![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Out of the darkness
Jun. 27th, 2009 10:54 pm
... brighter than a thousand suns - or at least, as bright as a 400W metal halide lamp.
For about the last 3 weeks, as well as other ongoing tasks, I have been working long hours on designing parts of the first prototype of a luminaire that is intended to be as bright as a 400W metal halide lamp, though the one model we tested only consumed 150W of electricity. Photometry, to be carried out by the prospective end user, will discover if the intended light output is reached, but it is very likely, because the overall designers of the luminaire have done the maths, and consulted optical designers.
It was too bright to look at, even wearing sunglasses.
There are seven LED printed circuits, each with two series strings of 12 while LEDs, independently dimmable. The LED strip is made on a type of printed circuit consisting of a thick aluminium substrate, with a very thin insulating layer, of a material chosen to be thermally conductive, and with the copper tracks etched on top of that. There is only one layer of connections, which makes the design difficult, but this composite material is needed to remove the heat from the LEDs. They are specified to lose only 20% of their light output over 60,000 hours if run at the maximum specified temperature, but above that level, the reliability declines rapidly.
Energy savings alone might not render this product cost effective, but when the close to zero maintenance costs are included, it should be very good in its intended us (initially fuel station canopies). In real use, the lighting is likely to last until the building is subject to major maintenance or remodeling.
This was an interesting project to work on, but had a difficult deadline, because the requirement to deliver all the electronics for three units was brought forward by 6 weeks. I predicted and avoided one electrical problem associated with the unusual PCB material for the LED strips, but had to retrofit extra parts to correct for another one (very high capacitance between LED and substrate). It was a first prototype.
The (switchmode) electronics to dim and control the current in the LED strips is on the same PCB as the LEDs, but is very constrained because most of the area is covered with reflectors, so that parts more than 3mm thick can only be placed in an 8mm strip at one end of the long narrow pcb. It is for this reason that the capacitance between the LEDs and the substrate was an issue. Had it been possible to fit the inductors in the ideal place, the capacitance would have been irrelevant.
The control and comms electronics is on a 21mm wide pcb that plugs into all the LED strips and carries power and control signals from the processor to control brightness, measure temperature, and so on. That contains, among other things, a memory chip carrier 19mm wide, so I had to design it with four layers. Even the on-board switchmode regulator worked cleanly, which is good because there was no time to make a revised control pcb.
It is done for now, finished yesterday evening, and I will not know if the project is to proceed for about a month. Now to catch up with the major backlog of other tasks. If anyone is expecting any kind of response from me, and I don't catch up soon, please remind me, because I have not been noting everything while concentrating on this project.
For about the last 3 weeks, as well as other ongoing tasks, I have been working long hours on designing parts of the first prototype of a luminaire that is intended to be as bright as a 400W metal halide lamp, though the one model we tested only consumed 150W of electricity. Photometry, to be carried out by the prospective end user, will discover if the intended light output is reached, but it is very likely, because the overall designers of the luminaire have done the maths, and consulted optical designers.
It was too bright to look at, even wearing sunglasses.
There are seven LED printed circuits, each with two series strings of 12 while LEDs, independently dimmable. The LED strip is made on a type of printed circuit consisting of a thick aluminium substrate, with a very thin insulating layer, of a material chosen to be thermally conductive, and with the copper tracks etched on top of that. There is only one layer of connections, which makes the design difficult, but this composite material is needed to remove the heat from the LEDs. They are specified to lose only 20% of their light output over 60,000 hours if run at the maximum specified temperature, but above that level, the reliability declines rapidly.
Energy savings alone might not render this product cost effective, but when the close to zero maintenance costs are included, it should be very good in its intended us (initially fuel station canopies). In real use, the lighting is likely to last until the building is subject to major maintenance or remodeling.
This was an interesting project to work on, but had a difficult deadline, because the requirement to deliver all the electronics for three units was brought forward by 6 weeks. I predicted and avoided one electrical problem associated with the unusual PCB material for the LED strips, but had to retrofit extra parts to correct for another one (very high capacitance between LED and substrate). It was a first prototype.
The (switchmode) electronics to dim and control the current in the LED strips is on the same PCB as the LEDs, but is very constrained because most of the area is covered with reflectors, so that parts more than 3mm thick can only be placed in an 8mm strip at one end of the long narrow pcb. It is for this reason that the capacitance between the LEDs and the substrate was an issue. Had it been possible to fit the inductors in the ideal place, the capacitance would have been irrelevant.
The control and comms electronics is on a 21mm wide pcb that plugs into all the LED strips and carries power and control signals from the processor to control brightness, measure temperature, and so on. That contains, among other things, a memory chip carrier 19mm wide, so I had to design it with four layers. Even the on-board switchmode regulator worked cleanly, which is good because there was no time to make a revised control pcb.
It is done for now, finished yesterday evening, and I will not know if the project is to proceed for about a month. Now to catch up with the major backlog of other tasks. If anyone is expecting any kind of response from me, and I don't catch up soon, please remind me, because I have not been noting everything while concentrating on this project.
