Shop Products
Houzz Logo Print
doubletagger

48v dc led wiring for a new build; xm-ls, etc.

doubletagger
11 years ago

Well, I'm new here, but my wife isn't. I just thought I'd create my own account. My wife is probably thankful for that. :)

We're about to start construction on our new home. I'd like to design the lighting, and, inspired by the few but bold that have taken on the challenge of "purpose-built LED", I march forward. The idea behind a "purpose-built LED" system is to maximize efficiency and match well one's design requirements with a solution, rather than just using off-the-shelf stuff. If we wanted recessed cans and dimmers and the likes, that "normal people" want, then probably using off-the-shelf pre-built stuff at 120V AC would be the best bet, anyway. But, alas, I'm not a very "normal" person.

Anyway, I'd like to run some numbers by whomever out there has the guts to tackle them, and get any feedback you may have.

First some assumptions: we'll assume that I an LED solution for a variety of reasons, and that CF or Ts or halogens or other alternatives aren't really options. We'll assume that I have an acceptable (to my wife and I) physical design, which involves strips in dropped crown-molding ("mini-soffit")-like configurations, run around the perimeters of rooms (or portions thereof) where LEDs are bounced off the ceiling (indirect). There are a couple of variants, and, in the kitchen, a direct down-lighting design prevails, but one that will still take the form of "strip". Putting those assumptions together, one option for component LEDs are the Cree XM-L's have now been around a couple of years. Now the picture is becoming clearer: I'm interested in buying some aluminum plates (or other substrate / cooling-configuration) and spacing the emmitters on them myself, as the given room/application dictates. We'll assume that I have some electrical experience, but inadequate LED experience. We'll assume that we're not the pickiest people when it comes to temperature / color. Finally, we'll assume that I'm _not_ interested in dimming. I will in some cases bank the lighting for 2 or maybe 3 brightness configurations, but that's it. However, there are some instance in which I'd like 3-way switching, so to speak, and I think I'd like to run the switching, if possible, with signal-wire, to the wall switches, which would be momentary switches of some sort (no state). More on this later.

I'd like to run things at 48V DC - just under the 50V threshold for obvious reasons, but high enough to minimize line loss if I need to take the DC any great distance. I'm open to transformer options and configurations, and, of course, have access to all the insides of the walls, can put boxes wherever I want, etc. Likewise, I'm confident that I can easily find physical housing for drivers. I know that if I use #12 Romex then my 5% drop distances for 4, 6, 8, 10 Amps are, respectively, 132, 90, 64, and 52 feet, which are all acceptable in my application. I will, of course, take distances and transformer locations into consideration when making final wiring choices. I'm not convinced that using a single transformer for the whole house's DC lighting is the best way to go, and I'm ok with having several regional transformers, say, each on their own little 10 or 15-Amp circuits (A/C, from the panel to the tranformer at 120V), to provide the DC to a given, say, few-hundred square-foot region of the house.

For a given room / application I have an approximate lumen requirement, albeit, since I'm not the most familiar with lighting, my calculation is based on incandescent equivalencies. E.g., in one room, I'd like the equivalent of about 2 75-Watt incans for whole-room lighting. No work areas or anything. This example happens to be a bedroom, in which we'll have reading lights at the bed. So, that's about 2360lm. If the XM-Ls are 250lm at 1Amp (I'm not sure if that's exactly right, but approx.), I might need about 10 of them to do the job. Possibly 5 along one wall and 5 along another, spaced a couple feet apart would do it, right?

And, if I'm paying about 6.5-cents per kwh right now, and wanted to compare, then my 2x75-watt incandescents would cost me 10-cents a day, or $35.59 a year if run 10-hours a day. Roughly, at 100lm/watt, I guess I'd be running the LED equivalent at about 25 watts (compared to the 150-watt incandescents), which calculates out to 2-cents a day or $5.93 a year. I know, I'm leaving out the consumption of the drivers, but if constant-current drivers run about 96% efficiency, I'll just let it go for the calculation for now. If my house was over-simplistically scalable and the whole-house load was 15-times this, then we're looking at a $1.20/day or $444.90/year difference (savings). I see XM-Ls on ebay for $60/10pcs ($6/pc) (is this a decent price? I haven't looked at vendors yet). If my whole-house scenario was over-simplified and I decided that I needed about 150 of these little guys, that would come to $900. The power-supply(s), wiring, boxes, mounting, etc. might all add up to a couple thousand (very rough). So I'm looking at a few thousand, perhaps, not including labor and soffit-building, etc., which I intend to do. So the ROI would be about 6-10 years. Well, big error, of course: I'm assuming that a bunch of incandescents and conventional wiring would cost $0, but that would just reign the ROI in a little. And if my LEDs last 15 years, then I only have to replace them at under $1000, likely, in 15 years, and the ROI on that is 2 more years, for another 15 years of life, presumably. Again, assuming that incands were 0$. I can handle that. I'd net more than $2k in the first 15 years, and then nigh $7000 over the next 15 years.

About switches... in a couple of applications I'd like switches at each end of a room/hall. I don't like state switches, and I'm not fond of touch-switches. I'd like a mechanical momentary switch. I haven't looked hard for one yet, but will soon. Any advice on a switch? Any advice on how to deliver a switch signal to the driver (I'd prefer to signal post-transformer, I think, but perhaps there's reason to consider 120V switches, even if I go with a momentary switch, or possibly low-voltage switches that go to latching relays on the 120V circuit?). Do I need to breadboard my own solution, complete with debounce and whatnot? My electrical experience of this sort is quite rusty, so I'd love it if there was a prefab solution I could make use of here.

An aside: I do have plans for PV-solar someday, but it might be 150' from the house, so I have to figure out how to carry it up high-voltage/low-current, and in big wires, to reduce loss; I'm sure it's still easier to step down to 48-V than it is to transform A/C, so, when the time comes, I'd like to have a way to patch that source in directly (that is, from the batteries). I don't think this will be hard, but if I'm transforming in the wall somewhere, I have to keep in mind how to get the new source there in the place of the AC; I may be able to just use the existing romex to do it, and be fine, and just bypass the transformer there, possibly replacing it with step-down circuitry. This isn't the most important piece of the puzzle, just something I'd like to think ahead on a little bit.

Now for color (temperatures, CRIs, etc.) I have XLampXML_HVW.pdf, the poop sheet which characterizes their "cool", "neutral", and "warm" white options, and I can take some jargon-filled advice from other threads about what "good" color might be, and I can, as advised, order some stuff and see what it looks like before ordering the big bundle, but does anyone have any basic advice for this product in particular, to get me started. Devoid of any advice, I'd probably try an assortment, but would stick closer to the warm end of things than the cool end. I like incandescent color alright, and the photographer in me likes anything between white and warm for portraiture, but I really am fairly liberal here; not even the CF "green" bothers me as much as it does some people (I wouldn't shoot portraits under it, but I can live with it). Of course I don't want green hues, but, I'm thinking I might have a pretty easy job of choosing here; my design will reduce glare problems, and my lax color attitude might lead to happiness with anything in an acceptable spectrum. Color-crats have it harder in life, I'm sure.

And finally, cooling: I really like the idea of running some water in tubing to which the LEDs are mounted. I know copper would work best, but it's not the most cost-effective. Otoh, it offers a rigid mount point (I could solder the emitters directly on). I know PEX is more insulative, but cold well water in PEX should still help deliver off the heat a bit, if I can figure out how to mount the things well. I have a couple of options for what to do with this water, but, for the sake of argument, just assume that it gets fed, thus pre-warmed, into our water-heating system. So, an assumption is that maximum efficiency occurs when light usage and water demand are in concert, or if water-demand is somehow constant. Since we're doing in-floor radiant, demand may be somewhat constant, if we go open-loop. But most likely we'll run closed-loops through heat-exchangers. In this case, we'll have to rely on the coincidence of water-demand and light-demand, but that may work out pretty well, actually. Any advice on doing something so unconventional would be greatly appreciated.

Comments (11)

  • David
    11 years ago
    last modified: 9 years ago

    A couple of comments for starters.
    1. A LED lamp/ strip is more than just the emitters soldered onto some wires (or copper tubing) that are in turn connected to some power supply.
    2. The LEDs will need to be powered at the voltage (and current) ranges specified by the manufacturer, which may not be 48V. The current might have to be lower than 1A for the sake of longevity.
    3. The cost of a LED lamp is higher than just the cost of the emitters. Dismantle a Cree CR6 and some other LED strips to get an idea of the Bill Of Materials (BOM) and cost.
    4. The BOM cost goes down when millions of devices are built. An emitter that you might pay $6 (+ shipping) would cost significantly less for an entity like Philips.
    5. The wiring and power supply setup for an extensive low voltage system will probably be a significant undertaking. There will be significant design, ... and regulatory hurdles (i.e. - building code) to clear.
    6. It might not make practical monetary sense to custom design an extensive, unique low voltage lighting system.

  • brickeyee
    11 years ago
    last modified: 9 years ago

    There is a very old engineering joke that the color of the 'bleeding edge' of technology is green, the color of money.

    Dealing with ANYTHING that is not in wide use and VERY standardized is expensive to develop, and then expensive to maintain.

    Try getting an electronic part more than a few years old.

  • doubletagger
    Original Author
    11 years ago
    last modified: 9 years ago

    Thank you very much for your post davidtay.


    1. A LED lamp/ strip is more than just the emitters soldered onto some wires (or copper tubing) that are in turn connected to some power supply.

    My bundled figure for transformers, driver circuits, and wiring may have been a little low, but I think it's close. I didn't break it out b/c I was already a little long-winded. :) I did leave out the cost of copper, though; I'm not sure that's the way I want to go anyway.


    2. The LEDs will need to be powered at the voltage (and current) ranges specified by the manufacturer, which may not be 48V. The current might have to be lower than 1A for the sake of longevity.

    Right, the particular one I mention can be driven at 2A, but it'll last a lot longer at 300mA. I was thinking about driving it at about 700mA. Of course, I would have to design/purchase a driver that will, among other things, step down voltage if necessary and supply the right current to drive at the proper amperage. And, of course, the things burn brighter at 2A than at 300mA, but last longer at 300mA, meaning that you have to buy more to achieve the desired lumens. Definitely a balancing act.


    4. The BOM cost goes down when millions of devices are built.

    However, the manufacturers, especially for something as hot as LEDs, and requiring as much design expertise as they do, set their markups liberally, too. Still, the case can be made (elsewhere) for the value of a purpose-built custom system, when it comes to LEDs, and that case has been made elsewhere, so I won't belabor the point.


    5. The wiring and power supply setup for an extensive low voltage system will probably be a significant undertaking. There will be significant design, ... and regulatory hurdles (i.e. - building code) to clear.

    I believe in most jurisdictions electric code compliance starts at 50V. Anything you do under 50V DC you're welcome to without constraint. Of course code has to apply to the transformers, but generally not after that.

    At least, those are some of the motivations I've been riding on when trying to decide whether or not to do this.

  • David
    11 years ago
    last modified: 9 years ago

    If you're still intent on doing it,
    1. There are newer emitters from CREE which are cheaper and more efficient. That could eliminate all the cooling requirements. You could try asking them for reference designs.
    2. The controls are cheapest if you use either ELV or MLV switches to throttle the low voltage power supplies (either ELV or MLV).

  • David
    11 years ago
    last modified: 9 years ago

    You may want to consider the possibility of getting the emitters from Cree's distributors instead of eBay since they would be able to provide some support. Pricing is ~ $1.50 or so per emitter in lots of 1000.

    I would take that route if I were building my own LED lamps.

  • doubletagger
    Original Author
    11 years ago
    last modified: 9 years ago

    Thank you brickeye.

    My personal experience doing several projects that are not "VERY standardized" is contrary to yours, it seems. I've usually highly enjoyed such projects, and often found them very fruitful. It is true that they often take more time. Anyway, fwiw.

  • doubletagger
    Original Author
    11 years ago
    last modified: 9 years ago

    Thank you very much davidtay - some really helpful pointers!

  • David
    11 years ago
    last modified: 9 years ago

    You're welcomed. Good luck on your project. I hope you get to build your own >=150 lumen/watt lamps.

  • brickeyee
    11 years ago
    last modified: 9 years ago

    "My personal experience doing several projects that are not "VERY standardized" is contrary to yours, it seems."

    If you work for yourself for free it is hardly a job, more like a hobby.

    Try and find parts in a few years.

    I make a lot of money now chasing down 'equivalent' parts for older systems.

    From building controls to satellites.

    A 10 year old electronic part (especially semiconductors) of known provenance is painfully expensive.

    Parts that cost only a few dollars ten years ago are now hundreds of dollars, if you can even find them.

    And counterfeit parts are being made all the time.

  • andersonsb
    10 years ago
    last modified: 9 years ago

    Hi doubletagger,

    I'm very curious about whether you decided to go ahead with this project, and if so, how it turned out. It looks as if this thread was last active about a year ago (?).

    Would you mind dropping me a line via the "email a member" link in my profile? I'd be happy to correspond with you offline about this if it's OK with you.

    Thank you.

  • brickeyee
    10 years ago
    last modified: 9 years ago

    "I believe in most jurisdictions electric code compliance starts at 50V. "
    Only if the power is VERY limited.

    Power limited control & communication circuits get a lot of relief (like doorbells, thermostats, and phone lines).

    A 12 V, 30 amp circuit is NOT power limited enough to not be a source of ignition.

    Your phone uses -48 VDC and then stacks about 90 VAC (at 20 Hz) on that for ringing.
    It is still a very power limited circuit though (it cannot deliver much current) so it is not a source of ignition for fires even if you short the wires (that just irritates the TELCO).