We would like to have several "tubular skylights" in the construction we are planning. These are skylights consisting of a circular opening in the roof with attached dome-shaped or circular window, attached to a cylindrical duct with a reflective material on the inside, which eventually reaches the interior through the ceiling, terminating at some kind of window. These skylights not only deliver the light from outside, but they also "disperse" the light (using the tube's inner material) so that upon exit at the ceiling the light is spread out over a larger area and not as concentrated as it would be otherwise.

As far as I can tell, 10" diameter is the smallest size offered among today's tubular skylight products.

We are going to be using a concrete roof technology that has a fixed spacing between its concrete beams. I currently understand it to be the case that 10" would not pose structural problems in the roof, and it may not necessarily interrupt one of the beams, but we would prefer to opt for more numerous openings of a smaller size given that strength and storm resistance is a priority for the building.

I suppose that the "volume of light" that is delivered, both physically and with regards to our perception of it, diminishes exponentially as the diameter of the skylight decreases. This may be one justification for the minimum of 10" among the available products. Aside from that, it is not clear to me whether there would be a big issue with hitting some kind of other physical diameter threshold, smaller than which we would be severely limiting the function (light delivered and/or proper dispersion of the light) of the skylight.

There are two particularities to our situation that make me think a smaller diameter would be OK. (1) There will be no bends in our skylight tubing, whereas generally the commercially available skylights are designed to accommodate bends in the tubing (and the eventual decrease in delivered light); (2) Our tube will be short, maybe 4 feet from exterior to interior, whereas existing products are designed to accommodate longer lengths. So we already have two factors which appear to boost the effectiveness / functionality in our case.

Other possible benefits of producing our own skylights: We can adapt it to our specific use, and specifically we can make it more storm resistant or adapt it to the particularities of our roof (which is more-or-less flat and will be topped with a concrete slab, which is different from the average roof with shingles etc). Another is that existing skylights seem to be expensive, though I haven't researched prices much.

I wonder what people think about that reasoning so far. I am thinking of having openings that are as small as 4".

Now, aside from whether it is a good idea or not, I am looking for advice on constructing the skylights. It seems straightforward to create a cylindrical tube, and apply a reflective material to the inside. However I am not sure about:

  • what to use for the exterior light collecting window / dome, which are usually designed to collect light from all angles. We do not need to take into account storm resistance necessarily, because we can install separately some kind of shutter that encloses the dome. I am thinking that we may be able to just install a 10" dome even though the opening that it will sit over will be smaller.
  • How best to connect the tube to the dome, in a way that minimizes leaks into the roof opening/canal; or if they are not connected, how to best secure the dome to the roof to minimize leaks.
  • What to use for the interior circular window.
  • General construction advice for assembling these items.

I apologize if this question is very long. Thank you.

  • Contact the companies that make them - they will be able to answer : size, design etc
    – Solar Mike
    Jul 16, 2019 at 22:10
  • True. I did contact one company once and managed to talk to someone who knew something but I don't think I learned much. I haven't done any more digging since though. Maybe I should have tried to get more advice from them before seeking it here.
    – user98368
    Jul 16, 2019 at 22:16
  • The light received does not decrease "exponentially" with the diameter of the roof opening, but rather decreases with the change in the square of the diameter, that is, decreases with the cross sectional area. The reason why 10" is the minimum is that the disruption in the roof is about the same for 12" dia, 10" dia, 8" diameter, but the amount of light decreases significantly Jul 16, 2019 at 23:20
  • The light volume depends primarily on the collector size, not the tube diameter. I've installed a number of these, and the highly reflective nature of the tube means that bends and reductions have little impact. A larger collector (and a suitable diffuser at the other end) are key goals.
    – isherwood
    Jul 17, 2019 at 18:36
  • @isherwood True, but you can't "concentrate" a large diffuse collector into a small pipe; you'll get the same intensity no matter how large the diffuser is. And, I'll guess that the resulting intensity decreases with the cube of the diameter; the amount of light that enters is proportional to the tube area (square of the diameter), and the bouncing inside the tube increases inversely with the diameter (each bounce cuts the light by a percentage). Jul 17, 2019 at 21:22

1 Answer 1


Your question is very complete. But, for a flat concrete roof? Yes, I know, it won't truly be flat and it better not be. Either way is No Way...you'll be re-caulking every year and it'll take all year long with that Swiss Cheese.

Please have the mason customize the roof for these. I'd like to see a 2" to 3" (twice that if you regularly get snow and ice) ramped cone poured for each location that's part of the initial monolithic pour...not some add-on guaranteed failure afterthought nonsense that actually collects and retains water.

Also, please don't run short-lived Romex under the slab for the night and cloudy lighting. Use the aluminum armored cable, unless it's attacked by the concrete. Then you might need the PVC or epoxy coated armoring.

I'd actually suggest a sidewall stovepipe looking singular tube that feeds a fiber optic lighting box. Of course, it could be a nice feature like a faux gable or corner turret. But, this would allow for experimentation and scalability. Maybe 4-inches is too bright in a fiber optic setup and you just need a 2" periscope that operates as a flagpole.

  • Thanks for your response. I think I understand the first half, you are recommending that the concrete slope up toward / around the opening so that it is elevated relative to the rest of the roof slab. That sounds good. I wasn't completely sure what you meant in the second two paragraphs but I think it is about the functionality that some skylights have to charge a battery during the day which powers the night/cloud light you are referring to. We will not have anything like that so not to worry.
    – user98368
    Jul 21, 2019 at 13:10
  • Very good, though I still think you'd be best served by fiber optics...still do your tubular experiments and no leaks, no maintenance and nothing threatened by weather nor even trees. Best wishes either way.
    – Iggy
    Jul 21, 2019 at 20:32
  • I had actually never heard of fiber optic lighting so I see what you mean now. I have always taken fiber optic to refer to data cabling. I will check it out.
    – user98368
    Jul 22, 2019 at 18:22
  • Do all fiber optic light solutions require electric power for the rooftop light collector? I am hoping for something that is as "standalone" as possible. Do you have any recommendations for vendors or specific category of fiber optic lighting? Can I just purchase the filaments? Exterior appearance is not important but I need to decide beforehand on any openings in the building before construction.
    – user98368
    Jul 22, 2019 at 22:11
  • VERY GOOD! Thank you for looking into FOL and not just dismissing it. But, nope no power ever needed and you can make the light collector out of a shoebox if you want. And yes, you can get just the fibers themselves. Sorry, I only know the basics and played with it last century. Amazon has very cheap play-stuff, but start with fiberopticlighting.com for your goals. Maybe it works for you or it doesn't, it just sounds a whole lot better to me for you.
    – Iggy
    Jul 23, 2019 at 12:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.