Posted 12 July 2015, rewritten 3 April 2016
My power supply arrangements for the lighting rig have had some false starts but are now resolved and so I have condensed and rewritten this blog entry to suit
To recap from my first post on lighting, the lighting rig is about 69 inches long and has four parallel strips of LEDs. The strips are unbranded, supplied in a foil bag with a CE mark, a claim for RoHS, the WEEE symbol and something about quality management. The rig uses the whole of a 5-metre strip (cut into thirds) and a third of a second strip, making about 400 LEDs in all.
The original plan was to run the rig from the layout power supplies. The layout had about 40 VA available, which seemed “at the time” to be more than enough for two analogue controllers, ten stall motors and some LEDs.
In practice, the power requirements for the LEDs are extremely non-linear. I began by running the rig from the second channel controller in the fiddle yard. This is a Gaugemaster UD. The rig took 300 milliamps at six volts and 1.75 amps at 9 volts. I wanted to keep the controller for the rolling road in the fiddle yard, so I bought an adjustable linear power supply based on the LM317 to run the LEDs. Using this supply, the power needed for the rig varied between 0.6A at 8V and a whopping 2.5A at 9V.
This made me realise, these LEDs need a constant current power supply. This became very apparent when I tried out an old ECM Compspeed feedback controller to drive the rig - it was pretty clear, the LEDs would blow up very quickly. I had got away with the Gaugemaster UD because it is non-feedback controller and, although I don’t know much about how it works, it probably actually is a constant current source at any particular throttle settling.
I then tried a purpose-made but unbranded LED power unit from eBay. This made no claims to be a constant current supply, but it did work:
The increased brightness meant the rig was now a bit too bright - a bit like an operating theatre. So I raised the rig by shortening the two suspension chains (about three inches each), and I also pulled the rig away from the layout with some string.
The LEDs are now 34 inches from the nearest track, and 44 inches from the back of the baseboard. The illumination is very even (I suppose, with 400 unfocussed light sources it ought to be) and if I stand in front of the layout I do not cast a shadow I can see. If I put an object immediately in front of the backscene there are multiple overlapping shadows, but they are reasonably unobtrusive:
Using the unbranded power unit, the rig was drawing 3.0 A at 11.6 V. The power unit was supposedly rated 6 A at 12V, so it would have some capacity to spare for some similar LEDs above the work bench, but in practice it ran very hot indeed - too hot for me to leave it unattended. The unit came with an unfused 13A plug (which fell short of every relevant UK regulation) and I do wonder if the certification markings on the side were more ornamental than evidential. So I sent this power unit for recycling.
I finally bought a LED power supply intended for a domestic lighting installation:
This is a 36W, constant-voltage supply, and it also got murderously hot. A bit of investigation told me, these power supplies are designed to be run at up to around 80% of their rated power. My lighting rig was now drawing a full 3.2A at 12V - a little more than 100%. And so, I disconnected one of the four strips on LEDs in the rig. This has dropped the current consumption to around 2.4A (or 12 x 2.4 = 29W), which somewhat coincidentally is 80% of 36W. The LED power supply still gets warm, but bearably so.
I liked the illumination of the four strips for model-making, but the three strips are more gentle for operating. Next time, I'll try running LED strips from a generously-specified power source (perhaps, a car battery), measure the current consumption, and then buy a power unit to suit.