Signal Help

[rkcarguy]

I wonder if actual water coverage was a problem on some tracks, I could see issues if the track ran through a low spot that actually flooded it.
I'm on a pretty decent hill, and I won't be able to have any standing water or it will quickly turn to mud and swallow the track. Kind of thinking I'll use 5/8 minus gravel, weed cloth, lay track, then 5/8" clear infill. A few spots are going to need ditches and some pipes to give the rainwater a place to go under the ROW when it pours.

[Steggy]

I don’t know about creosoted ties, but the ties on the 7.5” railroad I maintain signals for are pine that’s treated to below grade spec with the old CCA, copper chromium arsenic, treatment.

There are many ties treated that way at our railroad. A sample of new and dry ones about 15 years ago revealed a resistance range of anywhere from 75,000 to 100,000 ohms in free air. Measurements were made by driving a galvanized roofing nail into the butt end of each tie to act as an electrode and applying low voltage from an adjustable power source for 10 minutes. Before disconnecting the power supply, the current draw was measured to compute the resistance.

As part of this research project, some ties were made up into a piece of track whose length was such that it could be placed into a bathtub and submerged to the rails in water. The track's tie count was such that it would amount to 260 ties per 100 linear feet. The tub was filled well water (approximate hardness was 13 grains) and the track assembly was soaked for one hour. During that time, regulated voltage was applied to the rails to simulate an unoccupied block under power and the current draw was monitored.

Based upon that test and after performing some extrapolation, it was concluded that 100 feet of track of the same construction in typical wet conditions would have a resistance of approximately 250 ohms. In an actual track circuit, that 250 ohms would be in parallel with the track relay coil, and that parallel combination would be in series with the track circuit ballast resistor.

In a 5 volt track circuit, the relay chosen for the application, an Omron G5V1-DC5, has a DC coil resistance of 167 ohms. The parallel combination of the relay coil and "worst-case" track resistance (250 ohms) would be 100.1 ohms. It was desired to limit the block occupied (short circuit) current to 0.5 amperes to operate the power supply regulator well below its maximum rating. Therefore, the ballast resistor would be 10 ohms.

Given all that, the track circuit can be treated as a series combination of a 100.1 ohm resistor (parallel combination of track leakage and relay coil) and a 10 ohm resistor, resulting in a open circuit resistance of 110.1 ohms. Such a combination would result in an open circuit (block unoccupied) current flow of 0.045 amperes, making 4.55 volts available to drive the relay. The G5V1-DC5 has a minimum pick voltage of 80 percent of nominal at 25 degrees C, which means the pick voltage is 4. Therefore such a block would clear.

Extrapolating to 200 feet, the parallel combination of track leakage and relay coil would be 71.5 ohms. Placed in series with the 10 ohm ballast resistor, open circuit current flow would be 0.061 amperes, leaving 4.39 volts to drive the relay. Again, the block would clear, despite the track being wet.

Further extrapolation suggests that a 300 foot block should clear when wet. At that length, 4.24 volts would be available to drive the relay. However, the pick voltage rating for the relay is at 25 degrees C ( 77 degrees F). As relay temperature rises, pick voltage will increase—and the same behavior affects the ballast resistor to some extent. so the 300 foot block might not clear on a 90 degree F day immediately after a rainstorm.

The point to all this is while the tie treatment chemicals do cause some conductivity, the resulting track leakage is manageable with proper track circuit design and good track ballasting techniques that promote drainage. Plus results tend to improve with lower track circuit voltages.

We normally don’t have issues with the gates false activating in the rain but during torrential downpours water will short across the rails and get them to activate.

A flow of water in itself should not cause false activation. Pure water is an insulator, not a conductor, and rainwater is very close to being pure from an electrical conductivity standpoint. Minerals in the water, which are absorbed as the water flows over the ground, are the source of conductivity. However, given the relatively small rail area in contact with the water flow, I have some doubt that that is the problem.

At what voltage are you operating the track circuit?

[rkcarguy]

I've found water to conduct quite well, don't know if I believe the insulator part. Either that or just most of the water around is not pure. Water got into my switch box on my lake boat and all the lights were stuck on until it dried out and then I resealed the box.

[Atkinson_Railroad]

Water does not conduct electricity. Impurities, or other substances combined with water allow water to pass “potential”.

You can prove it for yourself on your kitchen counter using a battery, suitable connections, a light bulb, salt, and a glass of water.
As you add salt to the water, the light bulb will begin to light.

A fascinating aspect to the experiment I recall is how one of the conductors immersed slowly blackens in color,
and the other slowly turns a brighter copper color. The water turns an ugly gray mess.

I vaguely remember the experiment from 5th or 6th grade.

Adding salt to a barrel of water is one of the oldest tests related to testing a generator output.
Salt is slowly added to the water to create a load on the generator output.

John

[ChuckHackett-844]

.... the track is 7 1/2" gauge aluminum rail on concrete ties, two ties per foot, crushed limestone ballast, and my longest block is about 400'. As far as the 40ohm number that is what I measured when the track is wet (it actually was 50 ohms but I was trying give a little margin) much higher when dry. I am trying to make my signals more reliable when wet. If I understand your calculations and comments it is not possible.

Sorry I forgot to address the five second delay, ocasionally I have trains run with very rusty wheels and the signals have a difficult time with them, the delay would keep the signal from blinking and also give the trains a little extra time to clear the block.

The controllers I described in my post concerning "Solid-State Viability" would easily handle these conditions. If you were just interested in a "Track Detector" you could use the controller to create an eight-channel track detector by simply configuring the track input (with an up to 15 second release delay) to be echoed on one of the PWM outputs or attaching an add-on relay board. This could be inserted into any existing signal system.

Not really using it up to its potential but you would get reliable track detection.

I looked into building a compact "Track Detector" (perhaps 2 or 4 channel with uncommitted relay contacts as the output) using the same train detection techniques used in the controller but I didn't think there would be enough demand for it. I could be wrong, let me know.

[ChuckHackett-844]

.... I am currently analyzing a new block occupancy detector design that will operate the track circuit at 3 volts DC, with a short circuit current of about 0.75 amperes. ....

Yikes! With our close to 100 sampled track segments that would be 75 amps @ 3v = 225 watts. Granted, not likely to have every track segment occupied but ...

At a typical meet we might have say 25 trains running, assuming that there will be many times when trains are each spanning two detected track segments (we also divide blocks and sidings into two segments to refine train location on the track display and allow the system to more accurately determine train direction so the track segments are somewhat shorter than average).

This means 50 shunted track segments, not infrequently, there might be a load of 37.5 amps @ 3v = 112.5 watts. This would require us to run either an additional conduit around the railroad (by code can't mix line and low voltage circuits) to supply 120 vac. That, plus the step-down transformers, etc. starts to add up to significant money.

We currently run the entire railroad with 16 volts @ 2 amps = 32 watts ... and that includes 60 signals that always have one lamp on.

I must be missing something here ...

[ChuckHackett-844]

Delay-on-clear timing is generally more useful with aluminum rail, but is also of value in maintaining block occupancy through turnouts and diamonds. Generally, a time-out period of about one second is more than sufficient for uninterrupted blocks. Reasonably reliable detection in a turnout could require five to eight second delay-on-clear to minimize the chance of a false clear failure.

I recommend that you do your utmost to NOT have 'dead' track segments no matter what "delay on clear" you have. Delay on clear should only be used to cure very intermittent rail contact. I recommend not using a delay on clear (what I refer to as "release delay") any longer than 1 second.

The reason for not having 'dead' segments (crossings, turnouts, etc.) is that if someone is running something like a beam engine or box cab electric and they stop to chat with someone on a crossing or, more likely, derail at a switch, the signal system will declare the block cleared when it is not clear and cornfield meets will result.

It's fairly simple to insulate a turnout in such a way that it does not have any 'dead' sections. Even concrete crossings can be addressed by using a concrete saw to slice down through the welded steel spacers typically used at 2 to 3 foot intervals to hold angle-iron, etc. while the concrete cures.

A signal system should never use a timer function to clear the block because it leaves the block in an unknown state. There have been attempts to do axle-counting to address signal systems on welded-tie track. People invariably include a "timeout" to clear the signals in the event they do not auto-clear by a correct axle-count leaving the block. If a car comes uncoupled in the block or a train derails and the signals get cleared by the timer and you come around the corner ... guess what ...

Even on some mainline routes in Europe they have some track segments that use axle-counting rather than track circuit (I forget why). Sometimes there can be a miss-count (e.g.: less axles leaving than came in). In this case, it is required by rule that a inspection vehicle or train at dead slow traverse the entire track segment before it can be reset to the clear condition.

[Steggy]

I recommend that you do your utmost to NOT have 'dead' track segments no matter what "delay on clear" you have. Delay on clear should only be used to cure very intermittent rail contact. I recommend not using a delay on clear (what I refer to as "release delay") any longer than 1 second.

At our railroad, the policy is that both stock and closure turnout rails be one uninterrupted piece, which prevents the use of insulated joints. If that policy weren't present it would be a straightforward matter to include all but a small section of the turnout into the track circuit. Not having that feature, we have some locations where longer delay-on-clear is unavoidable to prevent a false clear status when a short, slow-moving train is passing through a turnout. Delay-on-clear is adjusted to the minimum required, usually around 0.8 seconds on blocks that are uninterrupted by turnouts. In practice, it has worked out well.

The reason for not having 'dead' segments (crossings, turnouts, etc.) is that if someone is running something like a beam engine or box cab electric and they stop to chat with someone on a crossing or, more likely, derail at a switch, the signal system will declare the block cleared when it is not clear and cornfield meets will result.

You are preaching to the choir, Chuck.

A signal system should never use a timer function to clear the block because it leaves the block in an unknown state. There have been attempts to do axle-counting to address signal systems on welded-tie track. People invariably include a "timeout" to clear the signals in the event they do not auto-clear by a correct axle-count leaving the block. If a car comes uncoupled in the block or a train derails and the signals get cleared by the timer and you come around the corner ... guess what ...

You may be misunderstanding. The timer itself doesn't clear the block. What it does is delay the clearing of the block until the track relay has remained picked a certain amount of time. As long as the track relay is dropped the block occupancy detector will remain in the block-occupied state. Furthermore, the timer is reset to zero each time the track relay is dropped, even if only for a few milliseconds. Block occupancy detection works exactly like the full-sized equivalent, in which vital relays are dampened (in many cases, pneumatically) to produce a "lazy" response when the last wheelset leaves the block.

[ChuckHackett-844]

You may be misunderstanding. The timer itself doesn't clear the block.

I think we are saying the same thing. In my case (software rather than relays) the 'delay' is reset each time a shunt is detected (even for an instant). The track is not considered "unoccupied" until the delay runs out without any detected shunts. To be precise: the delay is on the track going from occupied to unoccupied, the signal is cleared because the track went unoccupied ... after the delay

[Steggy]

You may be misunderstanding. The timer itself doesn't clear the block.

I think we are saying the same thing. In my case (software rather than relays) the 'delay' is reset each time a shunt is detected (even for an instant). The track is not considered "unoccupied" until the delay runs out without any detected shunts. To be precise: the delay is on the track going from occupied to unoccupied, the signal is cleared because the track went unoccupied ... after the delay

Okay, we are now singing out of the same hymnal. ♫♫♫

[rkcarguy]

BDD, why does the real system "shunt" instead of having the train complete a low voltage circuit?
My experience with cars and fuel injection drives me to ask this question. Most circuits in a cars ECU are closed by the ECU grounding the coil of a relay, the fuel pump for example. Or sensory circuits like air intake temp, o2 sensor, throttle position, and coolant temp take 5 volts in, and provide a lessor variable voltage in return to indicate a value or position to the ECU.
I think very outside the box at times, and deal with a lot of people in my industry whose answer is no more than "because that's the way we've always done it". I even look at full size trains, and see that being the axles and wheels are one piece, there is scrubbing going on when a train goes through a curve as one wheel wants to go faster than the other. Seems like wear would decrease and efficiency increased if each wheel ran separately on their own bearings.