Reversing loop modules - gory details

[Andrew_au]

Splitting off from this thread on reversing loops, to explore some of the gory details and not derail the other thread.

Most of the discussion will be about the behaviour of reversing loop modules in various wiring configurations. It will focus on DCC, but there may be some DC questions also.

But first, a long-ish introduction!

Preamble​

Terminology​

• At the risk of causing confusion, I will use +/- when talking about fixed polarity wiring, and P/N when talking about DCC wiring and phases.
• When referring to specific wiring labels used by various manufactures, I will use 'quotes'.

DC vs DCC, polarity and phase​

Polarity refers to one "end" of an electrical connection having a positive (+) voltage compared to the other end, which has a negative (-) voltage.

In a DC (analog) system, one rail is permanently attached to the +ve end of the power supply, and the other to the -ve.

• Speed is controlled by altering the voltage across the rails.
• Trains always travel in a fixed direction with respect to the power supply. To reverse a train, the power supply reverses the polarity.

In a DCC (digital) system, there is no fixed polarity. The power supply maintains a constant voltage differential but flips the polarity (+ vs - rail) many thousands of times each second. The control station encodes data for decoders by altering the rate of change. Train direction and speed is controlled by the decoder normalising the direction and voltage applied to the motor.

In DCC, two wires are "in phase" if their DCC polarity is the same, and in "opposite phases" (or out-of-phase) if their DCC polarity is different. In-phase (same polarity) wires and rails can be joined together as if a single wire. Opposite phase (opposite polarity) wires must not be directly joined, as this creates a short circuit (which is bad).

The reversing loop problem​

A reversing loop allows a train to travel one way down a track, go around a loop, and leave on the original track facing the opposite direction. The train is not required to slow down, stop or otherwise do anything unusual while travelling around the loop.

Reversing loops cause interesting wiring issues when used with two-rail track-powered electric trains. With a two-rail system, the rails have opposite polarity, and connecting the two rails directly causes a short circuit (bad). However, if you consider a reversing loop and trace (say) the outer track, you will notice that the same "rail" goes out on one side and returns on the other, defeating a key requirement of a two-rail system.

Note that 3rd rail systems do not have this issue. In such systems, power flows between the centre rail or pantograph and either or both of the regular rails. Electrically speaking, both the regular rails are a single rail. Because the uniqueness of the "3rd" rail is maintained around the loop, there is no risk of short.

Obviously, systems that do not rely on track power (e.g. battery, live steam) also do not have issues with reversing loops.

Although not strictly a "loop", wye junctions have the same problem. Any track layout that allows a train to reverse its facing without breaking its connection to the track will have this problem.

See this post for good pictures.

Reversing polarity and DCC​

In a DCC system, train direction is determined by the locomotive's facing. If you pick a loco up, turn it around, and put it back down, it will head off in the opposite (track) direction.

So, consider putting a dual-track insulator at some point in the reversing loop. On one side of the insulator, the train goes forward. On the other side of the insulator, the rails have the opposite phase, but the train still goes forward.

The problem comes as the train crosses the insulator. If a wheel (or electrically connected set of wheels) touches both sides of the insulator on the same track, a short circuit occurs. Most locomotives are designed with multiple pickup points to protect against brief dead-zones, so the only way to prevent a short circuit is to create an even longer dead zone. Dead zones are also bad.

DCC does provide an unique solution. The controller is flipping the polarity of the track supply many times each second. So all the electronics is used to dealing with polarity change. If we add a circuit that allows us to flip the polarity on the fly, decoders will read it as unexpected noise in the signal but otherwise be unaffected. As long as extra flips are infrequent, all is good.

So, rather than a single insulator, create an isolated section protected by insulators on each end. This section must be long enough to hold an entire train (or at least the electrically interesting part of the train - wagons with plastic wheels can be ignored). What matters is that this section has the same polarity as the adjoining section when the train is crossing that insulator. When the train crosses one end, power the section so that the polarities match. Once the train is completely within the isolated section, flip the polarity to match the other end, and the train can leave without ever experiencing a short circuit or disruption of power.

Reversing polarity and DC (analog)​

Traditional DC (analog) systems have an additional problem. Not only do they need to prevent a short circuit, but from the train's perspective the polarity of the circuit must remain consistent. To reverse a train in an analog system, the main line needs to reverse polarity so the train can keep travelling back the way it came.

This works much the same as a DCC isolated section, but in this case we reverse the polarity of the section that the train does not occupy. Note that this means that anything on that track has just been reversed, so scope of reversing tracks is much more restricted in DC systems than in DCC.

Using blocks​

In a DC layout that supports multiple directions on the same track, it may make sense to divide the network up into blocks that can be independently controlled for direction (and maybe speed). Whenever a train is crossing a block boundary, both blocks must be synchronised.

Blocks are useful in a DCC layout, but are only required for two cases:

• Power districts if the power requirements exceed that of a single booster
• Reversing sections

Because there is no permanent polarity in a DCC network and only phase matters, reversing sections only need to support phase matching. In DC, any section that will be transitioned in both directions needs to be "reversible". In DCC, something needs to be able to switch phase to avoid short circuits, but this section can be at any point in the layout so long as it prevents electrical loops.

Reversing loop modules​

Reversing switches can be implemented manually using a centre-off DPDT switch, or equivalent electronics that can take a fixed polarity supply and pass it on with either fixed or reversed polarity. However, this requires the operator to be very alert, as the circuits need to be switched every time a train crosses them.

A reversing loop module does this automatically. On a DCC circuit, the module needs to detect that the train is about to a controlled boundary and synchronise the phase between the two sides of the boundary, reversing polarity of one side if required. DC reversing is more complex in that the current section must not be reversed - only the new section can be reversed.

Note that the reversing loop module controls only a single section - it is the module's responsibility to synchronise that section to the adjoining sections.

• Sections controlled by automatic reversing loop modules must be adjacent to fixed phase sections, otherwise the modules can confuse each other.
  • some DC (analog) layouts use sophisticated controllers to control multiple sections independently. This is way beyond the scope of this discussion.

There are three basic mechanisms for triggering a reversing loop module:

• short circuit - the module detects the current change as the train shorts the insulator, and reacts accordingly
• sensor tracks - the boundaries of the sections are not a single isolated pair, but instead a short section of isolated track. Current in this section indicates the presence of a train.
• non-track sensors (e.g. reed switches)

Not a "reversing loop"​

Not everything that involves playing with track polarity is a reversing loop.

Auto-reverse modules are modules designed to allow a train to automatically reverse direction on the current track. A "reversing loop" module maintains the train's motion but fixes the track polarity to avoid causing short circuits.

• In DCC, auto-reverse modules are entirely about controlling the train - the track is polarity unaffected. In DC (analog), the track controls the train, but the behaviour is still different to an reversing loop module

Trains can be reversed using turntables or the like. If the turntable only powers the track that is currently connected, there is no polarity issue for DCC. In DC, track rather than loco orientation matters, so you're effectively just "reversing" the train. A reverse loop module might be useful for a DCC turntable if you want to keep the turntable powered independently of any of the feeder tracks.

---

That's just the introduction. The questions are in the first reply.

 

[Andrew_au]

(continued - questions in this post)

Wiring Modern Reversing Loop Modules​

The trend in reversing loop modules is to use sensor tracks for train detection. Examples include LGB 55085, Massoth DiMax 8157001, and Roco z21 Multiloop (10797). All of these have the following connections:

• Power +/-
  • Power supply to module in analog mode. Can't use track power as the track is not always powered.
  • Oddly, the Massoth and LGB modules do not label the power pins as +/-.
• Track in +/-
  • connects to fixed polarity track
• Track out +/-
  • connects to controlled track
• Sensors 1,2,3,4
  • connects to sensors
  • not used in short-circuit mode
  • in sensor track mode, connected as pairs 1,2 and 3,4 on each end of controlled track
  • in alternate switch mode, only 1 and 3 are used, and the other end of each sensor is connected to mains.

Links to products​

Correct as of time of writing. Manuals can be found on product pages.

• LGB 55085
• Massoth DiMax 8157001
• Roco z21 multiLoop

Comparing models​

The manuals for the Massoth / LGB and Roco modules show very similar wiring diagrams, but there are differences.

Firstly, Roco uses P/N rather than +/- for track in / track out. This is generally consistent with Roco (and Zimo) documentation, which uses P/N for DCC track polarity. I suspect LGB and Massoth use +/- for compatibility with DC track.

Layouts​

• All models use the same 4 wires for short-circuit mode (2 to track in, 2 from track out). There is no diagram using short circuits for analog mode.
• In sensor track mode:
  • "track in P/+" connects to sensors 1 & 4.
  • "track in N/-" connects to sensors 2 & 3.
  • 1&2 are paired, 3&4 are paired
  • "track out P/+" is between 1 & 3
  • "track out N/-" is between 2 & 4
• Like sensor mode, track contact mode has no connection between the sensors and the track polarity.
  • All models use sensors in parallel on each side of the track split, connected to 1 or 3 and to "track out P/+"
• In analog mode:
  • the main control is wired to "in" and to the loop.
  • the shared line is wired to "out"
    • For anti-clockwise loop, all models show inner (left) track as +ve
    • Massoth / LGB - "out" matches main on entry
    • Roco - "out" matches main on exit
  • Track contacts are only on ingress to track split

Questions on wiring​

Theoretical model​

One could achieve the behaviour of a reversing loop by having switched connections in parallel with the insulators. Whenever a sensor is triggered, the connections for that sensor close and all the other connections open. But that requires 2x2 connections plus sensors at each insulation boundary (one connection on each side of each insulator per rail, x2 rails per boundary).

If we assume that all the external connections are ultimately connected to each other, we can simplify the wiring and just trigger our DPDT. We now only need 2x2 connections for the entire system (2 rails, one "in" and one "out" per rail), plus whatever is required for our sensors. But this requires that we know the relative polarities of the inside and outside rails at each boundary.

Determining polarity​

Question: Is there a mandated relationship between the rail polarity and particular sensors? For example, sensor 1 is always used when the external polarity matches the polarity of the reversing track (P-P), and sensor 3 is always used when the external polarity is reversed with respect to the reversing track (P-N).

The interesting case here is the track contacts case. In this case, the sensors cannot detect track polarity - only the tracks themselves can do this. The odd situation here is the Roco.

• On page 18 (digital with track contacts), a train triggering sensor 1 connects N-N, P-P.
• On page 21 (analog with track contacts), a train triggering sensor 1 connects N-P, P-N.

How does the module know which way to set the virtual DPDT in this situation?

• It can't use fixed configuration, because in these two roughly equivalent scenarios the "fixed" wiring is oriented oppositely.
• It can't let the switched track "float" and detect the train shorting the circuit, because in the ingress case the train has not necessarily reached the powered rail.

Related question: which of the following must be done "right" for the system to function?

• relationship between "in" polarity, "out" polarity and the location of the #1 and #3 sensors
• orientation of the 1/2 and 3/4 track sensors relative to their partners

Using sensor tracks​

Question: when sensor tracks are used, are they powered at any point?

• Could a train stall if the sensor tracks were too wide?

Question: are the sensor tracks used as part of the polarity calculation? They could be, but this wouldn't work for modes that do not use the sensors.

Using short circuits​

Question: In short circuit mode, I assume the reversing loop module just watches for a short circuit and flips the controlled track polarity if one occurs. Is this correct?

 

[Greg Elmassian]

Another small niggle... DCC does not alter the rate of change (that would be called rise time), but the duration of a single cycle, or simply modulates the frequency. One of the benefits of FM (frequency modulation) is that the signal strength changes do not change the data.

(remember AM radio vs FM radio)

Another point, well designed autoreversers can be programmed to ignore temporary shorts caused by a wheel spanning an insulated joint, which means that from a practical perspective, the reversing section can often be shorter than the total train length as long as there are not lighted cars or more locos at the end of the train. (really depends on the autoreverser, but with what we use in the USA, usually is no issue). this really helps on reversing wye (triangle) sections.

Also the better autoreversers we often use in the USA do allow cascading autoreversing sections, by having adjustable delays and current triggers. Again this is most common in the USA, since we rarely use the "sensor" style autoreversers.

Lastly, sorry, the modules you are referring to are not modern, they are old school design. Almost everything we use in the USA has a microprocessor analyzing the short circuit caused when polarity is not right. (I just have to give a dig, it's the original design when stuff could not be made fast enough, even the continued use of relays belies the old style design).

OK, I've had enough fun... for people whose world only consists of the "sensor type" autoreversers, this information is good for the choices within that world. You would have a heck of a time trying to convince the average DCC user in the USA.

 

[Andrew_au]

Appreciated, Greg. A lot of what I'm doing here is trying to understand the stuff that they don't tell you in the manuals.

DCC does not alter the rate of change (that would be called rise time), but the duration of a single cycle, or simply modulates the frequency

Changing the duration of a cycle does tend to change the rate of change . I was treating the rise time as negligible. Point is that throwing in an extra arbitrary phase shift doesn't meaningfully alter the behaviour of the system. Your power model remains the same, and your signalling model treats it as noise.

Almost everything we use in the USA has a microprocessor analyzing the short circuit caused when polarity is not right.

I'm interested in the details of how this works.

You've essentially got 3 signals:

• The reference signal coming in to your module
• The output from your module to the reversing track
• The signal being shorted from the external track

Meanwhile, you've got two power devices involved:

• The reversing module powering & monitoring the reversing track
• Whatever is powering & monitoring the external track

If the external track and the reversing track are in-phase when shorted, no harm no foul.

How does it detect an out-of-phase short? I assume you'll see a rapid rise in current? How do you ensure that the reversing loop module takes corrective action before the other device does?

well designed autoreversers can be programmed to ignore temporary shorts caused by a wheel spanning an insulated joint

I assume this is highly dependent on train speed? How would the module handle parking a permanent metal wheel set across the join (assuming, say, the loco was bridging the other end)?

 

[phils2um]

Question: when sensor tracks are used, are they powered at any point?

Yes, the sensor rail segments get electrically connected to the track the loco is leaving through the wheels of the loco when it first enters the sensor zone. Assuming multiple wheel/skate pickups on the loco.

Could a train stall if the sensor tracks were too wide?

Again, yes. The loco will lose power if the sensor segment is longer than distance between the electrical pickups in the loco. (ignoring any "keep-alive")

Question: In short circuit mode, I assume the reversing loop module just watches for a short circuit and flips the controlled track polarity if one occurs. Is this correct?

This is certainly how the LGB and Massoth modules operate. The microprocessor based reverse loop modules referred to by Greg such as the DCC Specialties PSX-AR have more sophisticated algorithms to determine when to reverse the polarity.

 

[PhilP]

The only thing I think needs wording changing is in your first reply, Andrew.

Early on, you say the unused sensor connections are connected to "mains".

Certainly, here in the UK, "mains", or "the mains", refers to the incoming AC domestic power supply to a property.
Not what you are suggesting, I am sure?

PhilP

 

[Greg Elmassian]

Thank you for your positive attitude Andrew, my comments are to make it more complete, and some niggles (fine points) on accuracy. I have several electrical engineers working for my, besides my own university education in this area and physics.

"Changing the duration of a cycle does tend to change the rate of change . I was treating the rise time as negligible. Point is that throwing in an extra arbitrary phase shift doesn't meaningfully alter the behaviour of the system. Your power model remains the same, and your signalling model treats it as noise."
That is true grammatically, but not the right way to express this for frequency modulation. Frequency modulation cycle by cycle is NEVER called rate of change. Even "pulse width modulation" is misleading, since it is cycle by cycle, not a continuous signal being modulated, for example DCC and RF motor drivers use PWM, but the modulation stays constant for a constant speed. DCC is clearly a FM Frequency Modulation scheme, and that is a fundamental and universally understood term.

How does a "smart" DCC autoreverser work?

You stated that the information you have is:

• The reference signal coming in to your module
• The output from your module to the reversing track
• The signal being shorted from the external track

Well, maybe I could rephrase this, since your terms are a bit nebulous, and add the very important ones you did not mention.

• you have the dcc input signal, which really does not help you control anything
• the output to the reversing track, which you either interrupt or leave alone (but more on this), from your perspective, this is controlled by relays, all on or all off, and a slow switching time since they are mechanical relays
• the signal being shorted from the external track... I cannot see how this is different from your first point,

First and foremost, the autoreverser not only can sense a short, but an intelligent one will measure the current
Next the most important issue is the ability to ANALYZE the amount of the short and the duration. Beyond this, analyzing the "waveform" of the short, is it briefly high current and then tapers off some, or ??
Next the ability to analyze the short over time.


So for example, not all of the inner programming of the DCC Specialties PSX-AR are made public.

The system, using the microprocessor and firmware (there is a program in it to analyze this) can:

• delay the tripping of the internal breaker, or delay autoreversing. This helps when autoreversers are cascaded, a loop within a loop, or to not "fight" the DCC booster circuit breaker or any other breaker "upstream.
• if the short detected is of very short duration (a metal wheel crossing an insulator, when the train's powered cars are in the reversing section but the train length is longer than the reversing sections), then the sytem can ignore it an wait a small interval to re-check.
• The system can do a "slow on" to the power, i.e. one feature is apply power, notice the abnormal current draw, and pull the power wait a small amount of time and reapply. This feature helps false tripping when a loco or sound card has large "keep alive" capacitors, which can draw so much current when first powered, they appear to be a short.

The possibilities are endless since the unit can analyzed the amount, duration, repetition, and differences in current draw.

Using a microprocessor that can sample information thousands of times a second (easily), the possibilities are pretty much endless. You LGB "sensor" modules are an abacus, the microprocessor driven algorithm based units are indeed computers.

Hopefully that helps describe the state of the art. In the US, this is what large clubs use. There are other competitors with similar hardware, but the company named is the one who has been evolving this technology over the years.

More reading, 10 years in development:

PowerShield X - Family of Intelligent Circuit Breakers and Auto Reversers | Products from DCC Specialties.

Greg

 

[Andrew_au]

Early on, you say the unused sensor connections are connected to "mains".

Certainly, here in the UK, "mains", or "the mains", refers to the incoming AC domestic power supply to a property.
Not what you are suggesting, I am sure?

Not at all. However, they (and the 'power' terminals on the module) are connected to a fixed voltage DC supply, external to the analog DC controller. I could not think of a better short name for it.

 

[Andrew_au]

that is a fundamental and universally understood term

Some irony here on the term "universal", in that it itself is a term of art.

If I tell my wife that the DCC signal is a "frequency modulated square wave" she'll go "huh?". Heck, my eyes glaze over somewhat reading NRMA S-9.1. But if I tell her that it works by flipping the polarity of a DC signal, and the decoders use the (variable) rate of flipping to read it as binary data, and anything that needs straight DC power can just rectify the flips away, she gets it, at least enough to understand what's going on.

In any case, my ultimate goal was to explain why wiring up a reversing loop for DCC is much simpler than DC. At least in that in DCC you just need to make sure the track power is phase-matched and the decoder takes care of everything else, whereas in DC the orientation of the signal matters and you're a lot more constrained on how you can structure your reversing blocks.

Well, maybe I could rephrase this, since your terms are a bit nebulous, and add the very important ones you did not mention.

• you have the dcc input signal, which really does not help you control anything
• the output to the reversing track, which you either interrupt or leave alone (but more on this), from your perspective, this is controlled by relays, all on or all off, and a slow switching time since they are mechanical relays
• the signal being shorted from the external track... I cannot see how this is different from your first point,

I was curious as to whether the DCC input signal is used as part of the calculation, or whether it's just a source of power and to pass through to the outputs.

The signal on the track being might be relevant, and is not strictly the same as the DCC input signal. For example:

• it might be out-of-phase with the input signal
• once shorted to the reversing track, the behaviour of the shorted track could well differ from the DCC input signal

Question: is there any significant difference in behaviour if the DCC input signal is from the track being shorted vs an independent DCC signal? Ultimate example of this would be from a different booster.

---

Still curious on what are the actual wiring restrictions for a sensor-driven reversing module. Looking at those diagrams, there are a number of wires that I go "if I were to swap this, would it make a difference"?

It does look like a more sophisticated short-circuit detection systems has a lot of advantages, though. Not the least that you just wire up the DCC source and the track and let the module worry about all the other details.

Question: How precise does the alignment on the insulators need to be? Given that the module is trying to sync the rails on each side of the insulator, it strikes me that they could be a couple of cm out of alignment and it shouldn't matter.

And practical questions:

Question: The US manufacturers I've seen (e.g. DCC specialities) seem to mostly sell unboxed PCBs, whereas LGB / Massoth / Roco are selling boxed units that are not strictly weatherproof but just need basic shelter from rain, runoff & excessive heat. How do people usually mount the unboxed PCBs for outdoor use?

Question: Does anyone know any good Australian distributors for US railway electronics, especially G-scale stuff?

Question: In what cases do I need to be careful about the difference between G-scale and smaller scales? There are some products which are for G only, or HO/N only, or don't care - is there a good principle to tell which is which?

 

[phils2um]

I was curious as to whether the DCC input signal is used as part of the calculation, or whether it's just a source of power and to pass through to the outputs.

The signal on the track being might be relevant, and is not strictly the same as the DCC input signal. For example:

• it might be out-of-phase with the input signal
• once shorted to the reversing track, the behaviour of the shorted track could well differ from the DCC input signal

Question: is there any significant difference in behaviour if the DCC input signal is from the track being shorted vs an independent DCC signal? Ultimate example of this would be from a different booster.

I think you're making this much more complicated than it actually is - at least as far as the LGB and Massoth reverse loop modules. They are simply double pole-double throw switches which couple the switched rails to the in-coming rails or out-going rails. The DCC power and signal is just passed through unaffected. The only "magic" is in how the modules sense whether the switch needs to be thrown to keep the track and hence the booster's output stages from being shorted when a loco crosses the boundary.

The PSX-AR is a different kettle of fish. - all I can say is read the manual. Here's one place where it can be found: https://tonystrains.com/pdf/dcc_specialties/PSX-Manual-Rev-N-Software-Rev-L-Booklet.pdf

 

[phils2um]

Question: In what cases do I need to be careful about the difference between G-scale and smaller scales? There are some products which are for G only, or HO/N only, or don't care - is there a good principle to tell which is which?

My two cents - If it is going to process/transmit large scale DCC power it needs to be good to a minimum of a 24V DCC signal. The current requirements are dependent on the items purpose. For example I converted my LGB Santa Handcart to DCC using a small ESU "HO/N" decoder. It could handle the voltage but was only rated for about 1.5 amps continuous motor current. It was perfect (and cheap) for the purpose.

For anything else Rule 8 applies!

 

[The Shed]

Simplified! (Edited) to avoid doubt, this is with reference to phils2um explanation in post 10, and has no reference to any other observation.

 

[Greg Elmassian]

Wow a lot of questions... not sure all really want them answered.

1. explain it any way you want to your wife. when you start talking details about electronics, then you should use accepted technical terms. For the layman, explain it with toothpicks and duct tape, whatever, but this discussion is down in the details. The term "signal" is being used ambiguously for example. I have had an extended discussion with an "expert" who wrote a book, and he refuses to call DCC AC, but "bipolar DC". Of course he does not have a degree in science, so I guess all the people in universities are wrong.

2. To Phil's comment... yes we were talking the PSX-AR and similar, not simple relays being triggered by the sensing of a loco's presence.... not sure why you insist on 24 volts, the more important thing is low resistance in the system, so the voltages and currents you measure are more related to the short and not connectivity issues or limitations. I'm sure you have heard of the famous "quarter on the rails" test that HO DCC people use to determine the worthiness of the power delivery?

3. To Shed: "simplified"? The PSX-AR with only insulators, 4 connections is simplified beyond your example and is available for around $44 off the shelf and has all kinds of settings that can be tweaked. No offense but why use a microprocessor for something you could do with diodes and flip flops or 555 timers?

So, in the USA we would never have this discussion except for "lgb foamers"... but I see it is almost a cult thing about avoiding "arcing" (sensing an approaching powered car, vs. detecting a phase conflict).... of course watch your track powered trains at night, and after seeing all the "normal" arching, no one has ever continued the argument.

Anyway, Andrew, not clear why you are asking more and more questions... with all the data available and a programmable microprocessor, what you can do is so far beyond triggering a relay it's almost laughable.

I was trying to complete your "article" assuming you wanted to cover the normal available options to people. Not trying to stir the pot.

Greg

 

[The Shed]

"diodes and flip flops or 555 timers?"

Ok, then show us the circuit design and the workings?

 

[Greg Elmassian]

Aw, come on! You showed a simple logic example. I agree it simple. But so simple you could do it with fewer components/less cost. Thus some diode logic to run relays, and if you needed some boolean states, you could use flip flops.

My point was that it would be even simpler with the short-sensing autoreverser, than with the sensing sections.

If my point was not clear, it has to be clear now.

Greg

 

[JimmyB]

Aw, come on! You showed a simple logic example. I agree it simple. But so simple you could do it with fewer components/less cost. Thus some diode logic to run relays, and if you needed some boolean states, you could use flip flops.

My point was that it would be even simpler with the short-sensing autoreverser, than with the sensing sections.

If my point was not clear, it has to be clear now.

Greg

This is DCC, I haven’t got the foggiest what is going on, so not clear, but then I don’t care

 

[Andrew_au]

Anyway, Andrew, not clear why you are asking more and more questions... with all the data available and a programmable microprocessor, what you can do is so far beyond triggering a relay it's almost laughable.

I was trying to complete your "article" assuming you wanted to cover the normal available options to people. Not trying to stir the pot.

It was not exactly my intent to write an "article". Rather, reading the various manuals and some other discussions raised a bunch of questions about the limits of what one could or couldn't do. But my experience has been that going directly to very edge-case questions results in a lot of unhelpful answers:

• Some people assume that I don't understand the background and insist on explaining it, actively refusing to address the questions being asked
• Some people don't understand the background or the questions and give all sorts of irrelevant or unhelpful answers

So I'm genuinely asking anything labelled "Question". Everything prior to that is a combination of "you must be this high to enter" and a chance for people to check my assumptions in case they impact the answers. While the first post and other background might be educationally useful, its goal is actually just getting to the point where I feel safe to ask "given all that, what about this?".

And most of those questions boil down to:

• these 'sensor mode' wiring diagrams are very complicated. How exact do I actually need to be?
• is there a better way to do it?
  • and if so, what do I need to be aware of when installing it?
• what do I need to consider when wiring a reversing loop section that's not a simple 2-entry loop with both ends connecting to a common set of rails (or equivalent Y junction). E.g. multiple entry points, connections with arbitrary phases, different boosters, or another reversing loop module

But to the level of fully understanding the theory, not just "do this and it will work".


But it is looking like the sensible way to go for DCC is a sophisticated short-circuit unit - they are easy to wire (2 from DCC power, 2 to track) and can handle anything as long as you only short one "end" at a time.

Still a little curious as to behaviour, as there seems to be a lot of tunable parameters (why?) and I don't see how you could put 2 reversing tracks adjacent without risking them "fighting".

 

[Greg Elmassian]

OK, I read your post very carefully.

I won't respond to any of the sensor-based questions, besides understand the theory of operation and looking at the examples presented and see they make sense, "how exact" is a question that I, as an engineer would not answer further unless I understood the "reaction" time of the system, like the minimum trip time, any restrictions on the current drawn, etc. I think that the experience that has been presented here seems sufficient from my perspective, i.e. if I was going to use a sensor system, I think there's enough answers.

Is there a better way to do "it"? Again, is the question about an off the shelf system, or a custom sensor system like Shed proposed (which should be eminently tweakable) or the short detecting systems that I find pervasive in the US, and I personally have a lot of experience with.

What you need to consider that's not simple, again I would limit my personal comments to the short detecting system, and I believe I already answered that question.

Arbitrary phases, different boosters, that is all much more fundamental DCC knowledge, and you might want to understand those too first. No offense but adding that to your autoreverser function is sort of like asking what is the 1/4 mile performance of your car on the planet Mercury. Clearly you want to know more about just being on Mercury before doing burnouts.

I'd be happy to participate on a thread about mixing boosters/multiple boosters because there are some gotchas, and the issues of "smart" boosters and boosters that can interact with the command station come up.

The tunable parameters in the short detection systme are normally just current trip and delay, the program analyzes the "short" to determine if it needs to reverse. Other than functionally, the designers keep their algorithm pretty close to the chest. I recently found out something about the PSX-AR reversers from a person who got a private message from the designer. It explains a lot about what the system does, but it is not published.

I already answered about a reversing loop within a reversing loop, that is "adjacent". Some units have the capability to handle and some don't.

So, please proceed, but with all you want to know, it will be an "article" or perhaps even a magnum opus. That's great. Perhaps it will become reference matter for people who want to know more than "just use this and it works".

Having fun following this, really.

Greg

 

[Andrew_au]

Arbitrary phases, different boosters, that is all much more fundamental DCC knowledge, and you might want to understand those too first. No offense but adding that to your autoreverser function is sort of like asking what is the 1/4 mile performance of your car on the planet Mercury. Clearly you want to know more about just being on Mercury before doing burnouts.

But let me try a couple of burnouts (pun somewhat intended).

Classic layout model has a P rail and an N rail (to borrow Roco/Zimo terminology), which propagate the opposite phases/polarities of the DCC power/signal. A "reversing section" (I'll avoid the word "loop" for completeness) provides a section of electrically isolated track that can switch P/N and thus can support situations (such as a wye or a reversing loop) where the P and N rails would otherwise fail to be distinct.

Thus, all of the following are electrically (mathematically?) equivalent:

1. reversing loop with common mainline (e.g. shown in post 12)
2. standard wye junction with reversing "loop" on one branch of the wye
3. standard wye junction with reversing "loop" on one stem of the wye (and nothing on the other end)
4. 3 track sections, with left and right connected but wired together but with opposite rails and the middle being a reversing loop
   • top rail: ---P--- | ------ | ---N---
   • bottom rail: ---N--- | ------ | ---P---

Question: what if we take case 4 and run the two halves off two boosters (driven by a single control station). Does putting a reversing loop between boosters add any extra complexity than a standard reversing loop? I assume a sensor-driven loop doesn't care - one end functions as any booster handover zone, but the actual switching is handled by the sensors. Does having one end be both a reversing zone and a booster crossover zone create issues for a short-circuit system?

(one can work around this by creating a short zone that is between the reversing zone and the next booster zone, but that adds complexity)

Question: are there any issues with driving the reversing zone off its own booster (distinct from both ends)?

Comment: one thing particularly going for short-circuit detection mechanisms for reversing loops is that they can handle an arbitrary number of ingress and egress points with no extra complexity. You only need a single pair of wires to the reversing zone and the reverser only needs to consider the interface between the reversing zone and whatever transition is currently being crossed. In contrast, a sensor based system needs to consider the P/N orientation at each interface in order to wire the sensors correctly.

Example: reversing loop on stem junction of wye with another with more track outside the loop.

One could use short-circuit reversing loops as "safety zones" between different parts of a layout. Rather than ensuring that zone A and zone B have the correct phase, just drop a reversing loop in between and let it figure it out as required.

Questions specific to sensor systems​

Does a sensor-driven system have any mechanism to auto-detect polarity? For example, consider the circuit shown in Post 12. If I swap the wires to the reversing section, can the system adapt automatically? Or does it get the polarity wrong and cause a short as the train crosses the junction?

Does it make a difference whether I use "sensor tracks" or "track contacts". A "smart" sensor system could read the phase through the track contact, compare it to the phase in the output, and switch accordingly. A track contact system has no phase information until the wheel shorts the tracks, and if you're relying on that why not just use a short-circuit system anyway?

 

[Greg Elmassian]

"Question: what if we take case 4 and run the two halves off two boosters (driven by a single control station). Does putting a reversing loop between boosters add any extra complexity than a standard reversing loop? I assume a sensor-driven loop doesn't care - one end functions as any booster handover zone, but the actual switching is handled by the sensors. Does having one end be both a reversing zone and a booster crossover zone create issues for a short-circuit system?"

OK, so you are powering the "main line" with one booster, and the tail with another?

Without a diagram, not sure what you mean.

So without an autoreverser this does not work... clearly.

What do you mean put an autoreverser between 2 boosters? Literally it sounds like you would be connecting two boosters with the autoreverser input connected to one, and the autoreverser output connected to the output of the second booster.

Makes no sense to me. Can you explain better?