists from being able to twist maintains their strength, it does NOT increase it, nor does blocking (except by helping to maintain what strength they have). Keep in mind that the subflooring provides the same general load sharing, and they do not rely on increased strength subflooring to account for joist deflection issues - it is required to provide strength BETWEEN the joists, not along them (especially since the max strength is directional on most subflooring, it is quite a bit weaker along the joist than between them).

Re glue between ply and planks...the whole reason you want the ply there in the first place is to decouple the tile from the seasonal movement of the planks...the planks are likely strong enough in deflection between the joists in the first place. Each stage of decoupling progresses to the maximum stability required for the tile. Gluing the ply to the planks just helps ensure that that movement is transferred through the floor, negating most of the purpose of the ply in the first place. They omit saying to glue on purpose. Same idea with the cbu...testing proves that over time, the screw anchor ends up crushing the cbu since it is tied to the less stable subflooring, but the cbu, if installed properly with the thinset and tape, remains monolithic (albeit with some microscopic crushed holes around the anchors). The thinset underneath cbu is not there to hold it down, it is there to hold it up with 100% support (i.e., vertical deflection). Glue is good between the subflooring and the joists. It is okay between like materials (i.e., ply), but takes careful application with the right materials and technique. All of the TCNA subflooring prep is based on no glue, but, as said, gluing the subfloor to the joists is a good practice - the TCNA generally takes over from a properly installed general application subfloor and converts it to one that can support tile. Laminating ply to ply is a lot of work for not that much improvement (yes, a laminated set of ply IS stronger than those just screwed together), but if you do it wrong, it's worse. And, it makes it a major effort on a remodel down the road as you'd likely have to cut and chop the whole shebang out and start over, increasing costs by a lot. If you had two idential floors, one with blocking and one without and put identical weights in the middle of the span, they would both deflect the same amount assuming that the joists were held properly to the subflooring. The blocking helps to keep them vertical and stable.

Re how to measure for joist deflection - yes, you must measure the joists between structural support members UNDERNEATH the floor (i.e., what is holding it up). Those structures could be the rim joist, a beam, or a structural wall (note, not all walls are built to be structural!). Take a worst cast, a very small room in the middle of the joists. Say the joists can deflect 1/2" under a 'normal' load. Take that same floor and shorten the total length, even though the only area that is tiled is still in the middle, and the deflection may drop to 1/8"...much more tile friendly. Tile is very brittle, nearly any movement will either break the bond, or break the tile, or break the grout, whichever is the weakest. A joist tends to bow - point deflections don't generally happen, so you need to take into consideration the whole length, regardless of where the tile is. Now, if you're only tiling around the perimeter, say right next to the support, you might get away with it since at the supported end it doesn't move much, but that is an atypical situation.

CX the guy from the John Bridge forum who uses "eh?" in awkward sentences all the time...

None of the structural tables indicate blocking helps in the least. As I said, blocking helps to MAINTAIN whatever strength the floor has, so in that manner, yes, blocking can make a floor stiffer (than if the joists twisted upon deflection rather than staying nice and straight). It also really helps keep things in line prior to installation of the subfloor. If you want to make your floor stronger, use more joists, or deeper ones, or stronger ones, or shorten their span. So, as I said, blocking will maintain the strength you had, but does not increase it. Gluing your subfloor To the joists does a lot to prevent the joists from twisting, blocking also helps a lot prior to subflooring installation. The subfloor spreads the load between joists just the way blocking does, and they do not include the subfloor in the overall joist deflection calculation. Next time you talk to a structural engineer, ask him if the blocking actually decreases deflection or just maintains the maximum available from the joist by preventing twisting. SO, in a backwards way of looking at it, it helps, but not for the reason and effect you are talking about. A floor, blocked or not, with a properly supported subfloor (that anchors the joists well) will have the same deflection with the same load. Sheathing the bottom of the joists helps too by holding the joists in nice alignment from the other side, but doesn't increase the overall strength, it just maintains it.

If the TCNA analysis and testing found gluing ply to planks helped, they'd include it in their guidebook as an approved method. Their goal is to provide methods to make a tiled installation reliable, includes more than one method to that effect, and does NOT include gluing for a reason. They insist on ply over planks not for increased strength, but to decouple from the movement the planks have with changes in moisture content during seasonal change. You want to minimize that, and gluing hurts that. ANother reason why you don't want to screw the top layer TO the joists, those being the stronger dimensional component and more able to transfer that movement to the top layer. You want each layer decoupled, and gluing prevents that. Laminating same materials is not the same thing.

And, you think building a missile system doesn't require structural analysis?

Blocking can change the resonant frequency of the floor. The effect of blocking helps immensely prior to subfloor installation - the joists are easily displaced sideways until the subfloor is installed. The joists are not perfect, some will be cupped, and the blocking will not be perfect in length nor anchored perfectly. THen, after several seasons where it swells and shrinks, the blocking will not be as tight since it is easy to exceed the resilience point of the wood. A tiled floor puts a distributed load on the structure, it's not a point load, so sharing with an adjacent joist has no effect on the deflection of the whole floor when every joist is getting loaded...this is where you're missing the point. And, because a tiled floor is more monolithic, any point load tends to act as a spread load, deflecting the whole thing. The way to decrease deflection from the tiled floor's distributed (significant) load is to shorten the span, decrease the spacing, widen the joist, or deepen it. Blocking will NOT decrease deflection on a distributed load, but will on a point load (somewhat).

Look at blocking this way: take a rope, hand a 50# weight on it, it will stretch or deflect a certain amount. Now, take a couple of ropes hung say 16" away and tie them also to the weight...it won't stretch as much. But, now add another 50# weight for each rope you add...they'll all deflect or stretch as much as the single rope with the weight on it. Tile puts a significant weight on the entire floor, just like hanging weights on each individual rope, regardless of whether they are tied together...bottom line, if you need a stronger floor, blocking won't do it. Thicker, deeper, shorter, or closer together will give you decreased deflection.

Blocking DOES change the resonant frequency of a floor...it does not change the deflection rating of the overall joists. It DOES help point loads, but does not change the effect of a total load on the floor and the amount it deflects. Place a piano on a floor with and without blocking, and it will deflect the same. Jump up and down on the floor and it will be sturdier (vibrate less) with blocking. The two are related, but not the same. Blocking is sort of like fingering a string instrument...shorten the length, the pitch changes, but it does NOT make the string stronger - it will break at the same point regardless. Not exactly the same, but a joist has sections in compression and tension...that effect occurs along the ENTIRE UNSUPPORTED LENGTH, blocking doesn't change that, nor does it change the ultimate amount it deflects from the applied load. Blocking does prevent the joists from twisting under load, so in that manner (as I've said before), it MAINTAINS the strength the joist has. The subflooring does that to a great degree, and if sheathed on the other side, that helps too. It really helps make things stable before the subflooring is installed.

The joists themselves must be strong enough to meet at least the L/360 rating, blocking won't increase it. It will change the resonant frequency. Studies have shown that a load on an inadequate floor will sag, and continue to bow to a greater extent as the structure ages. It can take a decade, but eventually, that floor or beam or whatever will take a set from the load. Take a look at the ridge of houses, it is rare to see a new one that sags. Look at some that are 10, 20,...50 years old, some are nice and straight, but some sag radically from one end to the other - the result of using too small of a beam. Crushing the fibers to jam in a long piece of blocking doesn't help in the long run, either. Crushed fibers aren't as strong as uncrushed fibers, so you've effectively created a weak point in the joist. Now, a gluelam is a different thing...crushed, but with glue to reinforce and hold the things together.

Hi guys thanks for the info, new to this board and had a question. Doing a main floor in 30" by 20" tiles in an older home from the 70's I believe. Joists then ship lap subfloor and then 1" ext plywood glued and screwed on top of that. Is this acceptable to tile on top of? Or should I be adding another underlayment such as ditra? What about thinner membranes such as easy at or mapegard?

Thanks

Tile wants no plywood with a 'D' face, and exterior sheathing may have that. Do you know? a 'D' face means voids, and those are not acceptable underneath tile. Gluing ply to dimensional lumber is also iffy - the goal is to provide a stable surface, and the dimensional lumber moves much more than plywood - gluing it will improve between joist deflection (not really an issue with that thick of material), but you really want some decoupling. With tile that big, you want a VERY flat floor or you'll be having problems with lippage. Also, lay some down on a flat surface and see if they have any bow or twist in them. Hopefully, they are quite flat. Your subfloor is stiff enough, if you don't have a 'D' face, but the joist structure is equally as important, if not more so. If the joists are okay, I'd consider using something like Ditra. You may want to look into something like TLS or QEP leveling system - these are clips you install during the setting of the tile that hold the edges in vertical alignment. After the thinset cures (well, not completely, but at the earliest the next day), you break off the vertical portion and can then grout it. They are designed to break off near the bottom of the tile, so do not show once grouted. Larger tile are much harder to get to lay perfectly level with each other since any small difference gets magnified over the long edge of the tile. Thus, the leveling system will speed things up and almost guarantee better results.

Thanks for the info, I am just doing the setting the contractor has done the 1"plywood sub. The tiles are quite flat I have checked them and going with a "stacked pattern" compared to a brick or offset so should be ok. So you do recommend going with another underpayment? Would you recommend setting on top of the 1" ply with a high performance mortar?

Tiling directly to plywood requires a second layer. You're better off using something like Ditra on what you have. If the existing stuff is something like C-D grade, I'd put at least 3/8" additional layer of ply on top of it before I tiled, something like B-C or A-C with an exterior or exposure I glue. Do NOT glue it to the existing layer, but do screw it down, oriented across the joists, not along them. You could probably get by with what you have with that large of a tile unless a corner happened to be right over a void, and you happened to put a big point load there. Not likely, but possible, that's why they specify no 'D' faces, as everything is either solid or plugged to fill in any voids on anything above a 'D' face.