DCC Blowout 95

Home Site Map
This report on the DCC Blowout 95 was written by Stan Ames.

DCC Blowout 95 Compatability Report

On February 4-5 a Digital Command Control compatibility demonstration was held at the Springfield MA train show. This demonstration was organized by the National Model Railroad Association and sponsored by the DCC manufacturers.

The purpose of the demonstration was to test manufacturers claims concerning compatibility and provide users the unique opportunity to try out the various systems in a side by side unstructured test. In time, the NMRA is expected to develop formal conformance testing. For now, the manufacturers advertised claims about compatibility are all we have to go on. Thus the motivation for the demonstration.

One should note that the first generation of DCC systems on the market were created before the standards and/or RPs were finalized. For these early systems, compatibility between systems was more important than adherence to the emerging NMRA standards and/or RPs. For example, many Lenz decoders have a "feature" that in order to emergency stop a locomotive, a stop command must follow the emergency stop command or else the locomotive will resume its last speed after a brief stop. While this "feature" is not included in the NMRA specifications, manufacturers who wish to be compatible with these decoders must implement this "patch" to hide the "feature" from end users.

As the number of manufacturers increases, and the number of different versions of the DCC software multiply, more attention is being focused on complying to the actual NMRA standard documents, as testing compatibility between all systems becomes a very difficult NxM problem. This in turn requires manufacturers of pre NMRA systems to upgrade their command station software. As an example of this is the light function. The NMRA RPs specify bit 4 of the function byte for the light function, while pre NMRA decoders in 14 speed step mode use the undefined bit in the speed and direction packet. System One decoders which properly implement this feature of the NMRA draft specifications will flicker the headlight on pre NMRA systems that fail to include the headlight information when transmitting the function packet.

The need to work out these subtle differences has and will continue to require a great deal of cooperation between all the various DCC manufacturers.

Consumers should note that compatibility and conformance are both important features of any DCC system.

Our demonstration was not an in depth rigorous test but rather a higher level test of what worked and what did not. (Items that apparently did not work were reexamined in further testing in the weeks following the show) It in no way should this test be considered complete or in any way a statement by the NMRA on conformance and/or compatibility or lack thereof. Please note that the intent of this exercise was to provide critical observations and feedback to the manufacturers. Once problems are identified, the manufacturers and/or the NMRA Digital Command Control Working group will be able to correct these problems in a timely manner.

The following Command Stations were made available for our test

The following decoders were made available for our tests.

What Worked.

All equipment interoperated very well in the basic mode covered by NMRA specifications S-9.1 and S-9.2. The only exception was that System-One version 1.0 decoders were not able to be controlled by FastClock+DCC using a Mac Powerbook 540c and appeared sluggish when operated on the Lenz Digital Plus system (the reason for this will be discussed later). System One Version 1.1 decoders delivered at the end of the test operated without problems. The operator interaction present in the newest generation of command stations is a vast improvement over the first generation but still needs further work based upon user reaction. The smoothness of speed control of the latest generation of decoders has improved and the number of glitches present in the first generation has substantially decreased. Several of the problems identified previously could not be duplicated indicating that these problems had already been corrected.

One pleasant observation was that all command stations worked with all power stations. The NMRA DCC working group has discussed from time to time constructing an RP in this area (an early draft exists). To date the manufacturers have achieved compatibility in this area even in the absence of any formal specification.

Areas needing improvement.

S-9 Compliance

Analog Operation

Present System One decoders do not operate on 12 volt DC. The decoders must be removed and the dummy plug reinstalled to allow operation on conventional analog DC power. (All system one decoders come with an RP 9.1.1 2x4 medium current plug)

S-9 power pickup

All manufacturers recommend double gapping the power sections of the layout. Digitrax, System One and Loys-Toys auto reversing techniques requires double gapping (Lows Toys requires a quad gapped portal). This works very well with locomotives with all wheel pickup but does not work with locomotives whose front wheels pickup power from one rail and rear wheels pickup power from the other rail. (common in most HO brass steam locomotives). S-9 needs to be upgraded in this area and the manufactures of DCC auto reversing units need to spell this out clearly in their instructions. (Note if auto reversing is not used, a common rail wiring technique between all power stations allows these locomotives to operate but at the cost of higher power levels when the locomotives straddle power blocks)

S-9.1 Compliance

In the week after the show extensive testing was done to identify why the System One Version 1.0 decoders worked with some command stations and not others. The NMRA standards specify a range of values for the length of "1" bits that decoders must properly read and a smaller range that command stations may transmit. We found that the System One version 1.0 decoders accepted an overly small range of valid "1" bits and that the Lenz and Fastclock+DCC command stations were apparently transmitting "1" bits at the edge or outside this range. After the System One developers extended the dynamic range of acceptable "1" bits the decoders worked reliably with all the command station tested. A new release of decoder code (version 1.1) has been released for System One. Fortunately initial System One decoders have the unique feature that they can be reprogrammed with this new version. Users of Version 1.0 decoders are encouraged to take advantage of this service. Of note is that both Lenz and System One content that they are transmitting/receiving "1" bits within the acceptable range indicating the need for more rigorous formal conformance testing.

RP-9.1.1 Compliance.

In an early pre version of the plug RP, pins 3 and 7 were allowed to be connected on both the socket and the plug. In the final version which was completed in August and just recently approved by the NMRA BOT, this connection was only allowed on the plug. This allows the ability to use pin 3 for a function (for example headlight dimming.) Unfortunately both the pre-production Atlas locomotive and the Harmon products plugs have these pins connected. The result is that if this pin is used for a function (such as on the System One Decoder) this function will be instantly shorted out unless the connection between pin 3 and 7 is cut. (experience speaking) Altas and Harmon are aware of this and plan to fix it immediately. Users of existing Harmon sockets should take note and cut the connection between pins 3 and 7. The NMRA mid amp plug is not keyed but if a function is implemented only operates the headlights and functions correctly if the plug has the correct orientation. Manufacturers of plugs and sockets are encouraged to mark pin 1.

System-One decoders use F1 to control the rear headlight and use the yellow wire for this. While technically correct it is confusing as the other color codes for functions are one color off from the Digitrax decoders. It perhaps would be better if System one did not use the yellow wire as it intentionally has no function that goes on automatically in the reverse direction controlled by FL (the light function).

Extended RP Compliance

Some problems were encountered in the use of more advanced features indicating that more testing is needed in these areas. It should be noted while draft RPs have existed for some time, the NMRA has not finalized all of the extended operations RPs yet so it is not surprising that some problems would arise in these areas.

Address Map

If you program a Digitrax decoder to have an address greater than 127, expect it to have serious problems if you are using a command station that used extended addressing (such as the System One command station) or occasional problems if you are controlling accessory controllers. These problems can be avoided by restricting the address for single byte decoders to be between 1 and 127 as specified in the draft RP.

Speed Table resolution

We have demonstrated in the past that 14 speed steps are more than adequate in most situations so long as the 14 steps are chosen to reflect the operating characteristics of the locomotives. We asked for and received locomotives from the manufacturers with decoders installed. It is clear that these locomotives were not tuned to the 14 speed steps. While many of these decoders allow the user to customize the decoder to the locomotive, it is apparent that the speed tables as loaded by the manufacturers do not provide smooth speed resolution across the entire 14 (or 28) speed step range for the common locomotives. The 14 speed step mode works well only if the speed curve is consistent with the locomotive the decoder is installed in. In most decoders, a significant jump in speed was noticed in the 4-7 speed step range unless the speed table was customized (a process that most users do not perform) Only the Lenz Knobby appeared to perform the speed doubling technique to smooth the speed steps. The result was that many users preferred the 128 voltage levels speed mode which consumes more bandwidth and does not allow power matching between locomotives. All this indicates that the manufacturer need to spend more time in constructing better internal speed tables and providing information on which speed table works best with each locomotive.

Headlight Control

The NMRA RPs specify bit 4 of the function byte for the light function, while pre NMRA decoders in 14 speed step mode use the undefined bit in the speed and direction packet. System One decoders which properly implement this feature of the NMRA draft specifications will flicker the headlight on pre NMRA systems that fail to include the headlight information when transmitting the function packet. What is happening is the speed and control packet tells the headlight to go on while the incorrect function packet tells the headlight to go off. Since packets are being transmitted to fast this flickering effect may not be noticed. This flickering effect is most pronounced using FastClock+DCC in 14 speed mode (not likely to be used as 28 speed mode is implemented correctly) and in Digital Plus when a very large number of locomotives is being controlled. The fix is relatively easy to implement in these command stations.

Accessory Control

The Fastclock+DCC 3.4 and System One command stations were not able to successfully control our NMRAF5 switch machine controllers. This seems to indicate that perhaps the specifications are unclear in this area or that the address map between different command stations is not fully understood. System One expects to have this fixed very quickly and we will investigate this further in the upcoming weeks.

Function Control

Digitrax Big Boy Command Station has chosen F2 to be momentary while F1, F3, and F4 are on/off. System One has chosen to use F1 for the headlight in the reverse direction. Most other command stations have chosen all four functions to be on/off except for the Lenz Knobby which can be customized to indicate which functions are on/off and which are momentary. This can lead to real problems. For example if you use F2 for a on/off function the Big Boy system will not be able to control it. (Function address maps are not implemented yet.)

Programming

The NMRA draft RPs are more specific than implemented in early decoders concerning when the decoder can accept programming instructions (used to alter its Configuration Variables (such as the decoders address)) Problems were encountered in the ability of some decoders to be programmed by other command stations. More work is needed in this area.

Compatibility Problems where no Standards or RPs exist

Cab Bus Wiring.

Each manufacturer has different bus wiring schemes as each has a different bus architecture. Unfortunately many have begun to use the same plugs with different wiring schemes. In a module setting this can cause havoc and could potentially cause damage to equipment. Clearly the NMRA technical department needs to provide some standards in this area, at a minimum the pin assignments of the various common plugs should be defined.

Accessory feedback

The ability for accessories decoders to provide feedback to the command station will become a real problem in the near future. There are multiple protocols for feedback decoders that already exist and the products are not interoperable at present. More disturbing is that one manufacturer (Digitrax) has decided at present to keep its protocol proprietary. This means that Digitrax accessory decoders will only be able to provide feedback to Digitrax command stations. Since feedback will become an important issue in the future, perhaps the NMRA should initiate activity in this area.

Stretched Zero Mode

The System One and Roco command stations were not able to perform the stretched zero operation necessary for controlling locomotives without decoders. Apparently there is a way to do this for the ROCO system but our documentation did not indicate how. {note stretching zeros is not required but is very beneficial if not all your locomotives have decoders}

Summary

DCC Blowout 95 was a tremendous success. It indicates a significant need for both NMRA formal conformance testing and a need to document existing glitches so that manufacturers can work around problems that exist with existing Decoders and Command Stations. It also demonstrates a real need for the NMRA to update S-9.

While multiple compatibility problems were found, the basic conclusion remains that we were able to successfully operate multiple locomotives using different manufacturers decoders from multiple manufacturers command stations. The most serious problem found ("1" bit reception by Version 1.0 System One decoders) has already been fixed by the manufacturer. Other compatibility problems found arise in the use of the extended features, and since all the command stations that use these features have software that can be easily upgraded, these problems should not pose a serious problem to DCC users once the RPs are agreed upon.

On the weekend of March 1Oth the NMRA DCC working group is holding a meeting with the manufacturers to resolve the outstanding compatibility problems and hopefully finalize the extended RPs. Problems identified to date should be corrected shortly thereafter.

Based upon the results of DCC Blowout 95, I recommend that compatibility demonstrations such as this be redone at least yearly to ensure that new problems do not crop up unnoticed.

Stan Ames
NMRA #L05357

P.S. While I do not often provide product endorsements, I will make an exception to endorse the Loys Toys decoder tester. This box allows the user to hook up a decoder (using a plug or quick connect for wires) and test each decoder function by observing diodes connected to each output. For dealers selling DCC decoders or users with many decoders, this devise has proven to be a simple yet effective test tool.

Copyright 1992 - 2001 by TracTronics, Incorporated.
Last updated: June 25, 2001
These pages maintained by: Richard F. Weyand
Please send all comments/problems to: weyand@rcnchicago.com