The LED must be positioned by the front of the headlight. It is therefore necessary to start by introducing the wires. At the beginning, I twisted them, but this increases the risk of breakage. Once the LED is in its housing, there is a good chance that it will get crooked. It is convenient to place again the optics in the machining support, held with a little adhesive tape, to avoid risky handling.
As soon as the LED is nearly vertical in the headlight, put a small clamp to stick the wires on the support and to prevent them from moving, then make a light check with the diode tester by stripping the free end of the three wires. In case of breakdown, it is better to start again now than when everything is glued!
With its opening facing upwards, fill the headlight with Kristal Klear — without overflowing! Personally, I add over a thin Plexiglas small cut-out glass. This slows the drying down again: I wait until the next day to go on; it gives me time to relax…
So, the next day, I slip the wires into a telephone wire sheath that I glue under the horizontal plate of the headlight. Here, I use Uhu Strong & Safe: it has to hold! Once the glue is dry, I bend the wires at 90° upwards and I put a point of glue.
It is not useless here to redo an LED lighting test. If this test is negative, all you have to do is to cry. You may try to remove the LED of the headlight by the front, the Kristal Klear being little tough… and start all over again.
If the sheath is properly stuck on the headlight base, the assembly should take its place as before with the light guide. In order that the whole holds in the housing, I place a little Patafix into it. The sheath is also glued to the PCB: the less the wires will move, the less likely they will break.
Then, the wires are appropriately shortened — it is good to leave some slack — then stripped and soldered to the circuit. I have retained my old modified circuit as explained previously. Obviously, as there is no longer any incandescent lamp, nor LED on this circuit, it could be much simpler (see below).
In spite of my own advice: I have kept a unique resistor to feed the two headlights (but still different for the whites and the reds), again not to have to remake the circuit. Fortunately, the brightness seems fairly well-balanced.
Once the decoder is properly set, the result seems satisfactory.
Note: the position difference between the red and white source is clearly visible. Therefore, be careful to place all LEDs in the same direction.
By the way, I was just forgetting one detail: now, of course, there is no more light leaks under the chassis, nor in the cab!
Click the picture to see the etching mask to the exact scale (PDF).
Warning: I have not yet implemented this new circuit — so I don’t guarantee its accuracy. The improvements include:
Here are the references and characteristics of the LEDs used, purchased, I recall, at Electron Discount. If they are different, it might be necessary to change the resistor values. In DCC, one can always act on the output dimming.
|White for the cab||111||1206||500|
|White / red for headlights||198||0605||350 (wh.) / 100 (red)|
This sheath diameter is well suited, but it is not flexible enough for my taste. If you can find a thinner one, it will be better. For example, the Viessmann electric couplings wire sheath is very good…
Warning: allow to dry thoroughly, otherwise there is a risk of copper oxidation by moisture. Or use no-solvent Pattex 100%, almost as transparent as Kristal Klear.