Dealing with Loss of RF Signal in Dead-Rail for Onboard DCC Decoders

Note: This post deals with details of various brands of DCC-compatible, wireless RF receivers operating in the 902-928 MHz “ISM” band that connect to onboard DCC decoders. Some aspects of the discussion may apply to other RF bands as well.

Typical application. In some cases, such as the Airwire transmitters, the throttle and transmitter are combined. Also, the receiver and amplifier may combined, such as for Airwire and Tam Valley Depot receivers.

The designers of various DCC-compatible RF receivers have a couple of strategies for what output to provide to the onboard DCC decoders when a valid RF signal is lost:

  1. Output the random pulses that the RF receiver naturally outputs when a valid RF signal is lost. This option will cause most DCC decoders to maintain direction and speed while the DCC decoder “sifts” the random pulses searching for valid DCC packets.
  2. Output a fixed, positive Direct Current (DC) voltage to one of the DCC decoder’s “Track” inputs and a zero voltage DC the other “Track” input when either a) RF signal is lost, or b) when the RF transmitter does not send sufficiently-frequent “keep-alive” DCC packets. The latter is true for the Airwire CONVRTR. How the DCC decoder responds to these DC “Track” inputs depends upon DCC decoder configuration and, unfortunately, DCC decoder manufacturer discretion.

There are several NMRA-specified Configuration Variables (CV’s) that affect how DCC decoders handle the loss of valid DCC packets and are important to understand when the DCC decoder is connected to the DCC output of DCC-compatible RF transmitters because the RF receivers may lose or receive corrupted RF signal from the dead-rail RF transmitter.

The NMRA standard S-9.2.4, section C “Occurrence of Error Conditions” states “Multi Function Digital Decoder shall have a Packet Update time-out value.” Further down on line 60 the standard states “A value of 0 disables the time-out (i.e., the user has chosen not to have a time-out)”. This part of the NMRA standard is not universally-implemented by manufacturers, and it affects how decoders will respond to the loss of RF transmission of DCC packets. To implement this requirement, the NMRA standard S-9.2.2 has defined the recommended (R), but not mandatory (M), CV11, Packet Time-Out Value. A value of CV11=0 is defined to turn off the time-out, but CV11 is frequently not implemented.

However, another CV that is often implemented addresses some aspects of loss of DCC. The optional (O) CV27, Decoder Automatic Stopping Configuration, is under re-evaluation by NMRA, but the NMRA has taken no definite action some time. Here is what the NMRA standard S-9.2.2 currently (as of 2019) states about CV27: 

Configuration Variable 27 Decoder Automatic Stopping Configuration
Used to configure which actions will cause the decoder to automatically stop.

Bit 0 = Enable/Disable Auto Stop in the presence of an asymmetrical DCC signal which is more positive on the right rail.
“0” = Disabled “1” = Enabled

Bit 1 = Enable/Disable Auto Stop in the presence of an asymmetrical DCC signal which is more positive on the left rail.
“0” = Disabled “1” = Enabled

Bit 2 = Enable/Disable Auto Stop in the presence of an Signal Controlled Influence cutout signal.
“0” = Disabled “1” = Enabled

Bit 3 = Reserved for Future Use.

Bit 4 = Enable/Disable Auto Stop in the presence of reverse polarity DC.
“0” = Disabled “1” = Enabled

Bit 5 = Enable/Disable Auto Stop in the presence forward polarity DC.
“0” = Disabled “1” = Enabled

Bits 6-7 = Reserved for future use.

Since DCC decoder manufacturers frequently do implement CV27, what electrical output the DCC-compatible RF receiver provides to the DCC decoder upon loss of a valid RF signal will influence how the DCC decoder responds. We will break this down for various brands of DCC-compatible RF receivers in the 902-928 MHz ISM band in the following subsections.

Note that some DCC decoders will not honor CV27=0; i.e., all auto-stopping features disabled. For example, with CV27 set to 0, the Zimo MX-696, and probably other Zimo DCC decoders as well, will continue speed and forward direction if positive DC level is input to the “Right Track” DCC input, and a zero DC level is input to the “Left Track” DCC input. Under these “track voltage” conditions, the locomotive will stop if originally moving backward. Some (but not all) DCC-compatible RF receivers, such as the Airwire CONVRTR, provide these DC inputs, if a valid RF signal is lost, but only if connected correctly.

The “correct” connection relates to how the user connects the DCC output from the RF receiver to the “Track Right” and “Track Left” inputs of the DCC decoder. Under normal circumstances, when there is a valid RF signal, which way the DCC decoder connects to the RF receiver does not matter. Under the exceptional case of DC-only output by the RF receiver, if it loses a valid RF signal, which way the DCC decoder connects to the RF transmitter does matter. The user will likely want the locomotive to continue forward with the loss of a valid RF signal, so some experimentation is required to determine which of the RF transmitter DCC outputs should connect to which of the DCC decoder’s “Track” inputs to achieve the desired behavior.

Example DCC waveform output from a DCC-compatible RF receiver when there is a valid RF signal
Example random pulse output from a DCC-compatible RF receiver when there is no valid RF signal. Note the waveform’s superficial similarity to valid DCC output.

As a further complication, the user should probably turn off the decoder’s “analog” mode of operation by setting Bit 2 of CV29 to 0 to force the decoder to use “NMRA Digital Only” control of ”Power Source Conversion” (see the NMRA standard here). If Bit 2 of CV29 is set to 1, and again we emphasize the user should probably not activate this feature, then “Power Source Conversion Enabled” and then CV12 determines the power source; the most common of which is CV12=1, “Analog Power Conversion.”

Airwire CONVRTR Series

CVP Airwire CONVRTR-60X tender installation. The CONVRTR operates on Airwire channels 0-16. Note that the U.FL antenna lead was later connected to the CONVRTR. The LokSound L V4.0 DCC decoder mounting harness can be seen mounted on the tender wall opposite the CONVRTR, and its Track Left/Right inputs are connected to the CONVRTR-60X’s DCC A/B outputs.

When the CVP Airwire CONVRTR loses a valid RF signal or receives insufficiently-frequent DCC Idle packets, it detects these conditions and outputs a fixed DC voltage to the decoder. Consequently, the user should set CV27 according to the description above.

While it may seem that the user would want the locomotive to stop if its RF receiver loses a valid RF signal, consider what might happen in tunnels or locations remote to the DCC RF transmitter. Getting stuck under these circumstances if a valid RF signal is lost is probably not what the user wants, so we strongly suggest that the user set CV27=0.

The user is cautioned, however, that some DCC decoders, such as the new ESU LokSound 5 L DCC, do not honor the CV27=0 setting unless the “polarity” of the “Track Right/Left” is connected “correctly” to the CONVRTR’s “A/B” output. Experimentation may be required to determine the correct connection, but my experience is: CONVRTR A <–> Decoder Track Right & CONVRTR B <–> Decoder Track Left

QSI Solutions Gwire and Tam Valley Depot DRS1 Series

The QSI Solutions GWire operates on Airwire Channels 0-7. If the U.FL plug (at the upper-left corner of the Linx Transceiver chip) connects to an externally-mounted antenna, the antenna wire at the upper-left corner of the GWire board should be cut off at board level, or better yet, unsoldered.
The Tam Valley Depot DRS1, MKIII, operates on Airwire Channel 16
The Tam Valley Depot DRS1, MkIV, operates on Airwire Channels 0-16 (as well as other frequencies). Note the internal antenna on the right-hand side of the board.

The QSI Solutions Gwire and Tam Valley Depot DRS1, MkIII and MkIV DCC-compatible RF receivers will output random pulses to the onboard DCC decoder when a valid RF signal is lost, so setting CV27 is probably of no use. On the “plus” side, most DCC decoders will maintain locomotive direction and speed in the presence of these random pulses since the DCC decoder is actively sorting through these pulses for valid DCC packets, which is usually the behavior the user wants.

A Blueridge Engineering webpage describes how to easily modify the GWire for use as an RF receiver for any onboard DCC decoder.

OScaleDeadRail ProMiniAir Receiver

OScaleDeadRail ProMiniAir receiver operates on Airwire channels 0-16. The ProMiniAir can also be configured to operate as a DCC-compatible transmitter that wirelessly transmits DCC from any DCC source on Airwire channels 0-16.

The OScaleDeadRail ProMiniAir receiver has a default long address of 9001. Like the ProMiniAir transmitter, the ProMiniAir receiver’s channel can be reset in “OPS Mode” by setting CV255 to a value in the range of 0–16. The ProMiniAir receiver has the following options when a valid RF signal is lost:

  • Output random pulses to the onboard DCC decoder: The user can set the ProMiniAir receiver to output the random pulses when it loses a valid RF signal by setting CV246 to 0 in “OPS mode” at the ProMiniAir’s address. In this case, setting CV27 for the onboard DCC decoder is irrelevant because the random pulses from the ProMiniAir receiver will cause the onboard DCC decoder to maintain the speed and direction of the locomotive while it is “sifting” through the random pulses for valid DCC packets.
  • Output either fixed positive or negative voltage DC to the onboard DCC decoder: In this case, setting CV27 for the onboard DCC decoder at its address is relevant. The user can set the ProMiniAir receiver to output fixed DC voltage when it loses a valid RF signal by setting CV246 to 1 in “OPS mode” at the ProMiniAir’s address. A positive DC voltage is output by setting the ProMiniAir receiver’s CV248 to 1 in “OPS mode” at the ProMiniAir’s address, or a negative DC voltage is output by setting CV248 to 0. If the user does not want the locomotive to stop with the loss of a valid RF signal, then set CV27=0 for the onboard DCC decoder at its address. Of course, setting CV27 to other values (see above) in the DCC decoder will determine how the DCC decoder responds to the fixed DC voltage that the ProMiniAir outputs to the onboard DCC decoder upon loss of a valid RF signal.

Wrap-Up

It’s an unfortunate fact of life that we can lose a valid RF signal from our DCC-compatible transmitter. However, with a little study of DCC decoder documentation, and possibly a bit of experimentation, gracefully coping is definitely possible.

Retro-Fitting Smoke Units with Thermistors and Set-up for LokSound Decoders

This is a slight modification of a post titled: O Gauge Forum Post on Smoke Units.

It is possible to modify a non-ESU smoke unit to connect to the LokSound L or XL decoders just as an ESU smoke unit does by connecting the smoke unit to the specialized ESU smoke unit terminals: HTR+/-, MOT+/-, and TMP+/-. This capability allows you to take direct advantage of all the LokSound capabilities provided for ESU smoke units. The missing component in some smoke units is a Negative Temperature Coefficient (NTC) thermistor.

What started me down this road was a “dead rail” conversion of a Sunset 3rd Rail Big Boy (3-rail, “Late Version”) originally outfitted with TMCC and a nice Lionel smoke unit with dual output (photo below).

Dual stack, fan-driven smoke unit for retrofit.

I wanted to retain this beauty and use a LokSound L V4.0 decoder that is controlled by an Airwire CONVRTR-60. THOR73’s posts on the O Gauge Form inspired me to work through using this smoke unit with the LokSound L V4.0 decoder. I thought that if I could figure out how the ESU smoke units created their “temperature” inputs to the LokSound decoder, then I could retrofit the Lionel smoke unit so that it would be “input compatible” with an ESU smoke unit. This retrofit turned out to be simple.

I reverse-engineered an ESU 54678 smoke unit by measuring the resistance between the heater resistor leads (HTR+/-): ~23 ohms; motor leads (MOT+/-): ~16 ohms; and thermistor leads (TMP+/-): ~100K ohm at room temperature. Each of these components is electrically isolated from the others. When powered by a 14.8V LiPo battery, the LokSound L V4.0 decoder I had on hand produced the following results on the ESU Profi board using the LokProgrammer (with ground measured at the Profi board’s ground terminal):

Terminal Smoke off Smoke on (Throttle=10)
HTR+ (not connected to heater resistor*) 13.4V 13.2V
HTR- Open Switched open/ground @500Hz ~30% duty-cycle PWM
Fan+ 0V Pulsed <= 5V (difficult to determine with low Frequency chuffs)
Fan- 0V 0V
TMP+ 5.1V 5.1V
TMP- 1.3V 3.7V
* Battery+ (14.8V) connected to heater resistor + input

Cut-away of a 54678 ESU smoke unit showing the thermistor

The difference in TMP- between unheated and heated conditions suggests, but does not prove, that the thermistor’s decrease in resistance with increased temperature is manifested by a voltage increase at TMP- as part of a voltage divider where the thermistor is in series with a fixed resistor resident in the decoder, possibly with a low-side voltage offset:

Probably thermistor circuit

Guesses: RFIXED~1.5K based on probe measurements and derived @ 25C Voffset~1.24

So right off the bat, the ESU smoke unit’s heater resistance (23 ohms) is similar to Lionel’s (27 ohms), and both smoke units use 5V fan motors. The Lionel was missing only the thermistor. Lower resistance smoke units (around 8 ohms) might be problematic to convert unless retrofitted with a heater resistor in the 20-ohm neighborhood or use an externally-supplied, lower HTR+ voltage. The heater and fan motor similarity between the ESU and Lionel smoke units made this particular Lionel smoke unit an excellent surrogate candidate.

Thermistors with 100K ohm resistance at 25 Celsius are commonly-available, usually with a “B” parameter of around 3900 Kelvin. You can Google what this parameter means (simplified Steinhart-Hart Equation: R(T in Kelvin)=R@TRef*(exp(B/T-B/TRef)) ). While I could not verify that the ESU smoke unit used precisely this type of thermistor, the testing described later supports this selection.

The photo below is the Lionel 27 ohm smoke unit PCB, part #610-PCB1-045, Rev C (Lionel replacement part #691PCB1045), that was retrofitted with an “axial,” glass-coated 100K NTC thermistor with a B of 3892 Kelvin. (Well, it’s a Lionel replacement PCB since I cut some traces retrofitting on the original PCB that I regret doing. Interestingly, the original PCB did not have the mangled lettering of the replacement PCB that some have noted.)

The 3-pin power plug on the PCB can power the heater resistor since the outputs from the rectifier/5V converter do not connect to anything after removing the fan motor plug. The ground on the PCB MUST be isolated from the heater unit metal case since the PCB’s “ground” wire will be connected to the LokSound L’s HTR- terminal that regulates the heating resistor’s current path to the electrical ground! Electrical measurements revealed good electrical isolation of the metal case from the heating element.

Lionel fan-driven smoke unit with a 100K thermistor added.

I drilled two holes in the smoke unit’s PCB board, and the thermistor was inserted and soldered to two-wire leads that connect to the LokSound L’s TMP+/- terminals. I used high melting-point solder because conventional solder might melt at the high operating temperatures of the heater resistor and thermistor (max around 250 Celsius according to documentation for the ESU smoke unit).

The two heater wires from the three-pin PCB plug connect to the LokSound L’s HTR+/- terminals. (Pins 1 and 3 are shorted together on the PCB and connect to one side of the heater resistor; pin 2 is ground and connects to the other side of the heater resistor.)

The motor wires directly connect to the MOT+/- terminals. Out of sheer luck, when the red motor lead from the smoke unit is connected to MOT+, and its black lead to MOT-, the fan motor spins in the “correct” direction.

As others have suggested, I replaced the original 27-ohm ceramic resistor with a Lionel 27-ohm replacement #6008141055.

Once you connect the smoke unit’s six outputs to the LokSound L’s ESU smoke unit terminals, some modifications are needed in the ESU sound files and decoder setup since they did not originally activate the ESU smoke unit. First, follow THOR73’s directions regarding the connection between sound and smoke chuffing under the “Smoke unit” menu. The smoke unit’s automatic power-off time should be reset since the default is 0 seconds. I don’t know if 0 means never turning off, but a non-zero setting seemed like a good idea.

LokSound change decoder settings menu

What differs from THOR73’s discussion is the sound file setup for an ESU smoke unit. Editing the sound files reveals that most “nodes” have the option to set the “ESU Smoke Unit” parameters. Frequently these settings are turned off, but there are some useful “presets” you can select and experiment with. An especially interesting preset is the “preheating” preset available in the stopped state.

Mute State:

LokSound Stop state settings.

Here are the other states I modified, but I am no expert or knowledgeable about these settings. Usually, I chose a “Preset” and then selected the “Steam Chuff” checkbox, which preserves the parameters of the preset (unless you change them), but turns off the Preset name.

Stop State:

LokSound Mute state settings

DCX State:

LokSound DCX state settings

Coast State:

After editing these sound nodes, the next step is to set an “F#” to turn the smoke unit on/off on the “Function mappings” menu. The “logical” outputs column provides an “ESU Smoke Unit” selection, so I selected F23 as the ESU Smoke Unit on/off toggle.

TESTING WARNING: The ESU 53900 Profi Decoder Tester does not appear to provide adequate power to an actual ESU Smoke Unit or surrogates described here! In deployed operation, the LokSound L is perfectly capable of delivering sufficient power, but in my experience (or inexperience), the Profi board is not able to do so. I initially thought the culprit was the puny AC to DC converter provided to power the Profi board. But, the power connection to a very hefty 14.8V LiPo battery did not solve the problem. The workaround uses either THOR73’s high-side MOSFET switch mentioned in this thread or the low-side MOSFET switch described in the same topic thread. Either way, you will need to take power (about +14V DC) from the source providing power to the Profi board and use the Profi board’s HTR- output to control the MOSFET switch. In turn, this switch controls the smoke unit’s heater. Using THOR73’s high-side FET switch, you connect the smoke heater as he describes. If you use the low-side FET switch, I presented, the smoke unit’s HTR- output connects to the switch control input, and the switch’s ground connects to the power ground. 

Reiterating, YOU ONLY NEED THIS SPECIALIZED MOSFET SWITCH FOR TESTING WITH THE Profi BOARD! In actual operation, the LokSound L adequately powers an ESU smoke unit by direct connection to the decoder’s ESU smoke unit terminals, as is the modified smoke unit described here. 

Here’s the “proof in the pudding” video:

Surrogate smoke unit in action

Please forgive the disassembled state. I haven’t finished the dead rail conversion, but this video does demonstrate battery power with the LokSound L V4.0 controlled by an Airwire CONVRTR-60 wireless receiver.

Follow up

To be pretty linear, my guesses on RFIXED and Voffset are 1.5K and 1.24V. The 1.5K came from an “off” measurement of resistance between decoder GROUND and TMP-, which is fraught with potential error.

These values will give you the following approximate curves: the left axis is the voltage at TMP-, and the right axis is the estimated thermistor resistance.

BUT THE VALUES of Rfixed AND Voffset ARE ONLY ENGINEERING JUDGEMENT GUESSES!

Example thermistor T versus V and T versus R curves