Yes, the losses are the same for the same length/width of wire, but the point is that you automatically get the 'right' amount of light whatever wire you use. That's not true with CV supplies. (I don't think tweaking the voltage output trimmer is really a relevant 'solution'. - you don't normally get a user-accessible one.). But it's true that wire losses in general for this applicatoin are not significant.

borpin. Yes my CC idea is not going with the industry flow. As JSH says, most LED 'units' are designed for a 12V (or 24V) supply. (People are used to doing thing like this - it's exactly the same as 12V halogen lighting).

To do CC lighting you normally need bare LEDs mounted on a heat-sink/optics. This ought to be cheaper and more reliable, but of course won't be (cheaper) if no-one is making these.

In practice it's more variable than that. I have a '12V' LED striplight that works just fine if you drive it with a CC supply at 700mA. So some devices will work fine driven either way. But ones with active switchers (rather than just resistors) inside may not like it.

All the LEDcentre drivers are CV. The way you can tell is that if the supply specifies a voltage (and a specified or implied current range) then it's CV. If it specifies a current (and a specified or implied voltage range) then it's CC. Simple really.

Now that is what you call 'going too far'!

Surely the point is that with a CC drive, you still only have 1 LED worth of current even if you have 26 LEDs in the circuit? So the losses won't be the same as with a CV drive. Or have I missed something?

On a slightly different aspect, I see you chaps are agreeing that a single driver for all the LEDs in a room (or at least a circuit in a room) is the best topology. But you have the driver associated with the switch. Is there any chance I could persuade you to separate the driver from the switch and have it mounted out of the way somewhere? Then the switch(es) could be low voltage either using bell wire or radio control.

<blockquote><cite>Posted By: djh</cite>Surely the point is that with a CC drive, you still only have 1 LED worth of current even if you have 26 LEDs in the circuit? So the losses won't be the same as with a CV drive. Or have I missed something?</blockquote>

It's not as big a difference as you might at first think. Take the case for three LEDS in series. If we drive them from a constant current source at, say 800 mA (assuming around 3W LEDs) and assuming we used a good switched mode 240 V AC to 800 mA DC supply, then we could probably get around 85% efficiency in practice, so a power loss of around 1.6 W.

If the same three LEDs were connected in series inside an MR16 fitting (for example) and driven from a 12 V supply with an 85% efficiency via a simple current limiting resistor in the MR16 base, then the resistor value would be around 1.5 ohms (to drop the required 1.2V at 800 mA) with an additional power loss of about 0.8W Not that big an additional loss when it means you can use off-the-shelf LEDs, light fittings and power supplies rather than custom made, DIY fittings.

<blockquote><cite>Posted By: djh</cite>On a slightly different aspect, I see you chaps are agreeing that a single driver for all the LEDs in a room (or at least a circuit in a room) is the best topology. But you have the driver associated with the switch. Is there any chance I could persuade you to separate the driver from the switch and have it mounted out of the way somewhere? Then the switch(es) could be low voltage either using bell wire or radio control.</blockquote>

My plan is to fit the power supplies after the switch, in an easily accessible space (probably the under eaves service space inside the thermal envelope in my new build). This retains the power saving feature of not having the supplies on all the time and also means that the light wiring for the wall switches will be fairly standard, meaning off the shelf wall switches can be used (mains wall switches aren't usually DC rated).

Posted By: JSHarrisIf the same three LEDs were connected in series inside an MR16 fitting

Cheat! That's avoiding the whole question. The question is whether driving a number of lamps in series using a CC supply dissipates less or more power in the cable than driving the same number of lamps in parallel from a CV supply. Claiming that three LEDs in series in a single lamp is a suitable test case for a bunch of lamps in parallel is cheeky!

Posted By: JSHarrisMy plan is to fit the power supplies after the switch, in an easily accessible space (probably the under eaves service space inside the thermal envelope in my new build). This retains the power saving feature of not having the supplies on all the time and also means that the light wiring for the wall switches will be fairly standard, meaning off the shelf wall switches can be used (mains wall switches aren't usually DC rated).

Right, I see the thinking. I want to avoid having mains running through the switches, partly because I want to be able to use battery-powered radio switches so I don't have to put wires in bale walls and I don't have to think about where to put switches (or at least I can easily add extra switches later). Hence I'd like to find a power supply that can be remotely switched and has low standby power in the off state. I guess with the radio systems, the receiver controls a switch that would switch the power supply on and off, so that wouldn't be a problem. And I guess with bell-wire or cat5 connected switches a separate relay in front of the power supply would be the answer. So I guess I don't have a problem after all

I'm working on something a bit more controllable, but undeniably more risky for my new-build.

All lighting wiring goes to a central place in the house christened "brain central" by the sparkie - all wiring is conventional 1.5mm T&E for future proofing / in case I get this wrong however I plan to run it all at low voltage.

The lights are (mostly) 12V mr-16 which I am glad are simple devices as I will connect them to a bunch of PWM dimming controllers - I would assume if there's nothing more than a resistor and maybe a bridge rectifier inside then there is not a lot to get upset about the strobing of the PWM

The PWM drivers are controlled by DMX and daisy chained with cat5 cables to an arduino which can turn channels on, off, up and down as required.

The light switches are going to be "bodged" up from a number of 16mm stainless push to make buttons drilled through a blanking plate and sat in a normal back-box - wired back to the arduino in 1.5mm cable again - the push button will effectively short ground to a signal pin to instruct the arduino to increase or reduce the appropriate DMX channel.

I did have a fancy expensive variable constant current DMX controlled driver but I struggled to find any reasonable lamps to drive from it - hence the switch to PWM and MR16s. As technology evolves I might switch out the control and lights.

The bonus of the arduino is that I can enhance it in time to allow it to be remote or internet controlled, play "occupied house" patterns when I am on holiday, only turn on lights at night at 1/4 brightness etc

Plan B, in case this doesn't work, is desk lamps everywhere and a very cross other half!

-Steve

Putting the power supplies after switches should extend their life.

Posted By: JSHarris

I've taken a few 12 V LEDs apart and all have had just a bridge rectifier and series resistors for current control. Mine all dim fine with a DIY PWM circuit.

Thant kind of makes sense, the bit that makes less sense to me is that some MR16 LEDs explicitly state "not dimmable" where as others are "dimmable" - no idea what the difference is.. I

I would have expected an LED with no/minimal electronics to be dimmable by PWM no matter what it says on the box - tt may be that the dimmable ones have MORE electronics inside them as they expect to be dimmed from a AC trailing edge wave stepped down to 12V AC - or maybe the people that make them have no idea..

-Steve

Posted By: JSHarrisIt might be cheeky, but it's the way 12V LEDs are very commonly internally wired!

I wasn't objecting to how the lamp was wired, I was saying that you'd missed the whole point of the discussion. And still have apparently.

The whole conversation was in the context of you saying "True, but as you suggest, running twenty six 5 W LEDs on a single cable run is probably a bit extreme! ...I'd guess most people wouldn't have more than maybe 10 of them on a single cable run" and wookey's "Cable resistance is largely irrelevant if you use constant-current supplies. With a 700mA supply you get 700mA down the cable however many LEDs you use - just the voltage changes, and will rise a smidgen to cover cable losses if needed. If you put in more than 12 LEDs in one run then you won't be in 'low voltage' (<50V) territory any more."

So I thought it was pretty clear the discussion was about strings of lamps rather than individual lamps, however those lamps are wired internally.

If you're only talking about off-the-shelf and you say that CC is unobtainable off-the-shelf, that makes for a rather short discussion comparing CC vs CV wiring losses!

I suspect the reason for not selling MR16 Leds designed for use with a CC driver is that they are worried about people putting them in fittings fed by a voltage source (eg AC). Result is a blown LED and a potential warranty return.

I found some other geeks doing CC LED lighting which led me to a useful supplier of PSUs: Reichelt: http://www.reichelt.de/LED-Treiber/0/16/index.html?;ACTION=2;LA=102;GROUPID=5143;SID=10TzMPcX8AAAIAAHdAcT432c133564dba033084e629b622b6776c Much better range and prices than LED-technik. Have some nice efficient meanwells at 350mA and 700mA and useful voltages - ie. just right for single-room lighting.

And off-the shelf CC bare-LED modules are starting to appear. Here's one from dealextreme: http://www.dealextreme.com/p/1w-cree-q2-110-lumen-white-led-spot-light-module-3-5v-dc-13081 for €12.4, although oddly they'll sell you a 90lm/W module _with_ a CC driver for $10, which lets you fit them just like mains lights. Now if they could just do that one _without_ the driver for €5 that'd be great. http://www.dealextreme.com/p/1w-1-led-90-lumen-ceiling-led-light-bulb-with-driver-white-220v-48095

djh - if you want to do low-voltage switching then the obvious way is on the output side. You just have to find supplies with sufficiently low quiescent currents.

No, I mean so the supply drives a loop of wire with LEDs wired in series along the loop and there is a switch in series too. When you switch it on you get light. When you switch it off the supply enters a very low power mode due to no output demand. You need two characteristics for this to work well: a) supply dropping to low-power mode when output power is zero, and b) supply soft-starting on the LV side so it doesn't blow your LEDs up on switch-on (people talked on candlepower forums of one CC supply which _must_ be switched on the HV side otherwise it _will_ nuke all your LEDs with big voltage spikes).

Posted By: CWattersA constant current source will tend to go to a high voltage if the output is open circuit bucause it's trying to push a constant current through the high resistance of the open switch. How high depends on the max number of LEDs it's designed to supply and it's internal design.

I've seen this with the CC driver I played with and wondered how quickly that voltage actually dropped off once you closed the switch - it felt a "wrong" thing to do

Posted By: wookeyI don't like those string things

Under the 17th edition wiring regs you can have normal light switches in a bathroom, provided they are outside the 'zones'. That just means putting the switch at least 0.6m from the edge of the bath or shower basin, or 1.2m from a wet room drain.

Nevertheless, many people still feel happier with a pull-cord switch.