How to avoid building a basement layout altogether

I recommend watching the YouTube video George Sinos' presentation to the OPSIG which he gave yesterday. The link to it is here. The short story is, he built a switching layout in N scale on a shelf in the family room that ended up about 11 feet long, and it is so fun to operate, taking 2 people about an hour, that he's seriously considering canceling his plans to build a large basement layout downstairs. Here is what his track plan looks like:

He has a collection of about 400 freight cars that he rotates on and off the layout between groups of four op sessions, and uses a 4:1 fast clock, so that in four op sessions he covers about 16 hours of switching jobs to the various factories on the layout. He uses two ProtoThrottles for realistic engine control and JMRI PanelPro for a touchscreen to control the turnouts. I won't go on - you can watch the video for yourself.

His presentation reminds me of something I've often said to visitors to my layout - after 20 years of effort building the mainline from Seattle to Bellingham, I finally got around to building the Burlington, WA yard, and found it so fun to operate that I wondered why I didn't just start with that and not build the previous 20 year's worth of layout! I say it as a joke, but every joke has some truth to it.

Victoria BC in N with colored and numbered dots

 

Never one to miss an opportunity to add complexity where none is needed, I tried dividing the three N scale industry tracks described in my previous post into two or three industry spots, and then put small numbers inside the colored dots. Then I set up the spots as separate "tracks" in JMRI ops, and added them in a "pool" for each track. Now, instead of getting a switchlist from JMRI ops with four different destinations for cars, there are now eight. Without making any changes to the track!

The diagram above shows the new named spots on each track, with the spot number in parentheses, and the dot colors for each track. I'm trying to represent downtown Victoria BC, based on an article in Layout Design Journal #62 (3rd Quarter 2018) by Cal Sexmith, so I picked industry names from the article and placed them on the three tracks in a quasi-logical order based on the order they might have been switched in real life. Then I placed colored 3x5 cards next to the track where the buildings might be, just to define the locations of the spots for now.

This photo shows the cars resting in their correct spots after the first (20 minute) operating session using the new scheme. It's amazing to me how fun it is to switch a small simple layout like this. I don't like the clunky layout of the JMRI ops switchlist, and haven't figured out how to improve it, so I spend a few minutes reading the switchlist and putting corresponding colored (and now numbered) dots on the cars. This is similar to a yard crew putting chalk marks on the side of cars. Then, I grab the throttle and enjoy about 20 minutes of switching fun.

The total investment in this particular N scale train layout so far: 30 min to paint the board, 30 minutes to prepare the switches and flextrack, 30 minutes to glue the track down to the board and connect wires to the NCE PowerCab, and 90 minutes to enter the track and cars into JMRI ops. - 3 hours total. $150 for the NCE DCC PowerCab, $50 for track, $300 for cars and locomotive - $500 total.

Sure, there's no backdrop, scenery or buildings, the rolling stock and track aren't weathered, etc. But the operation itself, switching cars around, is already very satisfying. That's all I'm sayin'.

using colored dot (car tabs) to improve JMRI ops

If you've been struggling with using JMRI ops switchlists, try using color-coded stick-on "dots" stuck to the car tops to help you keep track of which car is supposed to end up on which track. It's not exactly prototypical to switch colored dots around, but it does make the whole thing easier and more fun. I've been experimenting with a Lance Mindheim style of "two turnout" switching layout, and the colored dots have really improved my enjoyment (and efficiency) of the switching moves.

In the past year, I've been learning how to use JMRI ops to generate switch lists for my N scale railroad, as a way to "dip my toe in the water" before trying to implement it on my basement HO layout, which currently uses car cards and waybills for car routing. The JMRI computer program has a lot of complexity to learn, but it doesn't take that much effort to enter in, say 100 cars and then have it generate switchlists. But there are two things about it that have been bothering me. One is that the moves are generated by random number generators, so unless you tweak the program a lot, you end up with illogical moves. The other thing is that it doesn't help the yardmaster block cars in the yard for the next train. Here's what a typical switchlist looks like:

I found out from Cal Sexsmith's article in the LDSIG Journal last year that Victoria had some interesting switching areas, and several car barge services to Seattle. So I decided to try doing some switching in Victoria. I put two turnouts on a plank (previously used only as a test track) with 5 pieces of N scale flex track and hooked it up to my NCE PowerCab and voila! Victoria!

Obviously the scenery leaves much to be desired, but it wasn't too hard to add these four tracks into JMRI ops, so I started generating switchlists and trying to use them. The switching was very tedious. Some of the cars needed to be moved, and some didn't. Some were moving offline, and some were moving between spurs. It was hard to remember what I was doing from one move to the next, and the N scale reporting marks are very small and it was hard to read them over and over again while constantly consulting the switchlist. The switchlist itself was formatted in a way that made it hard to keep track of the moves on. (I need to look into switchlist formatting options). Then I remembered I had seen a local N scale modular club using small colored dots as car tabs, so I pulled some out and tried using them to keep track of which cars were supposed to go where, according to the switchlist. It worked well! I could go through the switchlist only once, reading the reporting marks on each car and matching it to the switchlist, and placing the corresponding car tab on top of the car. That done, the actual switching was relatively easy, and fun! I've heard that some people "don't like switching colored dots around", but I'm wondering how different it is from switching cars based on chalk marks on the side of the car, like many prototype railroads would have used for switching. (Before computers, at least).

This operation, with only two turnouts and ten cars, is much more fun to operate than I was expecting! It takes me about 20 minutes to complete the work, long enough to make for an enjoyable break from routine daily activities without requiring a major time commitment. And if I want to make it more "meaningful," I could always generate a ferry run to Seattle and hand carry some of the cars back and forth to the main N scale layout. The only problem I see is that it is completely distracting me from all the other aspects of model railroading that I had been thinking were more important. Oh, well!

Progress towards remote ops

Above is a screen shot of a Zoom meeting set up with six live video views of the layout, suggesting that we are moving closer to being able to try doing a remote op session. I feel like I know just enough about all this to be dangerous, but not enough to make good decisions. Also, I keep thinking that if I can pull off a remote ops session in HO, so can you! :) 

Anyway, what I’ve done so far is to set up 5 older iPhones on tripods around the layout, and log them all into a Zoom meeting as if they were individual participants. I also bought a cheap ($28) wide-angle (90 degrees) webcam and mounted it on the ceiling and plugged it in to my laptop serving as the Zoom “host”. This means that my internet router is pumping all that video up to the cloud. I tried connecting them all up to Zoom last night and it seemed to work with all six channels active, but I need to try it again with JMRI running its web server and real people looking at it from outside. I also suppressed audio from these cameras, to reduce bandwidth load, but I don’t really like that, because it would be fun if people could hear the engine sounds as their train moves past an area.

One of the old iPhones is pointed at the TV monitor that displays a live video feed from my one camera train. Crude, but it works. The camera train video output is three RCA plugs - one for composite video and two for right and left audio. I would like to plug this into my computer but don’t know how of if it is possible with some sort of converter. I also have an old video camera that probably still works, and has a similar output to the camera train. It’s too big to put on a flatcar, but I could use it to show a portion of the railroad.

But now that brings up the issue of how to bring multiple video feeds into a single computer, and then have that composite video be displayed on Zoom. I’ve gotten the impression from OPSIG meetings that a program called OBS is the only way to do that, and I’ve also heard that only the PC version, not the Mac version, will work with Zoom. I’ve been doing everything lately from a Mac, but have some old PC’s sitting around, so I might update them and see if I could combine a couple of video feeds on OBS and have them join the Zoom meeting as well. I have a security camera from DLink looking at my Tacoma staging yard, and would like to make that visible to my dispatcher online as well. All this seems like a lot of work. It might be easier to just find another used iPhone and point it at the monitor that we already use to watch the staging yard:

But the best thing of all for remote ops would be to have more than one live camera train. Some of the OPSIG guys have been mounting Arduinos (or was it a Raspberry Pi) and corresponding tiny cameras in well cars, having the Arduino broadcast the video stream over wifi to a computer with OBS. A technically simpler approach might be to set up a couple of mirrors and set an old iPhone sideways in a well car with the mirrors pointing forward. I know I've seen pictures of somebody doing a 3D printed holder for the phone and mirrors, but don't remember where. Getting the optics right might be tricky.

But switching is what I really enjoy. Switching is going to be very limited in a remote ops session. I’m thinking of just having one or two block exchanges per train, and not even bothering with individual cars. That way, I can do whatever uncoupling is needed on the physical layout without getting overloaded with requests. The only good part of this is that I have a lot of unit trains and special trains that we never have time to run during regular op sessions, because we’re always so busy switching. The remote ops can exercise those trains instead.

Can a first try at a remote op session be far off? Stay tuned!

a master plan for holding virtual op sessions

How do you hold a virtual op session? Where people stay home and enjoy operating on your layout by running trains or dispatching from their home computers and/or cellphones? I've found the question complicated and confusing, and have watched a number of OPSIG layout tours that gave examples of people doing it. So I put together the above "mind map" to try to wrap my hands around the whole thing.

Like many things, it's good to take it in baby steps, so I now see that step one would be to simply provide some cameras for live coverage of the layout in action, and IP addresses to operators so they can log in to a WiThrottle and run a train, when and where instructed by me. The second step would be to provide a remote dispatcher with access to my JMRI computer and have the dispatcher issue the movement instructions and have control over key turnouts. Beyond that, we get into challenging areas like how to provide Zoom access to cameras riding on moving trains, etc.

Here's the complicated diagram with only the first level items shown. The green boxes are already in place. As of this writing I have set up 4 old cellphone cameras as Zoom attendees, and a fifth for broadcasting the camera train. Also, I learned how to help people use my IP address for getting into JMRI with their cellphone throttle apps, by doing "port forwarding" on my home router. All that's left now is to write up some instructions and a new schedule of trains that we might try to run in such a session. Clearly they would mostly be through freights and passenger trains, since I would be the only switchman in the room helping with pickups and setouts. 

Stay tuned! It doesn't look impossible...

Yard switching times vs. mainline running times

Here's the shocking fact: It takes at least 30 minutes of real time to break down a 15 car train in my HO layout's main yard, Interbay (Balmer). That's assuming there's only one person doing the switching. In theory, it would be faster if there were two people, one working each end of the yard. In practice, they would have to spend time coordinating their moves and keeping track of what goes where, so it's hard to say how much real time would be "saved." I put the word "saved" in quotes, to question whether saving time is the purpose of the "work," when model railroading is supposed to be about play, right?

I discovered this (obvious fact) accidentally by filming and editing a video on "how to break down an incoming train". I was thinking I could do it in 15 minutes, working by myself, but it took 30. If there had been another incoming train during that time, it would have taken much longer.

At one of my favorite op sessions in memory, everything was going great, and the yardmaster at Everett (the next town to the north of Interbay) asked me "is it true there is no train length limit on trains between Everett and Seattle?" To which I said "yes." So he sent a transfer run to Interbay with a cut of 40 (unblocked) cars. Since the longest siding in Interbay holds about 25 cars, this incoming monster filled the only remaining track in Interbay and blocked the north yard throat and both main tracks, preventing any further train moves out of Interbay to or from the north (which is most of the layout).

I don't remember what happened next. I think the Interbay yardmaster broke the train in two, but the only available track to put the second cut on was the mainline. Now both the yard and the mainline were plugged. There was still a lot of switching going on around the layout, so I think what happened is we ran down the clock and let the session come to an end, without unplugging the railroad. It took me hours of (very enjoyable) switching to get everything straightened out, during the following days. But I learned my lesson - maximum train length 25 cars. No exceptions, except for unit trains that don't require yard switching.

But today I'm thinking about another lesson. If it takes over 30 minutes to break down even a moderate length train, how many merchandise trains is it feasible to run in an entire op session? It only takes 2 to 5 minutes to run the 1 or 2 scale miles between my staging yards and the first yard they come to, or between yards. If we are going to have fun with a dispatcher and trains passing each other on the mainline, it's going to have to happen with unit trains and passenger trains.

My operating sessions are based on a condensed version of the actual prototype trains that ran during a typical day in 1973 in the greater Seattle region (BN, MILW & UP). I started out trying to do that in a single op session of about 7 hours. (3 hours before lunch, 4 hours after lunch). We never came close to finishing it, so I moved to having 14 hours of op session, over two days, with a 2:1 fast clock. And we never finish that either. What I'm thinking now is, forget the fast clock, just don't run more than one merchandise train in from staging in each direction (N, S & W) until the yardmasters are caught up. Regardless of how long that takes. There is plenty of work going on with all the locals and unit trains to keep everyone busy, without clogging up the yards too much by excessive adherence to prototype merchandise train schedules. 10 or 12 operators is not enough to simulate hundreds of prototype railroad employees.