I am renovating a 60's bank style ranch with a walk-out basement in NC and am planning to put in hydro radiant heating. Since I had to demo a part of the slab to relocate plumbing, I know that the layers are: clay soil, 4" gravel, ~6 mil poly, 4" concrete. On top of this slab, I was planning to put down 3/4" unfaced XPS, 3/4" decking sleepers on top of XPS, run PEX between the sleepers, fill with dry sand, 1/2" plywood, then either carpet or engineered hardwood. Many DIY sites, however, call for a layer of poly on top of the slab (below the XPS?). Wouldn't this cause moisture to be trapped inside the slab that is sandwiched between two layers of poly? I have 3" of space to work with, so any recommendations on the best and safest way to do this would be helpful. If it helps, I am familiar with Building Sciences practices but did not find any info relating to retro installation of radiant floor heating. My below-grade walls are sealed tight with caulking, 1" unfaced XPS, 2x4 build-out, and unfaced fiberglass. The above-grade walls on the walkout side are 2x4 wood framing filled with closed cell spray foam. Thanks in advance.

I you have 6-mil poly under the slab, there's no point to an additional vapor barrier. The 3/4" XPS will also be fairly vapor-retardent, but would allow the concrete to dry toward the room. Concrete is pretty tolerant of moisture in the first place- even if it's blow the dew point of the room air all year long, so what? What the above-the-slab poly might give you is a reasonable slip-surface between the XPS and concrete, but I haven't read of problems occurring without a slip-material for XPS the way there can be with gypcrete radiant slabs poured atop wood subflooring.

Mind you, using only 3/4" of XPS gives you a scant ~R4 of insulation between the tubing and the subsoil in that stackup, so using the floor as the radiation will be pretty lossy. Rather than 3/4" of sand, add another 3/4" of XPS between the tubing and use aluminum heat-spreaders for the PEX under the sub-floor WarmBoard(tm) style (or even using Warmboard or similar product as the subfloor to hardwood rather than using half-inch ply) to get more R and better heat-extraction from the PEX & into the floor for a lower operating temp. Sand would only insulate the PEX slightly from the sub-floor, raising the water temp requirements, increasing losses to the ground. Most slab-radiant installations call for a minimum of R10 under the slab, and use embedded PEX in the concrete for the lowest-possible water temp requirements.

For the finish flooring, avoid carpet, since carpet insulates the radiation from the room, (more than doubling the R-value from tubing/heat-spreader to the conditioned space over what it would be with half-inch poly + hardwood) which also raises the water temp requirements (and losses.)

You don't really need the 3/4" sleepers- long-screwing the plywood sub-flooring plywood to the slab on a grid with a hammer-drill/roto-hammer and masonry screws should be sufficient- just be sure to set up your grids carefully so as not hit the PEX with the drill, and be very careful about fastening methods & fastener lengths on the finish flooring.

With at least 1.5" of XPS total between the heat-spreaders and slab you'll have nearly doubled the total R value to the subsoil for all but the narrow section by the tubing channels. But the water temp requirements will likely be at least 10F lower as well. The combined aspects of higher R and lower water temp will cut the average heat loss to the ground by half or more.

You may want to get more & better suggestions on how to do the radiant & insulation by posting a question on the radiant heating portion of this site. (Some real radiant-heating pros monitor and post to that forum regularly.)

Building Science Corp concentrates on the building envelope design, not the mechanical systems within. They trend toward the PassiveHouse school of thought, that spending the money on insulation & air tigthness is a better investment than highest-efficiency systems. When you insulate to where the peak heat loads are low enough, there's no "cushy-factor" to radiant heat since the floor temp is about room temp, and the efficiency of the heating system hardly matters. In a retrofit it it's much more expensive to super-insulate though, and radiiant flooring allows you to run the boilers well into the condensing range of temps, yielding 95% efficiency or higher when done right.

Thank you for the helpful response. I will post to the radiant heating subforum for additional help. Could you explain what you mean by a "slip surface" between the slab and XPS or gypcrete? Is it to decouple the two masses so that they can contract and expand independently of each other with temperature change?

Yes- slip surfaces are necessary between mediums that a subject to expansion/contraction due to changing temp & humidity. Wood will change dimensions seasonally with humidity, gypcrete/concrete has very little flexibilty, and will expand/contract when used for heating, and the combined forces can fratz the slab a bit. By contrast, XPS is fairly flexible, and can accomodate the mechanical strains of the temperature changes without undue damage. The concrete will stay at about the same temp throughout the year, but the XPS will be several 10s of degrees warmer on the warm side than on the concrete side during the heating season.