Rangachari Anand
April 2007
Introduction
I have been running Märklin model trains for over fifteen years and have built a number of layouts over the years. Lately however, I have found myself increasingly dissatisfied with Märklin.
I think my dissatisfaction starts with the track itself. The layouts were built with M and C track both of which incorporate a roadbed (metal and plastic respectively). I increasingly find this kind of track to be unacceptably toy-like. To be fair, C-track, which is beautifully engineered, can be made to look decent but this would involve a lot of work. In fact, it appeared to be not so different from whats necessary for properly finishing two rail track.
Of course, Märklin does offer the K-track line which does not have a built-in roadbed and includes flex track but having viewed it in a number of layouts, I see two problems with it. Firstly, it is very expensive - more than $12 for a 90cm segment of flex track! Secondly, its not very attractive visually - K-track turnouts look clunky and fragile to my eye.
Märklin uses a three rail system. The middle rail (called "pukos" in Märklin-speak) finally crossed some invisible threshold of unacceptability in my mind. I just don't like the look of the third rail, especially, the "ski-ramps" at turnouts (necessary to lift the center-rail pick up shoe over the side rails). The main advantage of three rail is that it makes wiring easier, especially return loops. However, this advantage has been mostly nullified since cheap and reliable auto-reversers are available for two rail layouts. This eliminates the need for manual switches or relays to change track polarity in return loops.
Click on image to expand
Now, on the topic of rolling stock, a big part of my discontent with Märklin's arises from their new proprietory MFX digital system which I think is a giant step in the wrong direction. Most new Märklin locomotives come equipped with MFX decoders. Now MFX is backward compatible to the extent that MFX decoders can be controlled by the older Motorola protocol but in the case of sound equipped locomotives, only functions one through four can be accessed. Now, I use an Uhlenbrock Intellibox (capable of handling both Motorola and DCC protocols) for controlling my layout so I could potentially replace the MFX decoders with equivalent DCC decoders but it seems wasteful. Finally, its interesting to note that Märklin's own two-rail brand, Trix, offers DCC decoder equipped locomotives. It would have been far better if they had simply adopted DCC for their three-rail line as well.
One of my main interests in model railroading is full computer managed operation. In order to allow the computer to control the layout, it needs to know whether a track segment is occupied. I have not been entirely happy with the way train occupancy detection works on my current Märklin C-track layout. The technique I used, which is standard practice, is to isolate a section of an outer rail so that it "floats" electrically. The isolated rail section is connected to an Uhlenbrock 63350 feedback decoder. When a train passes over it, the wheels bridge the rails and ground the segment of floating rail. This is sensed by the feedback decoder. Note that in the Märklin system, wheels are not electrically insulated and conduct electricity - this is true for all axles, not just the locomotives. In practice however, I have found that unless the sensor track segment track and wheels are kept scrupulously clean, there are glitches in detection. Often, when a train hits a sensor segment, it is possible for several axles to pass over it before it is triggered. In addition to these issues, the presence of a roadbed also prevents the use of an IR based occupancy detection mechanisms.
Project goals
After much thought, I finally decided to build a two-rail layout to determine for myself whether the grass is indeed greener on the "other side" of the tracks. The main goals for this project are:
- Improve my layout design skills, learn to install flex track and proper ballasting techniques. Until now, I have worked exclusively with sectional track.
- Determine how cheaply I could build a good looking and well running layout. In particular, determine whether it would be less expensive than Märklin.
- Investigate train detection methods appropriate to two-rail systems. The plan is to try both current-draw detectors as well as infra red based such as IRDOT.
Track plan
For reasons of space, I decided to build this layout within a relatively small space (8ft by 1ft). Most of the available space in my basement is occupied by my 10ft by 4ft Märklin layout. Obviously, a continuous run is not possible within so small a space in H0 scale so this would have to be a shunting layout. I wanted at least one run around loop and a siding.
I used 3rd Planit, a powerful model train layout design program, to design the layout. I had actually bought a licence for 3rd Planit way back in 2001 but until now, I had only used it with Märklin sectional track layouts. This was the first time I would use it full-strength. The layout that I came up with is shown below.
The layout has just three switch points and is itended to be a sort of shunting puzzle. The layout is supposed to represent the end of a branch line that serves an industry. The industry will be located (eventually) on the upper left side of the layout. The branch line ends on the left side of the layout. The branch continues (conceptually) to a mainline on the right side.
In typical action, a train consisting of an engine and a couple of loaded freight cars (with raw materials for the industry) will have arrived from the main line and is waiting in the loop. A couple of freight cars with finished material is in the dead-end siding. The locomotive has to drop off the new freight cars in the siding and pick up the old cars from the siding and return to the main line.
The full sequence of steps in 3rd Planit that I used in designing this layout can be seen in this album:
Framework
The framework was built from 3/8" plywood. I just needed only a couple of 2ft by 4ft sheets which cost a few dollars at the local Home Depot. I built the frame work in the form of two 1ft by 4ft boxes that are bolted together. The idea here is to make it a bit more portable by allowing the two sections to be transported separately. After laying the track, I will cut the track and roadbed at the joint between the boxes. Some alignment mechanism will clearly be neccessary to ensure that the tracks line up properly. I have seen wooden dowels used for this purpose in the Railway Modeller magazine (a Peco publication)
The framework is topped by some 1/4" plywood that I happened to have lying around. Normally, I would not have used such thin plywood but since it is very well supported, there is not much danger of it sagging. I would have actually liked to elevate the roadbed a little to be able to put in a culvert and a small stream but I decided to defer that for the next layout.
Roadbed and Track
After a little investigation, I decided to use Atlas code 100 track since it was both cheap, robust and widely available. Now, Atlas does offer code 83 tracks which are a bit more realistic but after reading in the Model Railroader bulletin boards that code 83 is a bit harder to work with (it kinks more easily apparently), I decided to stick with code 100. For my next two-rail layout, I will very likely use code 83. The two turnouts for the loop line are number 6 while the turnout for the stub end siding is a Snap Switch which I bought only because it was on sale.
Having constructed the framework, there now arose the problem of transferring the track diagram to the plywood. I started by printing the layout 1:1 and stuck the sheets together with tape. Now, the recommended approach to transferring the diagram to the wood is via a gadget called a pounce wheel. However, feeling cheap, I decided to try another approach - namely simply gluing the track plan to the wood and laying the cork directly on that! Overall this method has worked very well. I simply used white glue to stick the diagram but would definitely use spray-on contact adhesive next time. Note also that I extended some of the lines to the end of the printout to help me align the turnouts. I also drilled the holes at this time to allow lower-side access to the turnouts for the switch machine.
I then started installing cork roadbed directly on the paper using vinyl caulk adhesive. This was recommended in an article on the Model Railroader website. I too can recommend this adhesive since its very forgiving to work with and spreads well. I also used the same adhesive to stick the track to the cork. On curves, I used a few nails in the center holes in the track to get it to hold the curve while the glue was drying. I then removed the nails after the adhesive dried. In fact on shallow curves, I have found that there is no real need for the nails. The adhesive alone can hold the track in proper alignment. As the pictures show, I used a Xuron cutter to cut the rails.
Switch machines
Since this is an educational layout, I am investigating a number of switch machines - by this I refer to electrically operated gadgets for switching the turnouts. In the case of Märklin switch machines are all solenoid based. Being familiar with these, I decided to start with Atlas twin-coil switch machines since they are the cheapest.
Now, Atlas makes two types of machines, the regular and the "deluxe", the difference being that the deluxe version incorporates a double throw switch to change the polarity of the frog. As it turned out, the four axle diesels that I am using have no problem negotiating a dead frog so it appears that the deluxe machine is not really neccessary. The regular switch machine is considerably cheaper at half the price.
Installation was definitely a little fiddly but not much worse than I expected. While putting down the cork roadbed, I had taken care to drill holes for undertable access to the points. I was definitely relieved when everything lined up correctly. Overall though, I am not entirely happy with the Atlas switch machine, even though its really cheap. For the purpose of comparison, I have ordered a set of Tortoise switch machines (which are motor driven). I will report my experince with them after they arrive.
Rolling stock
I bought a Proto 2000 U30B diesel equipped with sound for just $125 from Walthers. It runs beautifully and is whisper quiet with the sound system turned off. The sound system itself is very nice - a QSI Quantum decoder is included.
When I went to a hobby store to get a few freight cars, I found a really cheap GE 70 tonner locomotive on sale for a mere $35. To test the limits of cheapness, I bought it and to my surprise, it runs very well.
April 4 2007 - Summary
I have really enjoyed the project so far and am now thoroughly hooked on two-rail. Every aspect from design to construction has been enjoyable. I still need to:
- Paint and ballast the tracks. I have ordered this gadget which promises to be very interesting.
- Try Tortoise switch machines.
- Try IRDOT occupancy detectors
April 28 2007 - Tortoise switch machines
Since my last update I got rid of the Atlas Switch machines and installed Tortoise switch machines. The tortoise motors are driven by Wabbit stationary decoders from DCC specialities. Unlike solenoid operated switches commonly used in Märklin turnouts, Tortoise machines require a constant voltage to be applied (i.e. the switch motor is stalled). However, they draw very little current (less than 25ma) when stalled.
Now, I do have a surplus Märklin K-84 stationary decoder that I could have used to control the Tortoise machine in conjunction with a transformer and diodes but I decided to try the Wabbit decoder for two reasons:
- I did not want the hassle of an extra power supply. The Wabbit decoder provides the Tortoise with digital power, but I was not very concerned since the current drawn is so low.
- Wabbit decoders have provision for manual operation. This is very easy to use - all you need to do is to connect a set of push button switches to some terminals on the decoder.
- The Wabbit decoder has a really nifty feature: It can sense the presence of a locomotive on a track segment and throw the switch based on the built-in sensor. This can be used to automate reverse loop operation (see this document for details).
Here is a comparison of the costs for a remotely operated turnout from Märklin ( taken from the EISENBahn web site) and in my current layout:
| Item | Märklin | Two rail/DCC |
|---|---|---|
| Turnout | $17.90 (24611) | $8.00 (Atlas code 100 #6) |
| Switch machine | $18.03 (74490) | $15.95 (Tortoise) |
| Decoder | $9.65 (1/4 Viessmann 5211 decoder) | $12.00 (1/2 Wabbit decoder) |
| Total | $45.58 | $35.95 |
As you can see, the selected two-rail components are a bit cheaper than Märklin but not by a huge amount. I could have saved a lot more money by using the Atlas twin-coil switch machine and reusing my Viessmann 5211 stationary decoders. But I really like the way Tortoise machines operate. The biggest cost saving in switching to two rail would be from the relatively low cost of flex track compared to Märklin sectional track.
May 5 2007 - Weathering track
I have always admired the appearance of properly weathered track that are shown in model train magazines so I was determined to weather the tracks on this layout. Now, I could have weathered Märklin C-track but somehow I never did - perhaps the high cost of the track dissuaded me. Its
I used the Rusty Rails Painter from Joe's Model Trains to paint my tracks. I also used the specially blended rust color latex paint from Joes's model trains. After a few false starts, I was able to get the rythm of using the gadget and quickly painted all the tracks including the turnouts. I am very pleased with the way the painted track turned out. The color is just right. Just for comparison, I tried using a fine brush and found the Rusty Rails painter to be much easier.
May 9 2007 - Ballasting
I have ballasted the track using Woodland scenics medium ballast and stuck with Woodland Scenic cement. Overall, I can only say that it looks ok - not as good as I would like it to be. It turned out be harder than I expected to spread the ballast evenly. It was a bit lumpier than I expected. Also, I found that ballast particles stick to the sleepers and the web of the track. After the glue dried, I carefully scraped away the extra ballast particles but it was a fair bit of work.
