Short-cut URL for this page: https://chrismolloy.com/skylight. This article is featured in ReNew magazine, issue 111 (April - June, 2010), pg 64 - 66, as well as on the Makezine.com blog.
'Success? Failure?' Rating: ★★★★★
I decided my 20W PV panel was not working hard enough. The solar-powered attic-to-basement fan system, plus a separate fan system used to redistribute heat from our lounge to our bedrooms in winter, still left a good deal of spare capacity when it wasn't hot or cold enough to need the fans running. The heat redistribution system also relied on a lead-acid battery to do its job, and I've never had much luck when it comes to lead-acid batteries. It was time to come up with yet another use for my PV panel...
A couple of articles by Lance Turner in ReNew magazine caught my eye: an article to build your own (mains-powered) LED light fitting (Issue 104/July-Sept 2008, pp 70-71); and an article about skylights, including how to build your own 'fake', LED-based one (Issue 108/July-Sept 2009, pp 26-28). What excited me about combining these two articles was the possibility of building a fake, LED-based skylight driven directly by a PV panel - with no electronics in between to complicate the system (and use some of the power being produced). This possibility hinges on the fact that a PV panel is a constant current source and an LED is a constant current consumer - a perfect match! A non-obvious (to me, at least) feature of this system is that the brightness of the fake skylight will vary with the level of insolation falling on the PV panel, and will thus mimic a real skylight amazingly well.
The basic plan is to hook up strings of Cree XR-E LEDs (rated at 3.3V - 3.7V, up to 1000mA) to match the rating of your PV panel. A good match (that allows a safety margin when driving the LEDs) is to hook up strings of five LEDs in series (i.e. with a total forward voltage of 16.5V - 18.5V), and then connect these strings in parallel, one for each 10W of PV panel output (i.e. ≈600mA per string). Based on this plan I needed two strings of five Cree XR-E LEDs in order to match my skylight unit with my 20W PV panel.
One possible addition to this plan concerns fuses. If one of your LED strings were to fail (for whatever reason), then the current from the PV panel will be split between the remaining strings. If this results in more than 1000mA (1.0A) flowing through the remaining strings then you risk damaging the LEDs in those strings. A solution is to put a 1.0A fuse into each string. The prudent approach might be to add this fuse into each string regardless of circumstances, but if it is unlikely to be needed (e.g. the loss of one string will not cause too much current to flow though the remaining strings), then the non-prudent approach might be to save the cost and complexity (not to mention a small drop in power) of putting the fuses in. I've written an LED Calculator to help you with this decision.
One further issue to deal with is one of heat dissipation. Cree XR-E LEDs run hot enough to do themselves damage unless they are properly heat-sinked (is that a word?). The ReNew magazine articles state that a good rule-of-thumb for this is to allow 100cm² [16"²] of 3mm [⅛"] aluminium plating per watt of LED. Ten Cree XR-E LEDs will thus require 3000cm² [465"²] of aluminium (e.g. 100cm x 30cm [39.4" x 11.8"]). Each LED should be securely fastened to your aluminium base plate, and should be bedded on heat-sink paste to ensure good thermal conduction.
In order to make my fake skylight look like a real one I wanted to have it mounted in a frame and behind glass. I decided to use aluminium 'U' channel to make the frame, and to ensure a good thermal bond between it and the aluminium base plate to further aid in keeping the unit cool. The frame was to be left open-ended to ensure good airflow through the unit. The glass was to be textured (unlike most real skylights) in order to blur the look of the LEDs somewhat. I also wanted to be able to tilt my fake skylight, rather than be forced to have it shine straight down, so I built a hinge/tilting mechanism into the frame as well.
LED Calculator
I wrote a wee calculator to help match your PV panel to the Cree XR-E LEDs. The calculator makes a recommendation, for your specific panel (as specified by entering its voltage and current at maximum output), about: the number of LED strings (such that no more than 1.0A flows though each string); the number of LEDs per string (such that there is no more than a 3.7V drop across each LED); and whether or not each string should be fused (i.e. would the loss of a single string result in more than 1.0A flowing though the remaining strings?).
To find the voltage and current at maximum output of your panel you can use the data from the manufacturer (which is unlikely to match your specific panel exactly), or you can measure it by pointing your panel directly at the midday sun on a cold, clear day (which should result in the maximum output you can expect from your panel). Apply your own judgement to the results from this calculator - no liability of any kind for the use, or misuse, of this information will be accepted by the owner of this web site.
Materials List
Here's a list of what I used to build my fake skylight. You may want to tailor this list to meet your specific needs:
- 10 x Cree XR-E Q5 LEDs
- 1 x 1000mm x 300mm x 3mm [39.4" x 11.8" x ⅛"] aluminium plate
- 1 x 1000mm x 285mm [39.4" x 11.2"] textured safety glass (ensure whatever you use complies with your local building code for overhead use).
- 2 x 1000mm x 25mm x 25mm x 3mm [39.4" x 1" x 1" x ⅛"] aluminium 'U' channel
- Ø3.2mm [⅛"] aluminium rivets: 6.4mm [¼"] and 9.6mm [⅜"] grip depth
- M3 [Ø⅛"] x 12mm [½"] button/dome/pan head bolts + nuts
- Cable/wire; a connector block; heat-sink paste; a means to connect your skylight to your ceiling (I used hinges and hasps)
- Optional: fuse holders and 1.0A fuses (one per string); a switch on the PV panel side of the connector block to allow you to disconnect the power more easily.
The total cost to build this unit was around NZ$300, excluding the PV panel. The single most expensive item was the safety glass (NZ$120), which could easily be omitted if cost was more important to you than aesthetic. As a comparison, a single Solatube-type unit (including installation) can be had for around NZ$800.
Building Steps
Here's what I did to put my unit together...
 |
 |
 |
 |
 |
 |
 |
 |
   |
Success? Failure?
★★★★★
Pros: By far the best PV panel / LED -based thing I've built to date - this unit is extremely functional and pleasantly illuminates an otherwise dim end of our lounge.
Cons: None that I can think of.
Enhancement Suggestions: Add a stick-on strip thermometer (the kind you see on the outside of fishtanks; available from any pet shop) to the outside of the aluminium frame so you can monitor the temperature of the unit. These thermometers typically only go up to 38°C [100°F] or so, but this unit should run well under that.
Update: I replaced the glass with a diffuser (½" grid, mirrored plastic) and the result looks a lot better, and is more effectively diffused. I suspect the ventilation is better, too (not that heat has been a issue at all).
¤ Copyright 1999-2024 Chris Molloy ¤ All rights reserved ¤