Wiring a house for LED

[ddrueding]

Looks like I'll be buying a house soon. And in my price range, that isn't a house you can move into. Since I'll undoubtedly be doing a fair bit of demo, and reworking the electrical and plumbing already, it has me thinking about LED lighting. I'm sure that JTR will be the star here, but I suspect that the other resident geeks might find the conversation interesting.

Here is what I know so far:

1. LED is too expensive to be used as the sole light source for a room, and is better used as accent and task lighting. In my application I plan to use linear florescents in hidden up-light fixtures as general flood lighting.

2. High-powered LED produce a lot of heat, and LEDs need to be kept cool to maintain lifespan and efficiency. Active cooling is sometimes used, but isn't the ideal.

3. The power input is tricky, instead of requiring a specific voltage, it requires a specific amperage. This involves some complex electronics, especially if you intend to make them dimmable.

So, if I have the opportunity to wire the house from scratch. What is the best wiring method to support dimmable recessed ceiling spots? What sort of enclosures will I be looking for? Is all this stuff certified so my home insurance won't be voided?

Thanks in advance...

 

[jtr1962]

I'd try to put in low voltage DC wiring as much as possible. Common voltages are 12VDC, 24VDC, and 48VDC. To convert 120VAC to these voltages a variety of options exist. PC power supplies are one if you want 12VDC. 48VDC is a common voltage in industry and I'm sure many options exist to supply that. The basic idea is to start off with a fairly low DC voltage, and then run LEDs in series strings with a total Vf a few volts less than the supply voltage. For example, if you're using 24VDC and XP-Gs with a Vf around 3.2V @ 700 mA, you might put 5 or 6 or 7 in series. You can then use a step-down driver to regulate current. When LED Vf is that close to supply voltage then a switching driver can be quite efficient ( 96% or more ). Dimming could be done with a little extra circuitry, and the dimming pot and knob mounted remotely from the LEDs. Or better yet, put the LED driver, dimming circuitry, and control knob in the same box (the usual 120VAC outlet boxes will do just fine here), and mount on your wall just like any other switch. This way if the driver goes bad, just open up the box, and replace it easily with going into the ceiling. And I can probably supply you with all the needed electronics here as I already have a few good LED drivers in production.

You can probably do something similar if you stick to 120VAC exclusively as well. 120VAC as the starting voltage just means the driver is a somewhat larger and more expensive (and might need UL listing for insurance reasons). On the other hand, if you stick with low voltage wiring for all your LED needs, then I think the power supply which transforms 120VAC to low voltage DC is the only part which needs to be UL listed. A low-voltage DC bus has other advantages as well. If you decide to go "green" in the future with solar panels, you can hook them up directly to the low-voltage wiring (or at least interface them more easily than you would if the wiring was all 120 VAC). The main disadvantage of low voltage DC is the higher current for any given power level. However, since this wiring would be used exclusively for LED lighting, that shouldn't be an issue. Naturally, you still need to protect the wiring somehow against short circuits. The power supply will probably already have some protection. A circuit breaker or fuse would provide an additional layer of protection.

Now the question of how hard to drive the LEDs, and how to cool them. Naturally, the cooler the better. Look at this document from Cree regarding lumen maintenance. Driving at 700 mA seems to be a good compromise between the lifetime and keeping the number of emitters to a minimum. If at all possible, the larger your fixtures are the better. It is actually possible to use LEDs for all your lighting if you have the bucks. As I've said many times the light bulb form factor isn't well suited for LED lighting. Instead, think in terms of using an entire fixture for passive cooling. The fixture can serve both a decorative and heat sink purpose. It's possible to get 1000+ lumens from even a modest-sized fixture with today's LEDs. And you can probably get upwards of 10,000 lumens from a fixture large enough to fit in a 2'x4' space in a suspended ceiling. But for accent lighting the fixtures don't need to be huge. You might find that you can actually get more than adequate lighting most of the time just from a few fairly small LED fixtures. Let's not forget that today's better LEDs output about 30% to 40% of their power as light. That means less heat to be gotten rid of.

Last thing for now is color rendering. If you use the most efficient LEDs, which are the Cree XP-Gs at this time, then they are roughly ~6500K with OK but not spectacular color rendering. I've found that adding one 640 nm red per maybe 6 whites, and maybe one 505 nm cyan per 10 whites, improves CRI markedly without impacting efficiency or color temperature much. Provided everything is diffused properly, the colors should mix in well with the whites. CRI should improve from low 70s to probably mid-90s. And on another note, don't let that low 70s number fool you. For whatever reason, I've found that LEDs seem more pleasing for any given CRI than fluorescents. Maybe it's because the LED spectrum is continuous, albeit with a few valleys and peaks, while a triphosphor fluorescent spectrum isn't.

I'm sure I'll have more to add later. That's all I could think of for now.

Oh, and congratulations on the house! :aok:

 

[ddrueding]

Thanks for the really quick and in-depth response! I'll do more research before my next questions (now that I know which directions to look).

 

[ddrueding]

I'm not sure where to aim on power/lifetime of the LEDs. Considering I plan to over-spec the lighting of the space considerably, 90% of the time they will be dimmed quite a bit. Also, how long will it be before efficiencies and light color have made major jumps again? Would I do better to run a smaller amount of LEDs harder to get my money's worth out of them before they get outdated?

From my very limited knowledge of electricity, a higher voltage allows the same wattage at a lower amperage, and therefore less loss across a wire, correct? So running 48 volt would be more efficient? And running 14 XP-G LEDs per series would be about right?

Depending on the binning, 14 XP-Gs would put out about 2000 lumens? (I only see info for driving them at 350mA.)

Did I mention that this room is big? 650 square feet? How about that it includes 2 major work areas (kitchen and home office)? All-in I can see wanting 16000 lumens in there. That would be 112 XP-Gs. A bit less than $1k? They could all be driven by a pair of 960W 48v power supplies that would run about another $500. Haven't talked about power control, dimmers, heatsinks, mounting hardware, wiring, etc (another $1500?)

Does my math look about right? If so, I think I'll go for it.

 

[Pradeep]

That's crazy prices for one room of lighting. A couple of 8 bulb T5 flo fixtures would be half the price, whilst allowing you to customise the color frequency - mix up 6500 and 4100K tubes etc. Aquarium store should have them for tank lighting to demo

 

[ddrueding]

I've been studying a bit on lighting design, and the the biggest cheat I've been told so far is to break up the light into as many distinct sources as possible, and make each group independently dimmable. Even if I were to count on T5s for the bulk of the light (60%?), they would likely not be on most of the time due to their lack of control, while task lighting was cranked up to compensate.

 

[udaman]

I've been studying a bit on lighting design, and the the biggest cheat I've been told so far is to break up the light into as many distinct sources as possible, and make each group independently dimmable. Even if I were to count on T5s for the bulk of the light (60%?), they would likely not be on most of the time due to their lack of control, while task lighting was cranked up to compensate.

As I blast through the 1k post mark (gotta get me post/day ratio up to 0.4 ) I would say, the nanny state would look down on your project, at least in LA Cty. Maybe it's different up there, but there a codes that must be adhered to, you'd need an licensed electrician to do this work.

Since you r doing a DIY project apparently, suggest you do *lots* more research, like months!

All too easy to see how little you've put into the process so far. Ever been by a construction site, arc welders? Do you like the 'quality' of light? Do you like starring into the glare of an HID light? Maybe jtr should send you the 3aaa flashing, with the zoom lens he built me, it's got short runtime, running 1.2A IIRC, and a R5 binned XP-G (which btw, cost about $7 ea). Put that light up on the ceiling anywhere, look at it, tell me if guests are going to like that glare and color renditioning...if you compare to a filtered/softened high CRI fixture...night & day, IYAM .

These rooms, you ever plan on having guests? You want massive glare shinning at them when they look up at the lights???

I suggest you find some type method or fixture that allows for substantial filtering/softening of the light.

With that you need less concern yourself with color. And never mind what jtr says about color, he has a M$/Intel SSD/Grado mindset when it comes to colors. Lots of low CRI LED's look BFin ugggggly, with that characteristic green/yellow tint that make some people want to puke...like my mother. Where as jtr finds the CRI 100 incand, cause him to be nauseated, but he's *not* the norm.

1/2 the effeciency, you can get high CRI LED's now, if you're going with *bare* emitters, I'd suggest you look into these.

http://bits.blogs.nytimes.com/2009/01/26/leds-to-the-rescue-not-so-fast/

Getting closer to making jtr1962 happy-High CRI at 5K
http://www.candlepowerforums.com/vb/showthread.php?t=249232

^take a loot @McGizmo's thread on the Nichia CRI 92 LED's (Ra=92 = CRI), you can't overdrive these however, 700ma will toast them>50,000hr down to >5min .

You might also look at jtr's tests on the SST-90, that will get you equivalence to 100w or more incand bulbs, @100w = 1600lm, only the SST-90 can do this as a single LED. 200W bulb = 2500lm, @ very high current the SST-90 is the only single LED that could do this. But the glare! Harsh, harsh, harsh shadows, worse that being outside in the daylight bright sun trying to get a decent photo w/digicam.

 

[ddrueding]

Color quality is important, but I've found that CRI is a pretty poor indicator thereof. I'm currently running 12 high-CRI 6000K linear tubes in my office, and my wife and I both like them. There is glare, but that is because it is direct light, not because of the source itself. I plan to make most of the light indirect, bouncing off the vaulted ceiling, fireplace, wall wash, etc. The only direct light would be onto the kitchen counter top and desktop, where there will be enough distinct light sources to distribute the glare.

I would be interested in the best color quality, but it seems this is harder to achieve at higher color temperatures. Low lumen count per emitter isn't a big deal, adding more point sources certainly isn't a bad thing, and being able to direct the light is a good thing.

Uda, I've noticed your perspective and approach is very common on the internet. And can be summed up with the following points:

1. It's a bad idea.
2. You need to do more research.
3. It's too expensive
4. Overlay all of the above with a condescending, elitist tone, to prove you're right.

I can understand this mentality from teenagers in their basement with no money, but we are grownups. Our money is our own (or our wives'), and our life is a game, with our belongings all being toys. Play with your toys! Have fun! Experiment!

 

[jtr1962]

I'm not sure where to aim on power/lifetime of the LEDs. Considering I plan to over-spec the lighting of the space considerably, 90% of the time they will be dimmed quite a bit. Also, how long will it be before efficiencies and light color have made major jumps again? Would I do better to run a smaller amount of LEDs harder to get my money's worth out of them before they get outdated?

If you're thinking in terms of eventually replacing the LEDs with better ones, then I'd say consider 1 amp drive currents, especially if you'll be dimming them much of the time. As that document I linked to shows, at 1 amp with enough cooling you can still get in excess of 60,000 hours ( > 13.5 years @ 12 hours per day ) until the LEDs dim to 70%. The advantage of using 1 amp is that you'll only need about three quarters the number of emitters as you would by driving at 700 mA.

As for when efficiencies and colors make major jumps again, that's hard to tell. As mentioned in the other thread, Cree recently lab-tested an LED at 186 lm/W. While spectacular, note that this result is only about 34% better than the best production LED. I'll go out on a limb and say that if you plan to replace the LEDs in your project in 10 to 15 years time, you might have 225 lm/W ones available by then. And you'll almost certainly have high-CRI ones in a variety of color temperatures. However, given the roughly 30% penalty for high CRI at higher color temperatures, figure any high-CRI LEDs might only achieve a bit over 150 lm/W, or only slightly better than today's best XP-Gs. The real improvement might come from conquering droop. Right now the best XP-Gs only do about 107 lm/W at 1 amp. If we can mostly conquer droop then we might see nearly double the output in the future, or perhaps 1.3 to 1.4 times more with high CRI.

I'll also add that part of the problem making predictions like this is that I simply don't know exactly what all the issues are with making LEDs more efficient. It's entirely possible that we may struggle for decades to get much past 200 lm/W. Or if we're really lucky, and achieve major advances with green and red LEDs, then we could see color-mixed RGB whites of >300 lm/W AND decent color rendering in a decade's time. But regardless of whether my predictions are off or not, we'll certainly see a notable improvement in LEDs by the time the ones you're using start to dim. Also note that as we pass 200 lm/W in production, don't expect to see much further improvements with phosphor-based LEDs. You probably can't make 100% efficient blues for one thing. And you have about 18% inherent losses in the phosphor conversion process. Package efficiency isn't 100% either. Add all these things up and you get the roughly 225 lm/W I mentioned earlier. However, if you can hold those kinds of efficiencies past 1 amp then you can double the output per emitter. And cost per emitter will almost certain come down, perhaps to as little as one-sixth of today's prices.

From my very limited knowledge of electricity, a higher voltage allows the same wattage at a lower amperage, and therefore less loss across a wire, correct? So running 48 volt would be more efficient? And running 14 XP-G LEDs per series would be about right?

Depending on the binning, 14 XP-Gs would put out about 2000 lumens? (I only see info for driving them at 350mA.)

Yep, higher voltage equals lower current. That's why railroads string 25 kV catenery instead of continuing to use 600 VDC third rail. Same power can be delivered but with 2.5% of the current. If you use 48 VDC and want to run your XP-Gs at 1 amp, then figure 12 or 13 per series string to allow a little overhead for the driver.

If you want 2000 lumens R5 XP-Gs, then based on my tests you'll need about 14 emitters. If you drive then at 1 amp, then you only need 6 emitters.

Did I mention that this room is big? 650 square feet? How about that it includes 2 major work areas (kitchen and home office)? All-in I can see wanting 16000 lumens in there. That would be 112 XP-Gs. A bit less than $1k? They could all be driven by a pair of 960W 48v power supplies that would run about another $500. Haven't talked about power control, dimmers, heatsinks, mounting hardware, wiring, etc (another $1500?)

If using XP-G R5s at 1 amp, then you need about 45 emitters to get 16,000 lumens. Figure four series strings of 12 off 48 VDC for a slight cushion. The LEDs will consume roughly 3.3V x 1 amp x 12 = 39.6 watts per string. The constant current drivers should be about 96% efficient, so total power consumption from the 48 VDC rail for each string would be around 41 watts. Total power consumption would be 162 watts. A decent 300 or 400 watt, 48 VDC power supply should suffice. The 48 emitters will cost about $5.60 x 48 = $264 plus shipping at Cutter's latest prices ( actually the R5 is currently out of stock for a few weeks ). The power supply might be $200-$300. The wiring, electrical boxes, and drivers would probably cost roughly the same. So now we're pushing $900. The heat sinks can be basically 1/4" thick aluminum plates perhaps 3" to 6" wide ( space the emitters a few inches apart ). You can probably find something suitable at Home Depot or Lowes for roughly $100. I'd guess then total for the project might be $1000. Allow maybe another 25% just in case, so $1250 tops. It might also be interesting to add 2 reds and 1 cyan per string of 12 to improve CRI a bit. That might add another $50 or so to the cost.

Does my math look about right? If so, I think I'll go for it.

You actually overestimated things a bit. The only unknown is the 48 VDC power supply. I estimated as much as $300, but it could turn out to be more or less.

 

[ddrueding]

Thanks for all the numbers JTR, this will definatly be a direction I will be going. The current offer on the house expires in about 4 hours...we'll see.

As an interesting aside, I was planning on putting in an air-source heat pump water heater. The fun part about that is that any waste heat I can send it's way will make it more efficient. Water-cooling the LEDs to a radiator next to the water heater?

 

[jtr1962]

Maybe it's different up there, but there a codes that must be adhered to, you'd need an licensed electrician to do this work.

Hence my suggestion of using low-voltage wiring. I think anything 48VDC or less is considered low-voltage, and not subject to the same regulations as 120VAC house wiring.

Since you r doing a DIY project apparently, suggest you do *lots* more research, like months!

Agreed. Perhaps even start a thread on CPF to solicit other opinions.

All too easy to see how little you've put into the process so far. Ever been by a construction site, arc welders? Do you like the 'quality' of light? Do you like starring into the glare of an HID light? Maybe jtr should send you the 3aaa flashing, with the zoom lens he built me, it's got short runtime, running 1.2A IIRC, and a R5 binned XP-G (which btw, cost about $7 ea). Put that light up on the ceiling anywhere, look at it, tell me if guests are going to like that glare and color renditioning...if you compare to a filtered/softened high CRI fixture...night & day, IYAM .

The lens on that light does make it appear somewhat bluer than a bare emitter, so that wouldn't be a fair comparison. And I did suggest adding reds and cyans to improve CRI a bit ( and reduce the CCT slightly from perhaps 6500K to 5500K-6000K ). You get far less of an efficiency hit that way compared to using high CRI LEDs like the Nichia 083/183, and can still get CRI into the 90s.

I suggest you find some type method or fixture that allows for substantial filtering/softening of the light.

Ceiling bounce works well. I actually did this with colored Rebels mixed to provide white light and the results are satisfactory.

With that you need less concern yourself with color. And never mind what jtr says about color, he has a M$/Intel SSD/Grado mindset when it comes to colors. Lots of low CRI LED's look BFin ugggggly, with that characteristic green/yellow tint that make some people want to puke...like my mother. Where as jtr finds the CRI 100 incand, cause him to be nauseated, but he's *not* the norm.

I don't like incands because typical 2800K household incandescent is like living inside a pumpkin, irrespective of the 100 CRI. CRI numbers are only meaningful when comparing sources at the same color temperature. CRI by itself tells you little about how well the source accurately renders colors. A very high or low color temperature source will render colors poorly, regardless of CRI. I'm all for higher CRI than most present-day LEDs can deliver, just not at 2800K or 3000K. 4000K is passable, 5000K - 6000K close to perfect for the lighting levels I typically like. 3500K might be OK at very low lighting levels but it's marginal at best. Anything under 3500K is puke city for me. uke-r:

You might also look at jtr's tests on the SST-90, that will get you equivalence to 100w or more incand bulbs, @100w = 1600lm, only the SST-90 can do this as a single LED. 200W bulb = 2500lm, @ very high current the SST-90 is the only single LED that could do this. But the glare! Harsh, harsh, harsh shadows, worse that being outside in the daylight bright sun trying to get a decent photo w/digicam.

The SST-90 might be good if you want a single, intense source. I personally wouldn't use one to light a room unless I was trying to duplicate lunar noon. Many distributed sources, like the 48 emitters I mentioned in my last post, would work much better here.

 

[time]

That's crazy prices for one room of lighting. A couple of 8 bulb T5 flo fixtures would be half the price, whilst allowing you to customise the color frequency - mix up 6500 and 4100K tubes etc.

I was thinking that as well, except that multi-tube fixtures would be way too bright - unless he has a 12' ceiling.

You only need 6 T5 tubes to deliver 16000 lumens, they're seriously bright - especially if you use high-efficiency reflector fixtures. If you arranged them in pairs (end-to-end, 4' apart), with 8' between pairs, that would solidly cover an area 32' x 20' (10mx6m). In practise, you'll be constrained by the ceiling joists if you want to recess them (which I certainly would).

 

[ddrueding]

Even recessed, T5 tubes have some ugliness going on. I was thinking of building soffits, so I could stick the tubes in shining upwards, but that would cost a fair amount as well...

 

[time]

Even if I were to count on T5s for the bulk of the light (60%?), they would likely not be on most of the time due to their lack of control, while task lighting was cranked up to compensate.

You can dim T5s you know, you just need (expensive) dimmable ballasts. It's well worth considering this anyway.

But yeah, the more sources you have, the more likely the result will be aesthetically pleasing. So indirect is always the ultimate - and that's best achieved with fluorescent.

Just a personal observation on dimmers; I used to have heaps of them, but to be honest, they were only ever used for two, or at most, three different levels. It eventually gets a bit tedious to have to twiddle a bunch of knobs every time. Stored profiles would be great for what you want to do.

Don't forget that LEDs are likely to change color as you change drive current - could be an argument against JTR's suggestion of going to 1A, but I bow to his experience.

 

[time]

Even recessed, T5 tubes have some ugliness going on.

Yes, there are square recessed commercial fixtures (shorter tubes) that can look highly attractive, but I doubt you could mount them in a domestic situation.

Nonetheless, all these are off-the-shelf solutions and preferable for an investment like a house, i.e. don't frighten future buyers.

Also, ceilings with 20 downlights don't look so good either.

 

[time]

Maybe it's different up there, but there a codes that must be adhered to, you'd need an licensed electrician to do this work.
...
Since you r doing a DIY project apparently, suggest you do *lots* more research, like months!
...
Put that light up on the ceiling anywhere, look at it, tell me if guests are going to like that glare and color renditioning...if you compare to a filtered/softened high CRI fixture...night & day
...
These rooms, you ever plan on having guests? You want massive glare shinning at them when they look up at the lights???
...
I suggest you find some type method or fixture that allows for substantial filtering/softening of the light.
...
1/2 the effeciency, you can get high CRI LED's now, if you're going with *bare* emitters, I'd suggest you look into these.
...
the SST-90 is the only single LED that could do this. But the glare! Harsh, harsh, harsh shadows, worse that being outside in the daylight bright sun trying to get a decent photo w/digicam.

OMFG, I agree with everything that Uda said.

I'm just going to go away and cry in a dark corner somewhere.

 

[jtr1962]

Just a personal observation on dimmers; I used to have heaps of them, but to be honest, they were only ever used for two, or at most, three different levels. It eventually gets a bit tedious to have to twiddle a bunch of knobs every time. Stored profiles would be great for what you want to do.

That should be possible with a microcontroller and a little programming. The driver ICs I use can have PWM dimming, and most microcontrollers do PWM.

Don't forget that LEDs are likely to change color as you change drive current - could be an argument against JTR's suggestion of going to 1A, but I bow to his experience.

Yes, I've thought of that. That's one reason why I suggested adding reds and cyans, with each color driven by its own driver. If/when the tint shifts with age, you can compensate for it to some degree. In fact, if you want to get really fancy, then you can also add ambers ( and/or additional reds/cyans ), and have some sort of variable color temperature. Of course, calibrating it is the really hard part. If you use a microcontroller, then once calibrated you can save the setting in flash memory ( most microcontrollers have a a few dozen bytes of flash on board ). Programming is the second hardest part, although I might do something similar for myself in the next few months ( i.e. a variable color temperature light source ). So once written I could more or less leverage this code for other projects.

OMFG, I agree with everything that Uda said.

I'm just going to go away and cry in a dark corner somewhere.

:rofl::lol::rofl:

 

[ddrueding]

Microcontroller indeed. I've been in charge of a massive LiteTouch installation for a while now, though I haven't gone to their 3-day training in London.

I totally agree that, even for a multipurpose room, more than 4 settings would be crazy. At the same time, programming a bunch of stuff would be a PITA on it's own. I was actually thinking about having 4 buttons, then having a bank of dimmers in a closet for each of the 4 buttons. Press a button, activate the settings on that bank of dimmers. One of the dimmers could control each of the different accent colors for management of that aspect as well.

The tricky bit would be if I wanted those 4 buttons at three different locations...

 

[Stereodude]

I think you're wasting your time and energy. Unless you're planning to do a complete DIY LED lighting setup just do the normal 110V lighting fixtures. 99.9999% of people have those in their house, so the LED lighting fixtures that will come out will be designed for the existing 110V household lighting.

 

[Bozo]

No matter what type of lighting you install, you might consider installing a relay panel. Every outlet and lighting box is individually wired back to a central location. All the switches and dimmers are low voltage and also individually return to the relay box. The relay panel also contains a programable controller with solid state relays to control the lights. You can program what switch controls what light or receptical you like, or multiples of them.
The beauty of this setup is you can change the voltage at individual devices at any time. Need 12VDC for a single lamp? Just supply the appropriate relay with 12VDC. Done! No rewiring or trying to separate circuts.
You can also program the controler to have the lights turn on and off automatically.
I haven't looked into the hardware for this type of setup in many years, but I'm sure you can find it online or at a well stocked lighting store.

 

[ddrueding]

No matter what type of lighting you install, you might consider installing a relay panel. Every outlet and lighting box is individually wired back to a central location. All the switches and dimmers are low voltage and also individually return to the relay box. The relay panel also contains a programable controller with solid state relays to control the lights. You can program what switch controls what light or receptical you like, or multiples of them.
The beauty of this setup is you can change the voltage at individual devices at any time. Need 12VDC for a single lamp? Just supply the appropriate relay with 12VDC. Done! No rewiring or trying to separate circuts.
You can also program the controler to have the lights turn on and off automatically.
I haven't looked into the hardware for this type of setup in many years, but I'm sure you can find it online or at a well stocked lighting store.

Quite right Bozo. A replay box, combined with low voltage conduit to all the boxes, allows maximum flexibility. The added cost and time is minimal if you have all the sheet rock torn out anyway.

 

[ddrueding]

Time for a revisit on this, now that I actually own a house. All thoughts are welcome and appreciated.

 

[LunarMist]

Fly JTR out there and have him design you a professional system.

 

[Stereodude]

You still have to follow the building code. And, face it... The vast majority of LED bulbs are going to end up being 120V E26 specials since that's what everyone already has sockets for in their house.

 

[ddrueding]

Trying to find the building code for low-voltage systems, and there doesn't look to be much. It doesn't even look to require a permit.

 

[CougTek]

Here, you only need a permit to play with voltage above 50V. Below that, you can do whatever you want.

 

[Stereodude]

Permit? Who bothers with those?

LED lighting isn't quite ready yet IMHO. There were quire a few name brand legitimate LED light bulbs for sale in Japan that had decent usable output on par with incandescent and CFL bulbs, but holy crap were they expensive.

The Cree XM LED might mix things up significantly. It can deliver 750 lumen with less than 7W of power.

Basically I'd do normal 120V wiring and E26 sockets use incandescent or CFL's for now and in a year or two (or more) and when LED bulbs reach the prices you want to pay start swapping them in.

 

[Sol]

My flat is mostly lit by GU10 bulbs, at 50watt each and up to 10 in a fairly small room they were obviously chosen for style more than efficiency. I replaced a bunch of them with 4 and 4.4watt LED bulbs (I'm renting so I'm hardly going to replace the fittings). There is definitely less light overall, but that's not a huge problem for me. In the bathroom the difference is hardly noticeable and my bathroom lighting now consumes 16watts instead of 200. Admittedly that cost me almost 100 pounds in LEDs but they need to be replaced an order of magnitude or 2 less often I'll take them all with me when I move.

My living room is a slightly different matter. I could get it from 500watts down to 44 and still have enough light but it would cost almost 300 pounds in LEDs so for now I just have one fitting with LEDs and use the others sparingly. (I don't really need them most of the time anyway).

So for some situations LED lighting is already a pretty solid improvement, but it's probably less of an improvement if your lighting setup was efficiently designed in the first place...

 

[ddrueding]

Yeah, one of the things I was talking with JTR about was a system that takes 120V (dimmable) in, and then dims large arrays of separate LEDs. That would allow for the individual LEDs to be just pinholes in the ceiling, giving an even distribution of light and no ugly fixtures or cans.

 

[sdbardwick]

Yeah, one of the things I was talking with JTR about was a system that takes 120V (dimmable) in, and then dims large arrays of separate LEDs. That would allow for the individual LEDs to be just pinholes in the ceiling, giving an even distribution of light and no ugly fixtures or cans.

Interesting possibilities with individual LEDs; both even coverage and a star ceiling!

 

[ddrueding]

I'm trying to decide whether that star ceiling is cool or....Vegas. But doing our bedroom would be about $4k.

 

[sdbardwick]

More Vegas than cool, except perhaps in a HT room. I just don't think it works for in general purpose rooms; I imagine that the dark ceiling would be distracting in daylight. However, depending on the amount of circuit control baked into the lighting grid, you could still do a fair emulation in a normal ceiling when ambient lighting is low.

 

[LunarMist]

I would just add some extra 120V circuits and switches, then install local lighting. In some years you may find an extreme, customized system rather obsolete or out of style.

 

[time]

Interesting possibilities with individual LEDs; both even coverage and a star ceiling!

They're not the primary light source, just decorative low-power LEDs.

 

[Bozo]

Why not leave the ceiling white? That way the LEDs become room lighting instead of special effects.

 

[time]

My flat is mostly lit by GU10 bulbs, at 50watt each and up to 10 in a fairly small room they were obviously chosen for style more than efficiency. I replaced a bunch of them with 4 and 4.4watt LED bulbs (I'm renting so I'm hardly going to replace the fittings). There is definitely less light overall, but that's not a huge problem for me.

I can buy GU10 CFL bulbs for AU$10-15 inc tax. Wattage ranges from 9 to 18. Decent 11W models have equivalent light output to cheap 50W halogens, the big difference is the spread is at least 110 degrees, so they're unsuitable for high ceilings or highlighting. Life expectancy is supposed to be 15,000 hours, but I've only managed half that before they became too dim for my liking.

At 60 lumens per watt, they're still way cheaper than LED and don't require different fixtures.

 

[Sol]

I can buy GU10 CFL bulbs for AU$10-15 inc tax. Wattage ranges from 9 to 18. Decent 11W models have equivalent light output to cheap 50W halogens, the big difference is the spread is at least 110 degrees, so they're unsuitable for high ceilings or highlighting. Life expectancy is supposed to be 15,000 hours, but I've only managed half that before they became too dim for my liking.

At 60 lumens per watt, they're still way cheaper than LED and don't require different fixtures.

Yeah, I looked at those too but all the cheaper ones I could find were no good in some of the fittings I have. In any case the LEDs cost about 3 times as much and are supposed to last 50,000 hours so it's roughly a wash price wise and the LEDs use a bit over half the power. (They're probably not as bright but they're as bright as I need)

 

[Pradeep]

If I had the walls open I'd say Cat 6 and HDMI to every room. Leave room in the conduit for more runs No basement I guess?

 

[ddrueding]

If I had the walls open I'd say Cat 6 and HDMI to every room. Leave room in the conduit for more runs No basement I guess?

CAT6 to every room, HDMI and conduit to the media rooms (no reason to put existing wiring in the conduit, that is for future expansion).

 

[Stereodude]

Multiple CAT6 runs to every room. I'd go with at least 4 to any room where you expect to have a TV / Cable / PC. I'd also run an RG-6 or two anywhere you think you might want to put a TV later.