15/02/202020/02/2020.

Separate light control by function decoder

The trick (see page 2) of using a reed switch to turn off the red front lights is insufficient to encompass all possible scenarios. When the railcar travels with a trailer, this solution is sufficient, because the train cannot travel with the trailer forward.

But, in the case of a multi-unit, the “Picasso” can be at the rear, and must in this case turn off its head lights and turn on its tail lights. However, this remains impossible in the current configuration. Note that the ABJ 3 Electrotren it runs with doesn’t have this problem, because I have equipped it with a Lenz Silver+ decoder which allows to control all the lights independently.

Impossibility of multi-unit

The only valid solution is to use a function decoder. But, for that, we’ll have to modify the LED circuits.

What got me started was the work of a member of the Loco-Revue forum, Alpiliguri, who fitted his analogue Picasso with a classic decoder. To do so, he identified all the useful connection points (rails, motor, lights, etc.) and connected everything to this decoder.

The problem

My problem is different: my railcar is already digitized and sounded. Only the lights are a problem for me, and, to a lesser extent, the impossibility of function mapping. I also had another complaint: that of not being able to do a “MU”. However, my new Lenz LH101 handheld offers an alternative to obtain this operation, without going through the NMRA method.

I will therefore keep all the current functions, except the light control. The trouble is, I won’t be able to use the function keys as I’d like to. Here is the intended assignment, corresponding to a Lenz Standard+ V2 decoder:

Key Current function Desired function Output Wire CV Value
F0 Lights White headlights A White 33 1
    White tail lights B Yellow 34 2
F5 Shunting lights Red headlights D Purple 39 8
F7 Cutting red Red tail lights C Green 41 4
F11 All white lights A+B   45 3

Modification of LED circuits

Here is the diagram giving the actual arrangement of the components. One may wonder why they chose such a complication: there is not a common conductor for all LEDs which could be connected to a decoder Vcc. I’ll therefore have to cut tracks and modify their connection, which, due to the smallness of the whole, will not be easy.

So, the track between resistors R1 and R2 will be cut, then R1 will be connected to track 3 (in orange). The two red LEDs will thus be wired in parallel and controlled via track 3. The positive commons will finally be connected (in blue).

The white LEDs are controlled by track 2, this being unchanged from the original diagram.

Modified diagram

Operations on flexible circuits

The photos (taken under a digital microscope) concern, with an exception, the front circuit. The rear circuit is identical, except for the length. Overview during work. We can see that the microscope has its weighted base facing backwards. A brown epoxy plate is inserted between the circuit and the railcar body to protect it as much as possible.

Overview

Click on the image for a closer view.

Cutting of the track with two X-Acto strokes.

Cutting the track

Elimination of the track segment by heating, which allows its separation.

Track cut out done

Connection to track number 3 using a strand of flexible wire (diameter 15/100). It remains to cut it at the level of the arrow. Note: this wire, like the following, is not insulated, but, using some precautions, this is not a problem, because the circuit is varnished.

Connection of R1 to track 3

Connection of tracks 1 and 4. To be able to solder on the tracks, we must scrape off the varnish. Note: track 1 was already cut due to the previous modification with reed switch.

Connection of tracks 1 and 44

Soldering of the wires connecting to the decoder. In accordance with the NMRA code, the positive common wire is blue. The white and red light wires are respectively white and purple. Here, the rear circuit is photographed, hence the yellow and green colours.

Soldering of wires to the decoder

The little white rectangle at the bottom left is a piece of paper that allowed me to make an almost correct white balance.

Tracking and wiring of rail connections.

It only took a few tries with a continuity tester to locate points on the PCB capable of powering the decoder.

Rail access points

Test of the modifications

It only remains to install and program the decoder. As I don’t currently have a Standard+, I test the operation with a Silver 21+ which requires the wiring of a socket made with a 1.27 mm pitch double row connector. Before connecting the wires of the right and left rails, I modify the decoder programming, which is identical to that of the Standard+, I make a first check of the lights. Then I assign it the same address as for the railcar. I connect the remaining wires. I then see that the lights are reverse in relation to the running direction. A modification of CV 29, bit 0, solves this little problem.

Lenz Silver 21+ decoder wiring test

Lenz Silver 21+ decoder wiring test

Move the cursor over the images to see the details.

Although it’s not visible here, the rear lights are off, while the front lights are red. Sounds simple, but it wasn’t with that damn proprietary card… A little regret: I can’t use the F1 function for red lights, which could then be automatically reversed, since this function is reserved for engine sound.

Decoder location

Finding a location is not easy. Obviously, there is not enough space between the electronic card and the roof. The planned decoder is 25 × 15 × 3.5 mm. Near the driver’s cab, above the luggage compartment, there is an opening which has a suitable width but which is not long enough. Through this opening pass the motor supply wires. This passage must be respected.

So, I’ll mill this opening to give it sufficient length. After removing the electronic card and clearing the cables previously mounted, I place the railcar into the milling machine’s vice, with thick cardboard jaws coming close to the handrails. There is no need to tighten very hard, since the low machining effort will be horizontal. I hatched the areas to open with a pencil: right, for the passage of the motor’s wires; left, for the decoder.

To mill the small opening on the right, I take care to move the motor connector away from the tool. It would be very annoying if it got caught in the mill bit: breaking the tool, tearing out the wires…

Starting the opening milling

The milling shavings are vacuumed off as they go, to prevent them from entering the body. Here is the opening’s final aspect.

Final aspect of the opening

Move the cursor over the images to see the details.

I try out a decoder borrowed from another locomotive. I realize that its rear, which has relatively large components (arrow), can’t be placed low enough. So, in addition, I have to mill the part of the interior fittings below, near the motor connector.

Milling the interior fittings

Now the decoder is perfectly flush with the false roof and will not interfere in any way with the position of the electronic card. In addition, its wires pass easily over the edge of the opening.

Positioning of the decoder

Installation and wiring of the decoder

I decided to use the borrowed decoder (the other machine will wait). In order not to risk interference with the original card, I did most of the programming in this machine before “transplanting” the decoder.

After unsoldering its eight-pin connector harness, I connected the light wires to it. One last try before tidying up all the stuff: it works correctly. I even allowed myself to modify the output dimming (for information: value 16 instead of 255 by default), without having thought to check beforehand that the corresponding CVs were not used in the original card… Fortunately, that was not the case…

Soldering the wires on the decoder

Decoder connection

Move the cursor over the image to see the details.

The resistor visible in front of the driver is the load supposed to replace the motor. It has been removed after the tests.

Last thing: despite some fears that I had, the roof is put back correctly…