Tehachapi Loop II

[ednadolski]

60 cars is "short"! :Dn

By Tehachapi standards, anything that fits on Walong siding is 'short'   That there is the whole motivation for the current double-tracking project.

Really tho this is one of the great benefits about N scale.   Can you imagine, trying to build an equivalent helix in HO?   (Maybe a club could do it, but for me it would not leave much space in the room for any kind of layout  )

I agree with the Gary about the 1' sections.   (Heck, maybe 6 or 8 inches is enough).  On our club layout we have set up 1' detection sections in front of each signal for the eventual purpose of automatic train running (which hasn't actually happened yet).    Generally this means a 1' detection section at the end of each block.

Do you use that with hidden staging tracks as well?

FWIW, we've also divided the remaining lengths of the block in two, the idea being that when a train gets half way through a block it will start slowing down to a crawl so that when it reaches the end section at the signal it will creep to a stop realistically.   The 'half-block' sections also come in handy for 'early tumble-down' when going from sidings to single track.    Not that any of this necessarily applies to your operational scheme, unless you also like the idea of automatic running...

I am intrigued with the idea of automatic running,  but I am not sure I want to take on the effort of implementing it (at least at the beginning).  However I do hope to avoid unnecessarily setting myself up any constraints that would preclude it.

Ed

[GaryHinshaw]

Looks like a good plan Ed.  Love the ascii art too! - though I suspect you could have banged out a jmri panel in less time.   

Before you commit to too many über-blocks, you might want to verify that 39'+ trains will work reliably enough to be "fun". 

BTW, jmri has provisions for automatic running if you have fully detected track-work.  Just fyi.

[ednadolski]

though I suspect you could have banged out a jmri panel in less time.   

I did start on that but found it counter-intuitive.... time to RTFM.... 

Before you commit to too many über-blocks, you might want to verify that 39'+ trains will work reliably enough to be "fun".

I'm not sure of a reliable way to verify, short of actually building it.   So I guess I am taking a leap of faith.

I do have some basis for confidence from running the 64-car train on my current Loop.  That runs pretty reliably with the body-mounted LEZ couplers and all-metal wheels.   The limiting factor on train length is the 16-inch radius S-curves thru East Walong, and at 34.5" radius the new Loop is not anywhere near so tight.  While the curvature in the helix is tighter (25.5" - 30" radii), the grade is also less (1.8% - 1.5%).  (I suspect that for most N-scale trains, the drag from excessive curvature is as bad or worse than that from the actual grade itself.)

I could hedge by gapping Track 4 something like this:


    ---|-1'-|----25'---|-1'-|--12---|-1'-|---12----|-1'-|--25'-|-1'-|---


which would let me reconfigure the detection for three short trains, if the mega thing does not work out so good in practice.


Ed

[ednadolski]

* For hidden staging, I would not count on current sensing detection to be 100% reliable about sensing freight cars at the end of a train.  It is very reliable for locos however, ...

I'm wondering, why should freight car current sensing be any less reliable than locos?  Is it something inherent, or is it more a matter of implementation, maintenance, or other such issues?

Seems to me that unreliable freight car sensing would also be a significant issue for signalling systems.  Otherwise, there would not seem to be much point in equipping cars with all those resisor wheelsets 

Ed

[GaryHinshaw]

I don't have enough experience base with that yet to give you a definitive answer, but I think it mainly boils down to locos having more weight and a lower resistance across the rails.  With freight cars, the contact resistance between wheel and rail is more problematic.  When I first installed a bunch of FVM 10K wheels a few months ago, I had to polish the treads to remove the blackening to get reliable detection, and that worked quite well.  But since then, I have had a few dropouts in short blocks (e.g. turnouts) where there might be only one active axle in the block at any given time.  In longer blocks with many active axles, it is still very reliable.  Of course, clean track is also a must, and there are also several ways to tune the detection settings such as current thresholds and time-out settings (e.g. remain active for N seconds after the current drops to zero).  I have not played with those settings very much yet.

BTW, if you have run 64 car trains around your current loop with no issues, I think you should be fine with 100 car trains.   

[ednadolski]

Thanks Gary.   There seems to be general, implicit assumption of one resistor wheelset per car, is that correct?   In that case I can definitely see some potential for contact issues to arise.  Even if all the wheel treads, flanges, and rail are immaculate, in our rigid N-scale freight car trucks it is not a given that all four wheels will remain in continuous contact with the rails at any given time. Considering the inevitable variations in tolerances, curvature, grade, etc. it seems reasonable to expect some occasional drop-outs when there is only one wheelset per car.  I have myself observed that under the right conditions, a freight car actually can roll for quite a distance along some of the (sharp) curves on my layout with both outside wheels of a truck lifted off the rail. The car will not derail (with body-mount couplers, the adjacent cars are holding it in place), but there is no way a resistor wheelset would be picked up by a detection system in that case.  (For the record, I don't normally run trains like that )

With locos not only is there more weight to reduce any lifting or tipping, there are also more wheels which are all conductively active.  So the rigid trucks are not as much of a factor (otherwise our locos would run pretty poorly). Plus, whenever a loco gets dirty enough to affect its electrical performance, we notice the flickering lights and stuttering drive, and we take it out of service for a good cleaning.  Seems a safe bet that doesn't happen nearly so often for resistor-equipped freight cars. 

Perhaps a good compromise is to install resistor wheelsets on all the axles for the last car on each train (on modern trains with a FRED, or on the caboose in an older era train).  A regular cleaning schedule seems advisable too.  It would probably not be practical to do most/all cars like that, aside from the cost and the work it would probably also put too much of a current load on the layout's booster system.

Ed

[ednadolski]

I think you should be fine with 100 car trains.   

Hopefully!      Very likely tho it will only ever be one train that long - I don't think I'll ever have the resources for more.   Anyways that figure really is more about the length of the train rather than the actual number of cars.   I used 4.2" as the average car length, but I expect to have lots of centerbeams, autoracks, reefers, which should reduce the actual number of cars by maybe 20% or so.  (I should test out how well the body-mount couplers work with out with those long cars on the smallest curves I will be using.)

Ed

[jagged ben]

Ed, we do have the 1ft 'stop' blocks in a hidden staging yard, and if the trackwork in the yard ladder were a little better I'd probably use them to run into it 'blind', i.e while looking at the jmri panel.

Also I've run 60 car trains over our version of the loop, which is about 24" radius and 2.5 percent grade.  Only caveat is no autoracks or other cars longer than 60ft.   I think you'll be able to push the limits with you design, although you may need 8 locos to get 100 cars!

[GaryHinshaw]

Thanks Gary.   There seems to be general, implicit assumption of one resistor wheelset per car, is that correct?

Correct.  In my case, an explicit assumption.  For the most part I install them on an inner axle to avoid having resistors possibly fouling coupler draft gear.  It is certainly likely that weight is not uniformly distributed across all wheels of a given car at all times.  But the scheme still works more often than not.

I have myself observed that under the right conditions, a freight car actually can roll for quite a distance along some of the (sharp) curves on my layout with both outside wheels of a truck lifted off the rail.

https://www.therailwire.net/forum/index.php?topic=24108.msg264568#msg264568   

It would probably not be practical to do most/all cars like that, aside from the cost and the work it would probably also put too much of a current load on the layout's booster system.

Back of the envelope calculation: take a 50 car train with 2 active axles per car, giving 100 axles of 10K resistors.  This is a net resistance of 100 ohms, and a current of 100 mA at 10 V (to keep the numbers round).  This is comparable to having one more loco in the train, so it really shouldn't strain the booster.  I think the main challenge is reliably detecting freight cars in short (turnout) blocks.  My next step will be to turn down the current threshold and increase the time-out setting for those blocks.

I think you'll be able to push the limits with you design, although you may need 8 locos to get 100 cars!

!!

[ednadolski]

Ed, we do have the 1ft 'stop' blocks in a hidden staging yard, and if the trackwork in the yard ladder were a little better I'd probably use them to run into it 'blind', i.e while looking at the jmri panel.

Interesting, but I'm not sure I quite understand your reservations in that case. Is there something about the trackwork that lowers your confidence in the detection? 

Also I've run 60 car trains over our version of the loop, which is about 24" radius and 2.5 percent grade.

Hawt!     Love to see any pics/videos that you might have! 

I think you'll be able to push the limits with you design, although you may need 8 locos to get 100 cars!

GMTA:

Mega: 100 cars + 8 engines: 468" (39' 0")  (equivalent proto = 7800')

Generally I find that on average one engine can pull about 12+ cars upgrade on my layout, so that's what I'm extrapolating to get to the 100.  It may even work out to a bit more, with the reduced drag from the larger curve radii. FWIW I recall hearing that the La Mesa club has some sort of guideline of about 2 cars per loco axle, so that seems to match up pretty closely (even in spite of the scale difference).

For the most part I install them on an inner axle to avoid having resistors possibly fouling coupler draft gear.  It is certainly likely that weight is not uniformly distributed across all wheels of a given car at all times.

Yes, at least not until we get sprung/equalized trucks in N scale ..... 

Perhaps another option is, to install the resistors on both axles of one truck.  It seems unlikely that both of them would lose electric contact at the same time (unless the track were getting pretty dirty).   It depends on exactly where the resistors are installed, but offhand I should think that the inside of the wheel flanges would hit a coupler box before a small SMT resistor near the axle could (and perhaps the etched pockets are small enough to make it into a non-issue).   The only other reservation that I could think of with that is, to ensure that the detection gaps give enough clearance to any fouling point so that it does not matter which end of the car has the resistor wheelsets.

Back of the envelope calculation: take a 50 car train with 2 active axles per car, giving 100 axles of 10K resistors.  This is a net resistance of 100 ohms, and a current of 100 mA at 10 V (to keep the numbers round).  This is comparable to having one more loco in the train, so it really shouldn't strain the booster. 

That is in each train, right?   So 10 trains adds 1A, etc., and they are always drawing power for as long as they are in a live detection block.

I think the main challenge is reliably detecting freight cars in short (turnout) blocks.

I presume that the distance between any two detectable elements (i.e. loco or wheelset-equipped car) cannot exceed the shortest detection block (other than stop blocks).  If that is not the case, then what other options are there that will ensure positive detection?   Delays seem rather unreliable, if the train could be undetected when it is stopped.

I was looking over the diagram you posted in this thread and I did a quick estimate of the hardware it would take to implement a current detection scheme with signalling control.   For the 35 detection blocks on my current plan, it will take over $500 in RR-Cir-Kits hardware (plus all the resistor wheelsets, wiring/cabling, actual signals & LEDs, etc.).   This definitely is a commitment, but hopefully it can be implemented in stages.  The most important thing is to get the block arrangements right, and try to avoid any need for big changes after the track & wiring go in.

One thing I notice from the diagram:  The detection relies on a single wire connected serially between each block track and the main track bus, wound thru the detector coil.   Some of the block tracks can be quite long (some between 25' and 38').  Each individual rail in a block should have its own dedicated feeder wire, so as not to rely on rail joints for conductivity, and to avoid voltage drops across long stretches of rail.  Therefore each coil wire has to be a large-gauge wire that can handle the full length run for the block, and all of the individual rail feeders should connect to that (as opposed to connecting to the main track bus).   The return side of the circuit could still use a common bus wire, since there are no coils on that side of the circuit.  Is this a sort of 'standard' approach, or are there better ways to wire long blocks for current detection (esp. that do not require quite as much bus wire)?

Thanks,
Ed

[GaryHinshaw]

That is in each train, right?   So 10 trains adds 1A, etc., and they are always drawing power for as long as they are in a live detection block.

Yes, and it is not restricted to detection blocks, but any powered blocks.  Basically, everything 1000 live axles on powered track will draw ~1 amp (a bit more at 12 V).

I presume that the distance between any two detectable elements (i.e. loco or wheelset-equipped car) cannot exceed the shortest detection block (other than stop blocks).  If that is not the case, then what other options are there that will ensure positive detection?

Yes, the spacing between live axles -- loco or resistor-equipped rolling stock -- must not exceed the length of the shortest detection block for 100% coverage (and assuming current detection is 100% reliable).   The only other options I know of are light blocking system (visible or IR).  This might be a simpler solution for your short helix blocks.  [Aside - I'm using a visible system for my helix detection: my eyes.     I am reserving current detection for signalling & dispatching the main line.]

As for wiring, I did indeed run a third (14 ga.) bus wire for each long block, as in the bottom half of this diagram:



Full-up block detection is not fast or cheap, but hopefully it's worth it!

[railnerd]

Seems like you marked the detector on the wrong "rail" in that diagram if you want to discriminate between the 3 sections of track.  Are you detecting all 3 sections as one block, or trying to independently detect each section.

Of am I misreading this diagram?

-Dave

[jagged ben]

Interesting, but I'm not sure I quite understand your reservations in that case. Is there something about the trackwork that lowers your confidence in the detection? 

No, it's not the detection I don't have confidence in.  I lack confidence that my train will arrive at the end of the staging track in one piece with none of it on the floor.      In other words, I want to know to stop immediately if I see a derailment happening.

Hawt!     Love to see any pics/videos that you might have! 

We'll I've really left myself open to 'pics or it didn't happen'.       All I seem to have are crappy cell phone pics of an relatively early run of about 54 cars and 6 locos.   I definitely have done more with the same number of locos.  (Heck, on a good day I can do 54 cars with no helpers.  See last pic.)   

Anyway here are some pics.  Yes, there is only one freight train in these pics.    (This train had to stop and meet a 'Tehachapi Starlight' detour!  Must have been work happening on the Coast Line.    )
[ Guests cannot view attachments ]
[ Guests cannot view attachments ]
[ Guests cannot view attachments ]

This was only 4 locos, but keep in mind two were Athearn cowl units.  Good pullers.
[ Guests cannot view attachments ]

Note our version of the loop can only hold maybe 40 or so 50ft cars. 

[GaryHinshaw]

Or am I misreading this diagram?

-Dave

Yeah, this diagram is probably confusing when shown out of context.  (Sadly, I can't find the post where I first discussed it...)  In any case, this is supposed to show how I wire for current detection in a long block.  In the top half, "A rail, 3 track sections" indicates the A-side rail of 3 sections of track that comprise one block, each with their own feeders from the main bus.  I don't use rail joiners, so each track section needs its own feeder from the bus, as shown.   The bottom half shows how I run a sub-bus for each block, so that I can detect current in any track section in that block.  Each block gets its own sub-bus, while the main bus can make a much longer run.  The B-side bus does not need a sub-bus, so I didn't bother to show that.  Hope that clarifies it.

-gfh

P.S. Nice shots jb.  I'd like to get back to Richmond to visit some day.  My brother lives just a few miles from there.

[ChrisKLAS]

My next step will be to turn down the current threshold and increase the time-out setting for those blocks.

THIS! One of the best changes I ever made to my CTC system was to implement a 10 second debounce on the "island" blocks before they report clear. Even with 2 resistors per car (one on each end) and powered frogs, I couldn't get keep the wheels and/or track clean enough to avoid having the control points drop out of occupancy from time to time. Not a huge issue when I'm just out there running trains myself and know where everything is, but I've recently delved into remote dispatching (where the dispatcher is sitting at a computer 1500 miles away) and those occupancy drops were completely unacceptable in that scheme of things. You haven't seen excitement until the DS accidentally lines a switch while the train is still passing over it!

The 10-second rule has worked very well for me and, since its implementation, I haven't noticed a single block drop occupancy. Best yet, it's incredibly easy to do with JMRI.