S-9.1.2 Review (Nov 1, 2020 Draft)

[bakerstu]

These are review comments that came in over email.

Line 115, Section 2.2.2 Header: Change “Statin” to “Station.”

Line 139, After 1st paragraph Section 2.3: 2.3 does not indicate that TIA/EIA‑422 and TIA/EIA‑485 have similar characteristics and are often used interchangeably. I would add the following paragraph from line 75 of TN‑9.1.2:

The TIA/EIA-422 and TIA/EIA-485 driver/receiver standards have overlapping operating ranges and are commonly used together and/or interchangeably.

Line 178, Physical Medium and Topology: The lack of termination and the plethora of topologies likely limits the maximum wire length significantly. This might be a problem if a command station is at one end of a long linear layout and a power station is at the other end of the layout. This could result in a wire length over 100 feet (30.5 meters) or more.

Has the lack of termination and topology types been verified for both Full Scale and Driver/Receiver interfaces?

[bakerstu]

...additional comments via email...

first I had to look for the most recent version. At https://github.com/nmradcc/documents/blob/S-9/standards/S-9.1.2%20Power%20Station%20Interface.doc it notes "Last Commit 20 Aug 2019" but the document header says July 2012. So I used the version from https://www.nmra.org/dcc-rps-standards

However, comments up to section 2.2 are regarding already the oldest versions of RP-9.1.2. As several things don't match my personal experience, I always somewhat ignored this RP or standard.

Something that always puzzled me is the Unipolar Signal interface. Is there a command or power station using this? In the current standard I can't find any description of a unipolar signal. It is just in the Terminology.

The only use I can think of are the 5V logic signals used by Märklin, but I doubt this will survive the 12 V maximum of the old Driver/Receiver (Voltage) Interface.

Who uses the Driver/Receiver Interface? NCE?

2.1.4 Power State Common Shouldn't it be "2.1.4 Power Station Common"?

I assume that "Märklin style" power stations with one output at ground and the other output changing between plus and minus are not conforming to this standard.

2.2.1 Command Station Output Signal "A bipolar signal must appear differentially on a two-wire cable ..." Is it allowed to combine the two lines for the power station interface with the (throttle) control bus in a single cable? This is done by most command stations. Should the wording not be like "... must appear differentially on two wires/lines/terminals ..."?

2.3 Driver/Receiver Interface Why is there a distinction between TIA/EIA-422 and TIA/EIA-485? Yes, there are minor differences, but do we need to specify both types? We don't need the bi-directional property of EIA-485, but we may want to have the higher maximum common mode. The only manufacturer I can think of using this interface is NCE. If this is true, shouldn't we just check what NCE is using?

I would really like to have as few options for the power station interface as possible.

3 Physical Medium "Manufacturers that utilize feedback are encouraged to provide their specifications to the NMRA for possible future standardization." The Lenz "E" line to communicate a short is available at least since I know of DCC. But it never became part of a standard. I think we can drop this as we will not get agreement on any feedback line for boosters.

5 Labeling The real world labeling I know of are

• "CDE" for the Lenz interface with C and D being the differential full scale interface and E for notifying short,
• "DCC" for a two line differential full scale interface and
• "Märklin" for a five pin interface with logic signals for data, short and power station on/off. I believe most users will not understand the terms used in this standard. To me anything other than a differential full scale interface input should have a warning sign - except for a five pin Märklin style connector.

If we list S-9.2.4 as normative reference, we should provide a note in 2.1.5 Power Station Fail-Safe. I believe the 30 ms are the only link to S-9.2.4.

[bakerstu]

These are review comments that came in over email.

Line 115, Section 2.2.2 Header: Change “Statin” to “Station.”

Fixed.

Line 139, After 1st paragraph Section 2.3: 2.3 does not indicate that TIA/EIA‑422 and TIA/EIA‑485 have similar characteristics and are often used interchangeably. I would add the following paragraph from line 75 of TN‑9.1.2:

The TIA/EIA-422 and TIA/EIA-485 driver/receiver standards have overlapping operating ranges and are commonly used together and/or interchangeably.

Added.

Line 178, Physical Medium and Topology: The lack of termination and the plethora of topologies likely limits the maximum wire length significantly. This might be a problem if a command station is at one end of a long linear layout and a power station is at the other end of the layout. This could result in a wire length over 100 feet (30.5 meters) or more.

Has the lack of termination and topology types been verified for both Full Scale and Driver/Receiver interfaces?

I have personally tested the Full Scale Interface through 1000 ft (300m) of twisted pair impedance controlled cable (Cat5e) without termination. I observed no significant degradation in signal quality at the receiver (Power Station). I have not performed the same test with the Driver/Receiver Interface. However, past experience leads me to believe that the extremely low data rate of the DCC signal will result in successful operation without termination. I have added a note to the complementary section of the Technical Note:

The Full Scale Interface has been tested successfully through 1000 feet (300 meters) of impedance controlled twisted pair cabling. The Driver/Receiver Interface option has received considerably less testing, and may benefit from 120 ohm termination at each end of the transmission line.

[bakerstu]

...additional comments via email...

first I had to look for the most recent version. At https://github.com/nmradcc/documents/blob/S-9/standards/S-9.1.2%20Power%20Station%20Interface.doc it notes "Last Commit 20 Aug 2019" but the document header says July 2012. So I used the version from https://www.nmra.org/dcc-rps-standards

However, comments up to section 2.2 are regarding already the oldest versions of RP-9.1.2. As several things don't match my personal experience, I always somewhat ignored this RP or standard.

Yes, you are correct. In fact, this is the main motivation for revising 9.1.2. After discovering many likely mistakes and self contradictions, I discussed the original intent with the original author. I was told by the original author that the document was written under extreme time pressure and the original intent was to document the two prevalent options in the market (full scale and RS-485/422). The original author was not surprised that I found significant deficiencies.

Something that always puzzled me is the Unipolar Signal interface. Is there a command or power station using this? In the current standard I can't find any description of a unipolar signal. It is just in the Terminology.

I'm not aware of any use of a unipolar signal, and that is why I removed it. If there is an example use case of this, then I want to know about it so that we can add it back into the document with appropriate commentary.

The only use I can think of are the 5V logic signals used by Märklin, but I doubt this will survive the 12 V maximum of the old Driver/Receiver (Voltage) Interface.

The old Driver/Receiver (Voltage) Interface was intended to be RS-422 (confirmed by the original author) and was improperly documented.

Who uses the Driver/Receiver Interface? NCE?

NCE uses the Full Scale Interface (nearly identical circuit to Lenz). CVP has used RS-422 since at least 1997. OpenDCC and Tams are now using RS-485 (BiDiBus) in all of their new products.

2.1.4 Power State Common Shouldn't it be "2.1.4 Power Station Common"?

Fixed.

I assume that "Märklin style" power stations with one output at ground and the other output changing between plus and minus are not conforming to this standard.

A power station common has no prescribed relationship to "Earth Ground". The standard is silent on the use of "Märklin style" power stations, which are typically fully isolated with a common "rail", often tied to the negative supply of the Power Station, as you have described. There is no intent for this standard to disallow "Märklin style" Power Stations. They are fully permitted.

2.2.1 Command Station Output Signal "A bipolar signal must appear differentially on a two-wire cable ..." Is it allowed to combine the two lines for the power station interface with the (throttle) control bus in a single cable? This is done by most command stations. Should the wording not be like "... must appear differentially on two wires/lines/terminals ..."?

I've changed the word "cable" to "interface" (three different places in the document) and added a note to section 3 Physical Medium:

Other power and/or signal wires may also be bundled in the same cable as the Power Station Interface.

2.3 Driver/Receiver Interface Why is there a distinction between TIA/EIA-422 and TIA/EIA-485? Yes, there are minor differences, but do we need to specify both types? We don't need the bi-directional property of EIA-485, but we may want to have the higher maximum common mode. The only manufacturer I can think of using this interface is NCE. If this is true, shouldn't we just check what NCE is using?

I would really like to have as few options for the power station interface as possible.

Both RS-422 and RS-485 are presently in use in commercial products. However, I have added the following note:

TIA/EIA-422 and TIA/EIA-485 driver/receiver standards have overlapping operating ranges and are commonly used together and/or interchangeably.

3 Physical Medium "Manufacturers that utilize feedback are encouraged to provide their specifications to the NMRA for possible future standardization." The Lenz "E" line to communicate a short is available at least since I know of DCC. But it never became part of a standard. I think we can drop this as we will not get agreement on any feedback line for boosters.

Actually, there is an active proposal right now that two manufacturers have already implemented. The proposal places a Bi-Directional cutout on the Power Station Interface through which the Power Station can transmit using a 5mA current loop (basically the same as Bi-Directional at the track but using 5mA). It turns out that Lenz is already doing something similar with a 30mA (I think) current loop. Lenz has not been interested in collaborating on this effort.

Unfortunately, the Lenz method isn't going to work very well across long transmission lines. This is why the proposed method uses 5mA instead of 30mA (tested through 1000 ft, 300m of transmission line). We have collected an extensive set of test data that shows 30mA performs quite poorly in long transmission lines.

Stan Ames, Erich Whitney, and Robert Manchek have been involved in the peer review of the proposal. Additional information can be provided upon request, including early draft documents.

5 Labeling The real world labeling I know of are

• "CDE" for the Lenz interface with C and D being the differential full scale interface and E for notifying short,
• "DCC" for a two line differential full scale interface and
• "Märklin" for a five pin interface with logic signals for data, short and power station on/off. I believe most users will not understand the terms used in this standard. To me anything other than a differential full scale interface input should have a warning sign - except for a five pin Märklin style connector.

The A and B polarity wording is taken from the original document, and is in common use in the market. Yes, I know Lenz doesn't use this. Users do not need to understand the terms, they just need to match like with like. The standard purposely falls short of being too specific on the guidance provided by manufacturers for their products. Manufacturers that provide good guidance will be more successful in the market. There is room for innovation.

If we list S-9.2.4 as normative reference, we should provide a note in 2.1.5 Power Station Fail-Safe. I believe the 30 ms are the only link to S-9.2.4.

I've added a reference to S-9.2.4 in the text:

In order to avoid a decoder conversion to an alternate power source, as described in S-9.2.4,

[bakerstu]

Actually, there is an active proposal right now that two manufacturers have already implemented. The proposal places a Bi-Directional cutout on the Power Station Interface through which the Power Station can transmit using a 5mA current loop (basically the same as Bi-Directional at the track but using 5mA). It turns out that Lenz is already doing something similar with a 30mA (I think) current loop. Lenz has not been interested in collaborating on this effort.

Unfortunately, the Lenz method isn't going to work very well across long transmission lines. This is why the proposed method uses 5mA instead of 30mA (tested through 1000 ft, 300m of transmission line). We have collected an extensive set of test data that shows 30mA performs quite poorly in long transmission lines.

I since have been party to a more extensive tear down and reverse engineering of the technique used by Lenz LV103 Booster. It appears that Lenz is very likely using a 5mA current loop, similar to the active proposal.