Atlas narrow-hood light-board with the E24 DCC connector decoded

[peteski]

That new board is a bit of a mystery. It has two unused 4-pin connectors with very little information available about their purpose. I don't own any models with that board, and contacting Atlas to get some details about the electrical diagram did not yield any info (no reply from Atlas). Also, until recently I had no luck trying to borrow the board from TRW members until Benjamin volunteered to send me that board removed from one of his models.   Thank you @GM50 4164  on behalf of all the TRW members looking for this info. With the board in hand, last Sunday I was able to decipher all the connections. There are some legends silk-screened by each connector, but they aren't very helpful in determined what AUX outputs they connect to.


AtlasE24boardConnectors.png

The diagram is pretty much self-explanatory.  Each of the unused connectors provides access to three AUX outputs (with 1.5k ohm resistor in series) and a common positive (U+) or a "blue" connection.

For those interested in more details, here's a schematic diagram.


AtlasE24boardSchematic.png

All the AUX outputs are powered (no logic level AUX outputs were made accessible), and all those AUX outputs include a series-connected 1.5k Ohm resistor.  The resistor is on the negative leg of the circuit. The only exception is the hardwired rear headlight LED where the resistor is  on the positive leg. That is inconsequential since the headlights are hardwired directly on the light board.  Also curious fact is that there are 1µF capacitors wired in parallel with each LED to be connected to AUX output.  I have never seen any decoder manufacturer recommend adding such capacitor. Maybe Atlas' circuit designers thought it was needed to smooth out the PWM pulses used to dim the LEDs, or to reduce any electrical noise.  Those caps should not have any detrimental effect on the lighting functions.

I also  noticed that there is a RFI (Radio Frequency Interference) suppression filter included in the motor circuit. It consists of two 100µH inductors and a 0.01µF capacitor.  Usually decoder manufacturers recommend removing such filters often included in many models, but here the circuit designer  obviously chose specific values of those components not to interfere with the decoder's BEMF functionality.

I was  also asked if the ground/common (GND) connection was available on the Atlas board, which along with the U+ raw rectifier voltage could be utilized for connecting a keep-alive circuit.  Unfortunately, Atlas did not have enough foresight to include a GND solder pad on the decoder.  It is too bad since GND is available on the E24 connector and only some minor design change of the PC board (with no additional cost) would have been needed to provide that.

As shown in the closeup photo of the E24 connector, 2 GND pins are there, but the solder pads are too short to be usable for soldering a GND wire at that location.  Even if someone was able to solder a very small wire to one of the connector's GND pins, that would not work.  The mating connector  on the decoder is wider than the one on this board, so when it's seated on the Atlas board, it leaves no clearance under it for any additional soldered wire.

If someone really wants to add a keep-alive circuit, the ground connection can be made directly to the decoder.  The positive lead of a keep-alive will be attached to any of the U+ locations on the light-board, and the negative lead to any of the GND locations shown below.


LokSound5nanoGrounds.png

Of course soldering wires directly to the decoder  components will void the warranty. Also five of the components I showed (on the left) are MLCCs (Multi-Layer Ceramic Capacitors).  Those are made from a sandwich of hundreds of metalized ceramic leaves fused together, and they are sensitive to thermal shock which can cause microscopic cracking.  Thermal shock can occur when the caps are rapidly heated with a soldering iron from room temperature to solder-melting temperature.  Damage is not guaranteed to happen, but I figured out that should give a warning.

Since no logic-level AUX outputs are accessible , the ESU 3-wire Power Pack keep alive cannot be utilized, since the Power Pack's "control" lead (white wire) has to be connected to a logic-level AUX output.  Only generic 2-wire keep-alives or simple capacitor circuits can be used (and those might affect programing of the decoder)

Last thing is what is needed to mate with those unused connectors to be able to connect more LEDs.  Those on-board connectors are JST 0.8mm spacing male headers.  There are mating female sockets are available as cable assemblies.  They come in different lengths of wire with a female socket on both ends.  Different lengths are available.  A link to a 12" cable at Digikey:  455-2994-ND.  Manufacturer's part# is A04SUR04SUR32W305B is someone wants to find another source.  Basically to make that cable assy. usable, cut the wires half way between the connectors. That will yield 2 connectors with 6" wire pigtails.

Happy Independence day to our U.S. members!

[tehachapifan]

Great info!! Thanks!!

[GM50 4164]

Again, excellent write up on the info we all can use Peter. I just purchased several of the connectors for future use!

[kiwi_bnsf]

Thanks for the super-detailed write up of this board 

Unfortunately the lack of GND on loco PC Boards with E24 connectors is becoming a bit of an unwelcome trend. This is extremely frustrating as soldering directly to a Loksound Nano Decoder is beyond my abilities 

Having got used to adding Keep Alive Capacitors via the U+ and GND pads on Loksound Micros, or via pads on host PC boards, I'm not willing to install sound without a capacitor — it's just too unreliable without one.


The new ScaleTrains N scale Dash 9s and AC4400CWs are also equipped with an E24 board that omits the GND pins.  I have emailed some pretty stern feedback to ScaleTrains as they do include a nice GND and U+ pad on their E24 boards in the C39-8 and SD40-2s. It's only the latest locos that omit them. It's clearly not an issue of cost, because adding a couple of traces and pads is not even cents per PCB.

I would encourage others to email Atlas and Scale Trains with feedback about the need for GND and U+ pins, or to add at least 900-1000 microFarads of capacitance to their boards.

[peteski]

Thanks for the super-detailed write up of this board 

Glad to provide that info to everybody reading my posts.

It's clearly not an issue of cost, because adding a couple of traces and pads is not even cents per PCB.

Exactly, since etching PC boards is a subtractive process, starting with copper clad board where all the unwanted copper is chemically etched away, there should be zero cost to include a trace from the GND pins at the E24 connector to a copper pad. Unless the PC Board design software charges per trace, which I highly doubt.

I suspect that the employees of model manufacturers simply don't think that are modellers out there who open the models and modify them (like adding keep-alive circuits).  They probably assume that modelers just buy the models and run them unmodified. Which is like true for majority of the model purchasers, but it would still be nice if they also catered to the minority that modifies their models, especially in this instance.

The other thing is that these boards are likely designed by people who aren't really familiar with the inner workings of DCC. They just get some basic electrical  specs (probably from the headquarters in the U.S.) and just draw up the design without any other considerations.  Just speculating here. . .

[GM50 4164]

Got the connectors from Digikey and wired a TCS KA-N1 to my SD7. It has definitely increased performance!