Cellar lighting experiments..

[uncle tom]

LED lights are a remarkable development in home lighting, and in recent years the cost has been coming down dramatically - that is, until you want to be able to dim them.

Those with built in LED drivers (typically those that are direct swap outs for standard bulbs) are mostly non-dimmable. Those that can be dimmed are much more challenging (and expensive) to set up as dimmable lights than for conventional filament bulbs.

You can get very low power LED lamps - I have used some 1W GU10 spots to reasonably good effect, but as I re-configure the lighting in my home cellar, I want to install some surface mounted lamp units that produce just enough light to be able to see what I'm doing, but with nothing to spare.

These lamp units use an internal ribbon of LEDs with a diffusing glass and have a separate LED driver module that plugs into them and hides behind the unit when installed.

These units are not officially dimmable, and even the lowest power 6W units are extremely bright. (The units I'm using have a lens that is about 4" diameter)

So, I thought, what happens if I stick a resistor between the driver module and the LED ribbon?

The driver module quotes an output voltage of 11-28V (they are designed as constant current devices) so after a quick bit of ohm's law, I tried putting a 220 ohm 5W resistor between the driver and the ribbon. Result: a very slightly dimmed light - and a very hot resistor..

Investigation revealed that with an elevated impedance on the load, the driver module's output voltage ramped up to nearly 50V - leaving my resistor dissipating over twice it's rated power.

Taking the resistance up to 440 ohms by placing resistors on both wires generated a useful soft light without glare for the cellar stairwell, and keeps the resistors power dissipation within spec. After several hours of bench testing, there is no sign of the driver module suffering from it's elevated voltage output (as it should be able to withstand an open circuit, this was to be expected)

Elsewhere in the cellar I needed a much lower light level, and this where the LED becomes a remarkable device. After experimentation, I am settling on an inline resistance of 10,000 ohms - with a power consumption of just 0.25W this little unit produces quite sufficient light to illuminate its vicinity.

At that level of resistance there is no need for 5W power resistors, but as they are cheap as chips on eBay, I am using ones that are 2W rated.

[JacobH]

Is the need for very low output LEDs to protect the wine (as opposed to just being for aesthetic purposes)? If so would there be any advance in using narrow-spectrum LEDs which do not emit any short wave / UV light?

[akzy]

One interesting thing is that LEDs in housing rings tends to be strobed at mains frequencies. LED hobbyist packs tend to be constant current. The implication is that the constant current ones tend to be higher power draw. However, with a microcontroller (This was my first FPGA project actually, please dont mock the appended code), we can achieve both dimming and less power through the wonders of pulse width modulation (PWM).

Code: Select all

//This module makes a breathing LED that goes from off to on in 16 levels of brightness. Fiddle with sclock[n] positions to find best timings. 
module breathing_led(CLOCK_5, LED);
	input CLOCK_5;
	reg LEDstate;
	reg [15:0] brightness=16'b0000000000000000;
	output wire [4:0] LED;
	wire [14:0] sclock; //11 is 1 seconds - 0 is 1/4096 seconds -  6 is 1/32 seconds
	clock1s clock1smod(CLOCK_5,sclock); //Pll module
	assign LED[3:0] = sclock[14:11]; //shows time
	assign LED[4]=LEDstate; //breathing LED
	always @ (posedge sclock[0])
	begin
		LEDstate<=(brightness>>sclock[3:0]) & 1; //reads the brightness level
	end	
	always @ (posedge sclock[5])
	begin 
		if (sclock[11]==1'b1)
		begin
			brightness<=(brightness<<1) | 1'b1; //bit shift
		end 
		else 
		begin
			brightness<=(brightness>>1);
		end
	end	
endmodule

In simple terms, if we turn the light on for 1 ms then off for 1 ms {0,1,0,1,0,1,0,1...} and repeat over a long time scale {0,1,0,1,0,1,0,1...}, we will see something that is effectively half brightness. For 25% brightness you might do on for 1ms and off for 3ms {1,0,0,0,1,0,0,0...}. This could also do the trick.


As for the wavelength of light, the power of UV will be very little from these LEDs. Additionally, UV really struggles to permeate through glass (I always like to remember by not getting a tan through the car window). If you're hyper paranoid, go for light on the redder side or add a <400nm filter which you can buy film of dirt cheap. Whilst I was looking this up, I found an excellent summary of UV and wine interactions.

Best of luck! Let us know how it goes!

[uncle tom]

Is the need for very low output LEDs to protect the wine (as opposed to just being for aesthetic purposes)? If so would there be any advance in using narrow-spectrum LEDs which do not emit any short wave / UV light?

My general working assumption is that light exposure should be reduced to the lowest practicable level. UV should (in theory..) be arrested by the glass of even a clear bottle, so I assume the hazard lies in the visible spectrum.

Aesthetically, the units I bought emit a 'neutral' 4000-4200K light. I thought about adding a pale straw theatre gel to warm the output, but when run at low power the output is very similar to moonlight and quite attractive.

we can achieve both dimming and less power through the wonders of pulse width modulation (PWM).

All the YouTube videos on the subject talk about PWM dimming, and then go on to promote commercial units that are both expensive and also present practical wiring challenges, given that this is not a new build. It seemed worth the experimentation to try sticking a very cheap bog standard resistor in the back of them to see if that had the same effect, and I'm quite pleased with the result.

Obviously I don't have the luxury of a dimmer knob to play with, but if I did want to increase or decrease the light output, swapping the resistor for a different value is a fairly simple operation.

[PhilW]

To dim an LED you've essentially got three options.

1. Operate the LED(s) at a lower fixed current
2. Operate the LED(s) at the fixed current, but using a driver which switches it on/off quickly.
3. Operate the LED(s) at the fixed current, but turn the power to the fixed current source on/off.

Of these, (1) is what you do when using resistors and a fixed voltage in circuit design; (2) is what you do when using a microprocessor to directly control the LED using a PWM output from a pin, and (3) is what you do when you use a typical main dimmer (which essentially truncates part of the sine-wave), if the controlled circuit allows it (many do not due to capacitance and smoothing).

From your description, you're in the position where you have main already routed to the positions where you previously had a mains-input bulb, and want to replace with LEDs, hence the driver & LED-strip combo.

Considering the options:

(1) can be achieved by several options:
(a) replace the constant current device/circuit with a lower one.
(b) connect several/all your LED strips in parallel (dividing the constant current)
(c) add something in parallel to sink the current, or
(d) (as you have done) insert resistance such that the drive circuit limits current as max voltage is reached.

The last of these is the way you have gone; obviously this is running outside of intended specification for the device and means that you typically have 50V around (which can be dangerous) rather than just ~12V so do make sure you're safely insulated etc prior to inspection.

In terms of working out the resistance needed, trial and error has solved this for you; noting you're using a 10k resistor and likely max voltage output of ~50V, you're likely looking at around 5mA current, just FYI in case considering other routes.

(2) is more onerous (adding uP/PWM device plus driver yourself) unless it were available off-the-shelf and packaged at a good price. I'm assuming this isn't available at low price, though I haven't searched exhaustively (I did see one such unit at ~£50/unit which included RF-controllable dimming).

(3) is simple, and allows controllable dimming, if you can use dim-compatible drivers e.g. https://aomlighting.co.uk/LED-Lighting/ ... 5272-00-00 . I don't know how this compares to those you are using for size, space, cost and hence viability; if cheap enough this should be safe and reliable with standard main dimmer control replacing main light switch.

[uncle tom]

The last of these is the way you have gone; obviously this is running outside of intended specification for the device and means that you typically have 50V around (which can be dangerous) rather than just ~12V so do make sure you're safely insulated etc prior to inspection.

Indeed - I've neatly shielded all all bare wires with heat shrink sleeving.

Given how bright these units are in their standard format, logic would dictate a market for alternate driver modules that can be swapped over to run them at lower power, possibly with an adjuster built in; but there seems to be a complete lack of such units on the market (setting aside bare circuit board modules and units that are too big to fit as replacements) - an opportunity for someone..

[PhilW]

Given how bright these units are in their standard format, logic would dictate a market for alternate driver modules that can be swapped over to run them at lower power, possibly with an adjuster built in; but there seems to be a complete lack of such units on the market (setting aside bare circuit board modules and units that are too big to fit as replacements) - an opportunity for someone..

The most common method would be to use low-voltage wiring with no driver needed at the lamp end, with dimming at the control/supply. For retrofit I suspect the dimmable drivers such as the one I linked are probably expected to be the solution.

Note that even for retrofit, if you're prepared & able to disconnect the wiring from the mains at the source end and put a simple controllable head-end (single driver with dim control or capability) inline there before the wiring splits, you could still then use the existing wiring as a low-voltage circuit with just the leds (no drivers) at the lamp end; this would likely be simplest and cheapest method using off-shelf parts only (probably using constant voltage rather than constant current option), if you're happy/able to do this and depending on your wiring layout.