Hi everyone,

I have completed adding 2"inches of xps to my basement floor based on the advice of some of the radiant folks on this forum. Originally I was thinking of going with 1/2 OSB sleepers and then running the pex inbetween. On top of that I was going to install 1/4 cement board and then tiling. I am trying to save ceiling height, adding the 2 inches of XPS to prevent heat loss to the slab obviously bit into the space.

So I was thinking it would be great if I could actually groove or melt the XPS and make the channels directly in the XPS and not have to use the OSB sleepers. 1/2" of savings is going to make a difference. Either melt the channels in or maybe run a router. The router option maybe too strong. Also thought of heating up a copper 1/2" set of pipe and then pressing melting in the groove.

Anyways wondering if anyone had any comments or ideas on how I might be able to get this down.

Attached is the a phone of when we started, basement in completed now with XPS.

Not a good idea. You will be insulting around a significant portion of the tube perimeter and your heat transfer will be significantly reduced. Always best to have concrete fully around the entire tube.

Saila thanks for the reply. I should mention this is a retro install, and I am not putting concrete on top. Plan on putting cement board 1/4 inch then tiling.

Using cement board for this floor assembly will result in even worse heat transfer…very little heat will get into the floor…it will just get circulated to/from the boiler… To maximize heat transfer, you want to maximize the amount of tube that is in contact with the floor substrate. If you are locked into this approach of embedding the tube into the insulation, you may want to consider using aluminum plates to get the heat from lower embedded portion of the tube up to the cement board…similar approach to Below-Floor installation.  Perhaps the radiant folks on this forum who initially gave you the advice to go down this path will have some better thoughts.

I reckon you could cut the grooves in similar fashion as is done for ICF...or perhaps heat the aluminum plates with a torch and use them to melt the insulation while placing them into the insulation. The aluminum plates would correctly position the tube and would provide for better heat transfer and better distribution of the upward heat gain.

Using a Below-Floor heat transfer coefficient might allow for some estimation of the design parameters. We have software on our website to try this.

Good luck!

Using a melt/burn approach will degrade the insulating qualities of the XPS in the affected areas- don't do it!

You can use a standard wood working router with a ~3/4" bit (round or straight) to cut channels sufficientlyl wide & deep to accommodate sheet-metal heat spreader plates for half-inch PEX (which are certainly called for here.) Leave at least a couple of inches between the heat spreader plates to give the PEX some displacement room for thermal expansion.

I would say a router also, and formed aluminum plates, or at the very least, heavy foil formed into the channel to help the PEX conduct heat across the top of the insulation.

Posted By Ronmar on 12 Mar 2013 04:14 PM
I would say a router also, and formed aluminum plates, or at the very least, heavy foil formed into the channel to help the PEX conduct heat across the top of the insulation.

Routing that amount of xp would certainly cause a mess and the possibility of a heath hazard. Check out the hot knifes eifs installers use to cut relieves in the foam prior to stucco.

Thanks for the replies everyone very interesting.

Could anyone explain to me why putting the cement board (1/4 inch thick) is a bad thing? If the pex is directly below the cement board and the XPS below is preventing thermal loss to the slab, wouldn't the cement absorb the heat since there is less resistance?

Would 1/2 OSB and then tile be a better option?

you need a conductive path from the pipe to the floor surface. if whatever you add is not in contact to a significant amount of pipe, it's just an additional layer of insulation.

Use plates at least... thermodynamically, you need that tubing to conduction contact. tile on top of that, thermodynamically, with cement board is fine, you basically just did a 'sandwich' install with foam instead of wood sleepers, nothing wrong with that at all. plate to tile will perform significantly better than plate to subfloor to tile. that said i have no idea if you can put tile on top of foam like this without cracking... if a tile guy would be ok with that, then great. I'd be concerned without a solid layer that there may be movement that would compromise the tile install.

I would have suggested Roth Panel in the beginning which is foam and aluminum and tubing grooves all in one, just in case someone stumbles on this later.

I would agree with Rob. The cement board is fine from a conductive point of view, but 1/4" cement board over foam does not give me much confidence in the durability of the floor. Probably tile cracks and at least grout line cracks. A better plan might be to lay extruded mesh over the foam, pipe, aluminum set up, maybe staple through mesh into the aluminum fins and then lay down a layer of thinset. Let this dry and shrink than set the tile into a fresh layer.

Posted By FBBP on 12 Mar 2013 04:54 PM

Posted By Ronmar on 12 Mar 2013 04:14 PM
I would say a router also, and formed aluminum plates, or at the very least, heavy foil formed into the channel to help the PEX conduct heat across the top of the insulation.

Routing that amount of xp would certainly cause a mess and the possibility of a heath hazard. Check out the hot knifes eifs installers use to cut relieves in the foam prior to stucco.

More of a health hazard than the inhaled HFCs & styrene vapor & polystyrene aerosols from a hot-knife approach?

Unlike the EPS used in EIFS, XPS is loaded with HFC134a, but vaporized polymer/smoke/aerosol issues would be pretty much the same with either.

With a router approach a dust mask would be sufficient for avoiding inhaling XPS crumbs, most of which would be macroscopic, not a persistent aerosol form.  There would be some HFC & airborne styrene release, but not as much as in a melt-out approach IMHO.

Posted By Dana1 on 13 Mar 2013 02:22 PM

Posted By FBBP on 12 Mar 2013 04:54 PM

Posted By Ronmar on 12 Mar 2013 04:14 PM
I would say a router also, and formed aluminum plates, or at the very least, heavy foil formed into the channel to help the PEX conduct heat across the top of the insulation.

Routing that amount of xp would certainly cause a mess and the possibility of a heath hazard. Check out the hot knifes eifs installers use to cut relieves in the foam prior to stucco.

More of a health hazard than the inhaled HFCs & styrene vapor & polystyrene aerosols from a hot-knife approach?

Unlike the EPS used in EIFS, XPS is loaded with HFC134a, but vaporized polymer/smoke/aerosol issues would be pretty much the same with either.

With a router approach a dust mask would be sufficient for avoiding inhaling XPS crumbs, most of which would be macroscopic, not a persistent aerosol form.  There would be some HFC & airborne styrene release, but not as much as in a melt-out approach IMHO.

Dana - thanks for reminding us of the chemicals, wasn't really thinking of them. I've routed Celfort before and had to wait for a warm day when we could open both doors to the shop and then used a leave blower to clean it out of all the nooks and crevices. We finally went with an enclosed router and tied a light vacuum line to the dust shoot. Kept a 3" line from the dust collector in the vicinity and that made it almost bearable. Maybe a dust mask will be enough in a basement but if he can get something more then just a shop vac on the job, it will save him a lot of grieve.

Thanks for the feedback. I see why aluminum heat spreaders make sense and the need for extra contact.

Good insight on the tile. I can't seem to find any reference to being able to tile on top of xps + cement board. Eventhough the fomular f-250 is rated for 25 PSI, I think the deflection could be a problem.

What if I were to run 2x4 (shaved down to 2" inches to match the XPS height) in between the XPS sheets. In essence creating bays at 24" inch on center. The 2x4's would have no deflection since they would be in contact with the basement floor directly. Create the channels for the pex and add aluminum heat spreaders and then I could add a layer of 1/2 OSB and then a layer of Cement board and then tile.

Does anyone think this approach would take care of the deflection issue with the tile and would the osb and cement board with spreaders still be able to radiate the heat up?

Well, 0.5" cement board is about R0.05, 0.25" ceramic tile is about R0.15, and 0.5" OSB is about R0.8, for a total R-value of about 1. You are still below R2 which is about where the required supply temp begins to get excessively high to generate the amount of radiant heat gain to replace the typical heat loss. So YES, it is doable and the heat will still radiate up. However, without knowing your heat loss, can't say with absolute certainty what radiates up will be enough. This also seems like a lot work and expense just to have radiant heat via a unconventional and perhaps questionable floor assembly. It pains me to say this, but sometimes it makes more sense to just go with electric baseboard heat.

Posted By cloves on 14 Mar 2013 10:53 PM
Thanks for the feedback. I see why aluminum heat spreaders make sense and the need for extra contact.

Good insight on the tile. I can't seem to find any reference to being able to tile on top of xps + cement board. Eventhough the fomular f-250 is rated for 25 PSI, I think the deflection could be a problem.

What if I were to run 2x4 (shaved down to 2" inches to match the XPS height) in between the XPS sheets. In essence creating bays at 24" inch on center. The 2x4's would have no deflection since they would be in contact with the basement floor directly. Create the channels for the pex and add aluminum heat spreaders and then I could add a layer of 1/2 OSB and then a layer of Cement board and then tile.

Does anyone think this approach would take care of the deflection issue with the tile and would the osb and cement board with spreaders still be able to radiate the heat up?

But the foam is also be in direct contact with the basement floor, so without full subfloor thickness you'd have the same deflection issues or WORSE, since the sleepers would become less compressive fulcrum points and lines of weak shear strength rather than allowing the microscopic flex to be distributed over a broader area.

Absolutely want to make sure the floor's deflection isn't a problem for possible tile install.

I appreciate everyone's input as usual. Attached is another idea I was working on.

- Tile (Porcelain Glazed or Ceramic)

- 3/4 Plywood

- 2x4 Sideways (1 1/2 x 3 1/12 nailing surface)(1/2 shim to bring it up to 2" to match XPS) Fastening into concrete using concrete screws)

- XPS 2' inbetween each bay (24" inches wide)

- Pex and aluminum transfer plates routed into XPS

- 6 ml poly on concrete Pex will be close to 12" OC and will vary a hair based on where the sleeper/frame falls. Tried to roughly show in bottom of drawing. Could this setup solve the deflection issue?

This floor assembly will likely have a significant amount heat loss through the 2x4. You should really work this out to determine the “whole” floor assembly effective R-value (i.e., combination of foam cavity and non-insulated 2x4s). The 4” slab will likely be at approximately 50 deg F and the heated tile surface will likely be approximately 80 deg F. So the heat transfer equation delta T will likely be approximately 30 deg F. These actual temp values will of course depend on your location, time of the year, and associated heat loss your building. A good hydronic radiant floor heating system designer will normally keep the downward floor heat loss to less than 10% of the required upward floor heat gain. I suspect that this floor assembly will not come close to meeting this design requirement. So you will have to decide if the resulting lower heating system efficiency and the resulting increased operational cost is acceptable to you.  Again, a little bit of analysis and planning goes a long ways in avoiding these sort of issues.

With 3/4" standard subflooring resting on foam it has LESS deflection than the same standard subfloor suspended on joists, as it done all the time. (Air between joists has a heluva lot more give than 25psi or even 15 psi foam resting on concrete, eh?)

Sleepers add little to the rigidity of the floor here, but they DO add a lot of thermal bridging and heat loss, as sailawayrb aptly points out. Adding the less-compressible sleepers provides a fulcrum for concentrating the pressure at the center of the foam between the sleepers rather than distributing it more widely (at lower overall depth) on the foam layer. It'll flex less if you go with a field of foam, no sleepers.

Absolutely want to make sure the floor's deflection isn't a problem for possible tile install.

Once you've got the foam down, another thing you might want to consider is Bekotec by Schluter. You just lay the Bekotec foam panels which have a knobby arrangement to hold the PEX tubing and then you screed mortar over the top. You will get your insulation, your heat transfer medium and a perfect surface to lay tile on without worrying about support or cracking. You will add a total height of 1-11/16" above the foam you have down now.

My guess is that 1/2" cement board + plates over foam (ie, no OSB) would work well with normal floor loading. But do a small test first.