Digitrax AR1 issue

[Spaceman Spiff]

Still on the road. I'll try to get track plan out soon. The Kato crossovers should be isolated as I do have insulated rail joiners. The single motor crossing causes the problem also. Like I said what is weird it will enter the loop fine but not exit.

 

 

Spiff

[KenS]

Watch the video and you will see the train shuts down as each bogie passes over the red tack marker. The AR flips, the power restores the motor car pulls and the next bogie trips it again.

 

I think that's it.  It's acting like you've swapped polarity at that point, but instead of reversing polarity when the bogie bridges the two track sections, a short is occurring.  Since you've swapped the AR1 for one that works elsewhere, it has to be either a wiring problem or a failure to detect the short due to high resistance. 

 

1. Re-check that the AR1s "to reversing section" outputs connect to the section of track between the two red pins (the loop).

 

2. Check resistance in the wiring from the power supply (command station) to the AR1. If the track is part of the supply line (i.e., if the AR1 is wired to "normal" rails the way its instruction booklet shows) you need to be measuring that.  High resistance can prevent a short from being detected as it reduces the rate at which current rises during a short.

 

3. Check resistance from the AR1 to the loop (the feeders to the loop).

 

4. Check both sets of track gaps to ensure that both rails are cleanly gapped and something fills the gap (otherwise one rail can expand into the other and cause a short without a train).

 

5. This seems obvious, but make sure there isn't another feeder on the loop you've forgotten about.  I've done things like that...

 

I'd ignore the crossover if the train isn't near it when the problem occurs.  If part of the train is on the crossover then that may be where the short is occuring (and the AR1 won't have anything to do with it).  If you think the crossover is a potential, more things:

 

6. Some switches make a point rail the opposite polarity to the rail it's adjacent to when it's thrown away from the rails, and a bent rail or out-of-gauge wheelset can bridge the point rail with the running rail and short it.  Inspect all wheels with an NMRA gauge to make sure they're spaced properly, and inspect the point rail gap to make sure its not too close.

 

7. If it's an electrified frog, check that it's set to the right polarity (matching the rails the wheel passing through it will use).

 

You said it was a Kato crossover.  Is this the double crossover (four Kato switches in one molding)?  That has the potential for #6 (see my diagram for the through route polarity on this page), although I've never seen it cause a problem on mine. I don't think it could cause #7.

 

Note that the Kato crossover needs the inside rails fed from both sides. It's isolated on that rail, and against the two rails at the crossing point.  But the outer rail isn't isolated, it's a solid rail from end to end.  I don't think that could cause this kind of problem though.

[Spaceman Spiff]

Sorry for the poor quality but here is the track plan for the area in question. I had to use an iphone to take the image.

 

Thanks

 

Spiff

[inobu]

Spiff,

 

The drawing gives us more information but the exact wiring configuration is required as it (wiring) could be root cause.

 

One question that comes to mind is where is the AR is getting its power from. The AR is basically a smart feeder. It takes power from the mainline/booster feed and transfers powers the reverse loop.

 

A switching action must occur either before a train enters a loop or after. This switching action sets the conditions of the loop. Either the loop portion of the AR is in phase or out. The wheels of the first current bearing car will identify the condition.

 

If the main line and loop are in phase the AR will do nothing and the train enters the loop. As the train circles around to exit the loop, the rails must be switched or a derailment will occur. This switching action also creates an out of phase condition between the main rail and return loop. This is why the AR should flip only once per rail switch.

 

When the wheels bridges the gap a short is created which triggers the AR to flip in phase. THe job of the AR is to make sure that the return loop is always in phase with the main line. That is achieved by having the main line as the power/reference and source.

 

The video shows power loss each time a bogie enters the gap. That means that a power flip flop is occurring.

 

So its hard to tell what is going on without a complete diagram. We need to know where the AR's reference power source is wired to and what is feeding it. That why the switches and crossovers are important. They can act as relays changing polarity up line based on its condition.  

 

 

This is from Digitrax

 

The reverse loop must be long enough to handle your longest train. This is especially true if you have lighted passenger cars or freight cars with resistance wheelsets. When the train exits the reverse loop section, the autoreversing device senses both the reverse section and the section of track which the train is entering and matches the phase of the DCC signal for both sections. This sensing function may need to be adjusted to prevent false readings by the reversing device.

 

There can be only one train in the loop at any given time. Because the autoreverser is sensing as a result of the train's movement into a new block, there can only be one train at a time to avoid false readings.

 

You may have to experiment with insulating gap locations. Although the track gap locations shown on a track plan are probably in the right locations, you may have to experiment. Many track plans were drawn at a time when toggle switches were the only answer. In most cases, these reversing sections were drawn as small as possible to minimize operator inconvenience. With DCC, you may need to expand the size of the reversing section to accomodate your longest train. See the example below:

 

 

 

 

In this example, the original reversing section was limited to the area between the turnouts. Once the railroad was in operation, it was discovered that trains with lighted passenger cars would "confuse" the AR1, so the reversing section was expanded a bit. Since the AR1's operation is automatic, it didn't matter if trains were operating on the reversing section or on the normal mainline track. With the defined reversing segment made a bit larger, trains with lighted passenger cars would be completely inside the reversing segment before the AR1 operated.

 

In another situation, traditional model railroaders "see" two reverse loops when, in fact there only needs to be one. Consider this situation:

 

 

 

[Guest Closed Account 1]

Are the reversing units wired exactly alike?

 

Meaning is the common rail wired to the same point on the reverser or do both units fight each other?

[KenS]

Some things you could add to the diagram:

 

1. Any other gaps.  Also, if you use "common rail" wiring (one rail not gapped) then for each location, note which rail has the gap, inside (closer to the middle of the loop) or outside.

 

2. Any other feeders (including the first ones to the right of the crossover).

 

3. Feeder "phase" per rail.

 

4. What are the inputs of the AR1 (Track A and Track B contacts on the circuit board) wired to? An adjacent block, or a track bus direct from the power supply (Inobu asked this question earlier)? I'm presuming the outputs (the "out to reversing section" wires) go to the feeders in the loop.

 

For #3, what I mean is which bus or supply wire connects to which rail in each location.  Say that your power supply is putting out two lines we'll call RAIL A and RAIL B and you use Kato feeders with BLUE and WHITE wires.  For each feeder location note if BLUE or WHITE goes to the outside rail, and if it is connected to A or B.  From that and the location of gaps relative to the switches we can tell if there is a point where a phase mismatch is occurring.

 

Webskipper raises a good point, use of common rail wiring can add complications.  Even with common rail, both rails in the loop must be gapped at both ends and connected only to the AR1 outputs.  This is implied by the diagram in the AR1 instructions, but not clearly stated. 

 

Additionally,  if two adjacent blocks are both controlled by AR1s they're going to fight each other. Note if any of the feeders mentioned above connect to a different AR1.

[Spaceman Spiff]

I will have to investigate further when I get home.  The AR1's are getting power from the main bus wires under the table and not directly from the rails. Should I change this? In regards to the gaps both rails per line (main and siding) have gaps.

 

Spiff

[inobu]

Spiff,

 

 

I think that is your problem. The AR needs to be tapped off of the rail as the rail can change polarity by way of the switch or crossover. Whats on the bus verses whats on the rail can be opposite depending what you have switched up line.

 

Inobu 

[Guest Closed Account 1]

Click below for AR1 Instructions

 

http://www.digitrax.com/ftp/ar1.pdf

 

All 4 loop rails are gapped. See the diagrams.

 

The AR1 is usually located close to the point where the gap is cut for the revers- ing loop. The length of the reversing loop is determined by the maximum length of the train that will use it.

1. Turn off track power. (Note: Failure to turn off power before connecting

your AR1 may damage your AR1.)

2. Connect Track Power from Rail A and Rail B mainline to terminals 4 and 5 on the AR1. (Figure 1)

3. Connect the wires from the reversing section to terminals 1 & 2 on the AR1. (Note: terminal 3 is not used.) (Figure 2)

 

Think we figured it out.

 

Polarity of the tracks changes once the Ar1 is involved so it is smart for the Ar1 to know what the polarity is on the mainline before it does its voodoo.

[Spaceman Spiff]

Thanks for the help guys. I really appreciate it. I will rewire the two loops to get power from the rails as opposed to the bus lines. It's really frustrating as my other loop works flawlessly and the loops in question work some of the time.  I shot this video with my iphone so apologies for the quality.  The trains ran fine during the video. I ran the trains 10 minutes later and they kept stalling as previous videos.

 

 

 

 

Spiff

[KenS]

I think that is your problem. The AR needs to be tapped off of the rail as the rail can change polarity by way of the switch or crossover. Whats on the bus verses whats on the rail can be opposite depending what you have switched up line.  

 

I'm not convinced.  A DCC reverser typically works by detecting excess current flow in a short.  There's a short if a wheel bridges two rails at different phases, regardless of where the AR1 is getting its power from.  There's more going on here.

[inobu]

[smg id=1424]

 

 

This is why/how the crossover can effect the AR if the conditions are right (in the bad sense).

 

If the power into the crossover (top) BA would enter into the lower switch by the bridge a polarity switch would occur. Because the AR reference is AB it will always detect a short.

 

When the first car hit the AR it will flip, because the AR is referencing the bus, the AR will revert back to the AB comparison. When the next lighted car hits the gap, What is on the wheels? BA, What is the AR reference AB. whats is going to happen, Trip.

 

 

The same thing holds true if that top crossover is straight through. What ever the configuration is beyond the crossover or switch will be propagated to the rails at the gap.

 

In the diagram you can see that if the inner rail (Bottom) of the crossover is thrown the AB values will be present at the (top) gap which is a polarity match and there will be no AR trip.

 

 

This is where the problem lies. The AR does not truly know what is on the rail in all actuality the AR is looking at the bus. 

 

Inobu  

[KenS]

If the power into the crossover (top) BA would enter into the lower switch by the bridge a polarity switch would occur. Because the AR reference is AB it will always detect a short.

 

I think you've diagrammed the inputs correctly (if and only if there are feeders between the crossover and the gaps, otherwise theres a dead rail for "A" outside the gaps, which could be our problem).

 

But I don't think you correctly understand the operation of the AR.  It's not testing phase on the rails of its output side against a reference phase on its input side.  There is no phase on the rails (without a train) except what it puts there from its input, so the output will always be in phase with the input.  What it's sensing is the current flow that occurs when a train shorts some out-of-phase rail with its output, causing current flow through the AR from the input (whatever it is).  The source of the adjacent rails and its relationship to the ARs input is irrelevant. All that matters is phase at the point where a wheel bridges the gap, and that's either in phase (and nothing happens) or out of phase (and an inversion happens on the AR outputs to bring them into phase).

 

Digitrax even notes that one of the adjacent sets of rails can be powered by a completely different booster.

 

Problems occur when the sensing fails (high resistance prevents the short from raising current fast enough or when the trip current setting on the AR is wrong) or when it tries to match against two separate and conflicting events (a train bridging both ends of a too-short reversing section).

 

When the first car hit the AR it will flip, because the AR is referencing the bus, the AR will revert back to the AB comparison. When the next lighted car hits the gap, What is on the wheels? BA, What is the AR reference AB. whats is going to happen, Trip.

 

You're assuming the AR reverts to original phase the instant the short goes away. That's possible, but I wouldn't expect it to work that way as it would prevent any vehicle with multi-wheel pickup from working at one end of a reversing section that wasn't part of a closed loop.  And Digitrax's documentation clearly suggests it will work for those.

 

The same thing holds true if that top crossover is straight through. What ever the configuration is beyond the crossover or switch will be propagated to the rails at the gap.

 

Just a nit, but the Kato double-crossover doesn't propagate anything on Rail A.  That's one of its annoying (but sometimes useful) features.  Rail A is insulated no matter how the thing is set. And Rail B is solid metal and always propagates the same track from the far side of the switch, even if the switch is thrown and the other side is different.

 

BTW, Digitrax's TSB has an article that suggests that repeated switching by an AR indicates that the trip current is set too low.  Since we've eliminated the AR itself (and thus the cause of that problem) that implies that there's a wiring problem restricting current flow during a short (which also prevents detection).  That can be:

 

a. Too few feeders in a long reversing loop (not the problem since the error occurs near the AR1 output feeders).

 

b. Track outside the loop being fed through switch points rather than directly via a feeder (switch points can be a high-resistance path).

 

c. Loss in bad rail joiners (which shows up as high resistance across a unijoiner if you check with a multimeter with feeders disconnected)

[Spaceman Spiff]

I can't believe the willingness of the folks of the site to go out of their way to help. That's what makes this a great site. Thanks to everyone for their help.

 

I removed the track and put down new track and also rewired the power leads to come right from the rail as opposed to bus lines (my lower loop which runs with no issues is powered from the bus lines so I left it as is). It now seems to be ok. I will run it abit more to see what happens. In regards to the crossover in the diagram I have feeders on both sides of it.  Using inubo's labeling the rail polarity is:

 

A

B

A

B

 

not

 

B

A

A

B

 

Not sure if that makes a difference or not.

 

KenS mentions feeders for the loops. Using the AR1 output, can I have 2 or more feeders to feed the loop or should I just use 1 feeder?

 

 

Spiff

[inobu]

Spiff,

 

You responses clarifies you configuration and suspect components are eliminated. That's the only way to nail it down.

 

 

KenS your comment

 

"It's not testing phase on the rails of its output side against a reference phase on its input side."

 

The test is written in the trip result script in the PIC.

 

in all actuality that is what it is doing.

 

For example:

 

Input rail 1 = positive

Input rail 2 = negative

 

(The input should never change)

 

The AR will make output rail

 

output rail 1 = positive

output rail 2 = negative

 

a trip means that

 

 

output rail 1 = negative

output rail 2 = positive

 

The AR will make 

 

output rail 1 = positive

output rail 2 = negative

 

The input will never change but a polarity flip will occur based on the position of the switch.

 

Here is what it looks like.....

[smg id=1425]

 

 

This is from Digitrax.

 

The reverse loop is managed by an AR1 Automatic Reverse Controller. When a train enters into the reversing segment, the AR1 senses track power phase between the reversing section and the rest of the railroad and then automatically matches the two.

 

http://kb.digitrax.com//index.php?q=1313

 

There is no phase on the rails (without a train) except what it puts there from its input, so the output will always be in phase with the input.

 

You are forgetting that if the switch is thrown off the main line, the inner rail on the main is connected to the outer rail of the loop this is a polarity switch and the AR has to flip the polarity on its reverse loop.

 

Switches are the polarity shifters. Crossover can do it too on the top rails (outer)

 

Inobu

[KenS]

Inobu, I think I might understand what you're saying.  But I still think you're wrong. 

 

I wasn't talking about the external rails being flipped.  But even if that were the case, it would be a matter local to one end of the reversing section (except in the one case where you have track that loops back on itself through a single switch, which isn't the case here). 

 

An auto-reverser has to be able to detect a phase swap at either end of the reversing section (well, I suppose someone could make one that synchronized with only one end and only worked for loops, but that's not what Digitrax's documentation says they did).  The only way I'm aware of that it can do that is by sensing current flow when a wheel (or car with multi-wheel pickup) bridges the rails across a gap.  And that just depends on the phase being inverted; the cause of that inversion might be local, but it can detect it no matter where its inputs are wired, as long as that's a valid copy of the track DCC, even if it's coming from a completely different booster from the adjacent track (as Digitrax says it can).

 

Since Spiff seems to have fixed the track, whatever the problem was, there's no point in discussing this further here.  If you want to discuss it, PM me and I'll draw some diagrams to make my points clearer and PM you back.

[Spaceman Spiff]

KenS, using the AR1 output, can I have 2 or more feeders to feed the loop or should I just use 1 feeder?

 

 

Spiff

[KenS]

You can take the output of the AR1 and split it as many times as you want, as long as it only connects to the rails of the "reversing section" (the part of the loop between the gaps).  If you want to run a separate feeder to each piece of unitrack you can (some people recommend that for permanent layouts built with sectional track, as it eliminates all long-term problems with rail joiner contact)

 

Unless the sectional track is permanent and inaccessible, that's overkill. For flex track at most you need a feeder to each 3' (or 1m) section. And barring any serious probelms in the rail joiners, the usual rule of thumb for DCC is that you need a feeder at least every 6'. Plus you can just solder the rail joiners in between feeders if you want to ensure good contact.  You can use a larger feeder spacing, but power loss in the rail starts to become a problem after that (nickle-silver rail is a terrible conductor compared to copper wire).

 

In my own experience, even 10' of good unijoiners isn't a major problem, but one bad Unijoiner can change that and 20' of even good unijoiners will show a very noticible drop in train speed due to voltage loss in the rail.  You can find bad unijoiners by testing with a multimeter on either side of a joint with the feeders disconnected. If resistance across the joint is more than 1-3 ohms, the joiner is worn and should be replaced (if replacing it still doesn't get resistance below about 5 ohms, it's the track, but 5 ohms is still okay; the really bad ones have resistance significantly above 10 ohms).  The same logic likely applies to other kinds of sectional track rail joiners, but I never tested those.

[Spaceman Spiff]

Thanks for the info and help. I'll some more test runs to make sure it's all good before adding more feeders.

 

Spiff

[KenS]

Make sure you test the crossover in both "through" and "crossing" arrangements and run a train through it on all four paths.  If the feeders aren't right there, you'll either have a short or a dead section.

 

BTW, when I do this kind of testing, I use an old loco I picked up on a bargain table so if I fry it or its decoder, no harm will be done to an expensive model.

[Spaceman Spiff]

Thanks for the tip.

 

Spiff