Wayne Roderick, 3rd Division, PNR, NMRA (life)
01/17/98 rev 01/16/07
Browse the text while the pictures paint
We've seen many attempts at making a reliable turntable with indexing, few successful, some monstrosities and quite a few failures. We suffered through our own errors too, but eventually answered the challenges and built a self indexing turntable that has stood the test of time. I'm sure there are other good designs, but in my travels, I've only seen one other that I would call successful. It was built long ago by Larry Keeler, Kansas City using a primitive computer, IMSAI?. He used a non slip timing belt to link the the table and motor, a revolution counter on the motor and software to position the table from a known zero point and to account for system lash. Today, New York Railway Supply makes an indexer, $359 plus options, with a stepper motor that attaches to the shaft using the same digital principles. The remaining problem is finding or refurbishing a commercial turntable to minimize "wobble, slop or play" so you can take advantage of an accurate positioning device.
The turntable at Malfuntion Junction was built more 20 years ago, in 1978, and it has served with nearly zero maintenance, and retained its accuracy. Many visitors have crawled under the benchwork to make notes on it. The goals we set out for it back then are listed for your consideration.
RUGGED: It's built with 3/4" plywood, 1/2" shaft, ball bearings etc. The entire roundhouse complex is on a 3/4" plywood base. I feel that anything less lacks the necessary long term stability, especially after you chop out a 12" hole for the turntable.
ACCURATE HOMING: A 100 scale foot table is over 13 inchs in length. The distance around is over 40 inchs and you need to align the track within .020 inches, or .0005 (.05%). With optical homing, we attain that goal easily.
INEXPENSIVE: There is only one set of optics for the 20 tracks. The redundant device for each track is a very simple sheet metal vane. with an amazingly simple adjustment mechanism.
EASY CONSTRUCTION: You don't need machine shop precision.
ACTION: The control panel has a Single Pole, Double Throw rotary switch with center off. We rotate it in the direction, CW or CCW, that we want the table to go. and the table moves at its fastest, RUN speed in the chosen direction. When we turn the switch off, it drops to medium, APPROACH, speed and continues to the next track. As it nears the track, it drops to slow, HUNT speed and zeroes in to place.
The sketch below is intended to give you ideas. It is not a detailed construction plan, because like so many devices on the TSL, our table was built from the junk box. The only dimensions that are significant is the shallow depth, only about 5" deep and the width, about 15". The table is about 100 scale feet long in HO scale. You can make it any size you want, for any scale, G or smaller. The dimensions of the table (bridge) above have no relationship to the drive and homing hardware below, except for the radial positioning of the tracks. Our tracks are positioned 10.8 degrees apart, and that was determined exclusively from the roundhouse design.
THE OVERALL CONCEPT is that the homing devices are under the benchwork and accurately slaved to the radial position of the table above. To accomplish this, we mounted a 12" disc of 3/4" plywood on the same shaft with the table using techniques that insured essentially zero radial lash or slop. This can get tricky without a machine shop, but we discovered the tricks. Our shaft was salvaged from an antique Teletype machine and it had a flange that could be used to attach the 3/4" plywood homing disk, so this becomes the heart of it. It is important that the disc rotates reasonable true, with less than 1/4" wobble at its rim. The shaft has ball bearings that were set into to 3/4" wood stucture- The flange and bearings are shown solid in the cross-section. OK, now head for the junk box with imagination.
Jan 2001 update- The bad news is that we didn't make photos of the original TSL turntable while it was under construction and I have no desire to remove it for picture taking. The good news is that it has been so successful that we're building another one with 48 tracks for the Pocatello Model Railroad and Historical Society, circa 1920.
For the historical student, the prototype of the Oregon Short Line roundhoust at Pocatello Idaho had 60 track positions at 6 degree spacing. With the 100 foot turntable, the rails of the adjacent tracks just meet at the edge of the pit. The Pocatello roundhouse in 1929 had 53 stalls and six in/out tracks leaving one empty position. There stood the utilities, a pole that dropped wires to the gallows and the underground drain tile. It was too big to model and still be able to reach into the table so we cut it to 48 tracks at 7.5 degree spacing and limited the roundhouse depth to 100 feet, more typical of th 1914 construction. Similiar to the half circle one in 1914, Our replica only 26 stalls, so we have some garden tracks to show off pretty locos.
What? Twenty plus years later and we're building the same thing again? Yep! the Oregon Shortline (Now UP) TT at Pocatello that we are modeling was built prior to 1914 and it's still working. True, the bridge was modified to increase it's load rating by shifting weight from the center bearing out to the rim rail, back somewhere, perhaps in the forties or fifties. Ours is already built that way, so what's to improve? ;-)
This it what it looks like on the new 48 track table with the bottom bearing removed. The boards on the sides are simply legs to rest it on at the workbench
TABLE MOUNT: How do you make the table ride smoothly on the circular pit rail, so it maintains proper elevation, and do it with no radial lash. We soldered a hacksaw blade on the end of the 1/2" shaft. Soft solder will do, 'cause there's plenty of area and a torch would ruin the spring temper. We arced the blade down and bolted the table to the ends. Nuts are soldered to the blade where the mounting holes are, (it's too tough to drill) so that it's easy to remove the screws from the top and lift the table off. The table is a piece of pine 3/4" by 1-3/4" with gingerbread details, i.e. decking, bridgework, and guardrails added after the mechanism is working well.
The table has wheels that bare on the pit rail for support and electrical pickup, and the added spring pressure enhances the contact with the pit rail. With this support arrangement, the vertical alignment of the shaft is not very critical! The blade is shown in red in our sketch.
This it what it looks like on the new 48 track table with only the hacksaw blade and end rollers in place. The roller mechanism are simple brass wheels on a music wire axle soldered to supports made from printed circuit board.
TRACK POWER SWITCHING: How do you switch the power to all those tracks. Once upon a time, we had a typical 12 position rotary switch, but then we rebuilt the yard and ended up with 20 tracks. Idea! The turntable rotates like our switch, so get rid of the redundancy and let it do the switching. We made a simple switch by cutting some slices into a 4" square pc board with the Dremel tool, mounting it under the table, and fabricated a set of contacts to select the proper segement as the table turned. It's shown in green in our sketch. The rotor is a scrap of nylon that clamps to the shaft and has a couple of brush holders (like a motor) fabricated from brass tubing. Spring loaded brushes wipe on the copper. A 25 pin "D" connector makes it easily removable.
This it what it looks like on the new 48 track table. The bottom bearing is in place and the switch mounts below that. There is also a phosbronze wiper on the shaft to carry power to one rail on the table. The other rail gets it's power from the ring rail. Track power is selected at one end only of the table, so that end usually points to the roundhouse and garden tracks. IN and OUT tracks receive power separately.
RAIL ALIGNMENT: Plan on putting tracks on opposite sides of the pit that are accurately aligned, so the table can be turned end for end. You do this by constructing all the tracks on one side first. Then tack your rails to the table with spots of hot melt glue- no ties, we'll fill in between and alongside the rails with planking later. Now do the end-for-end game until you learn how to position a track on the opposite side and position the rails on the table. It's a simple matter of geometry, easier to do intuitively, than explain. You'll probably have to melt the glue with your solder iron and move the table tracks a bit. When you get the table and one pair of tracks aligned, the rest is duck soup. When you're done, feather the rail ends slightly to form a funnel or gathering effect.
RING RAILS: There are two of them. The topside one, the "pit" rail, is in the bottom of the pit, cut into two half circles. This is used for electrical pickup to the table rails. Contrary to some guidelines, an additional reversing switch is not needed- the split circle does it automatically. The gaps are cut opposite each other where there are no tracks, so that the table wheels will not cause a short.
09/30/98 Update for DCC: You ought to cut double gaps with enough space between them so you can't short out the constant DCC power as the table turns past the no-track area. Alternatively, you could wire a relay to cut the DCC power while the table is in motion.
02/03/01 The new 48 track TT doesn't have a no-track area where we can gap the ring rail so we leave it continuous and use the center shaft for the other conductor to carry track power to the bridge. You could of course drop the power in from a nearby pole to the gallows using a slip ring assembly. We do lose the automatic polarity feature so you'll have to have an auxilliary track polarity switch or device.
The underside ring rail on the homing disc is not used for electrical purpose, but is a reasonably precision 0.1" spacer for the positioning vanes above the homing disc.
POSITION VANES: The vanes are cut from sheet metal, about .010" thick, 1/4 wide and 2-1/4" long. One end is reduced to about 1/8" x 1/2". The other end has a hole for a spike or small brad. They're easy to make, if you rough cut them, squeeze them together and solder the whole mess into one block. Now drill, grind and shape the block. Finally, heat the block up with your torch 'til the solder is melted, drop it in a paper bag and shake. Wow! lots of pieces, all identical. Throw away the top and bottom pieces that have burrs on the edges.
Pin the vanes onto the homing disc at the same radial angles as the tracks and solder the narrowed end to the ring rail. Now, to fine tune a stopping position, all you have to do is heat the joint and slightly move the vane. Our mechanical system has essentially zero radial lash.
OPTICS: I used a couple of phototransistors and a "grain-rice" lamp bulb. These are assembled on the end of a piece of hard, springy aluminum that slides along the homing disc between the edge and the ring rail. This keeps the optics and vanes relatively positioned even with some wobble in the homing disc. With the metal polished and the wood track waxed, friction is nil. The vanes barely clear the lamp as they move past, alternately blocking the light from the phototransistors. The whole system needs to be reasonably shadowed from excessive outside light. By now, you can guess that we're going to move into a position to balance the light on the two phototransistors and this is the key to accuracy- It takes two!
This it what it looks like on the new 48 track table. The mount for the photo-transistors (PT) is a small piece of double clad pc board, one side soldered to the phos-bronze substrate, the other sliced up to make four solder pads. The substrate has a couple holes for the PTs to look through. They are fine positioned simply by bending the short leads. You have to look closely to see the lamp embedded in some silicon sealer. The whole assembly is soldered to a couple #4 wood screws embedded in the wood mounting block. Note the wax coating on the rubbing surface of the wood disk. That's all it takes to get smooth tracking. Like someone recently described the Russian space equipment relative to NASAs high tech stuff, "it's ROBUST" :-) Crude and simple- Strictly following the KISS principle:-)
MOTOR DRIVE: Back to the junk box. Our motor and gearbox was once a tv channel selector that happens to turn at a reasonable speed. Two wraps of electricians rubber tape around the homing disk, gives it some traction. On the motor shaft, we glued on a piece of rubber or plastic tube, painted it with contact cement and dipped it in sand or ballast. A little spring pressure to hold the two in contact, and we have a great friction drive. Seems like the rubber or glue should have deteriorated by now, but it still works.
For the new one, we couldn't find that old rubber tape so just glued a 3/4" wide strip of neoprene cut from a bicycle inner tube onto the wheel. Trying to think digital, I opted for a stepper type rather than looking for a conventional one with limited slop in the reduction gears. Rummaging around in the junk box, I found a bipolar one with the common 1.8 degree steps, 0.9 degree if half stepped. This one is about 1-3/4" square with 12.5 ohm coils and seems to have lots of muscle. Lots of other bipolars or unipolars would work and there seems to many on the surplus market. Got this one for 50 cents at Tek' surplus in Beaverton OR. The motor shaft with it's friction coating is about a 1/4" in diameter which means it moves about .002" per step and this is very nearly the accuracy where the turntable track meets the radial track(s).
ELECTRONICS: The electronic package is built on a 2-1/2" x 3" perfboard and attached under the table alongside the track switch. It needs only 4 wires to the outside world, +12V, ground, CW, and CCW directions. Examining the circuit you'll see that motor speed is controlled by pulse length modulation. At RUN speed, we have a 50% pulse width, and it gets less as we APPROACH and then HUNT. APPROACH speed happens when the manual switch is off and the vane is not blocking either phototransistor. HUNT speed happens when the vane blocks one of the phototransistors. STOP occurs when the circuit balances. Some tinkering with the optics and resistor values will be in order, depending on what devices you select and of course this should be done at the bench. Note also that the 12 volt/50ma lamp has 180 ohms of resistance in series. It glows a yellow-orange color and should last forever. I don't apologize for odd value resistors and unidentified components- they came from the junk box too. This is not a construction article, only a springboard for ideas and I'm sure you can improve on the 20 year old design using a Basic Stamp or similiar processor. I'm not going to do it 'cause its not broke and so I won't fix it.
That's it- If I put my 13 pound anvil on the table, it will overshoot and then back up once or twice to zero in. With an HO loco, you might just detect a slight overshoot and backup. Good luck in the junk box. sss
For the new electronics package, at first I was diligently trying to stay digital and apply a pic or Parallax Stamp to it when I decided that was the hard way to go and perhaps awkward for my peers to duplicate and repair in my absence. For the same reason, the new schematic separates the motor drive from the optics system so that it could be degraded to a manual control.
The stepper motor drive would work nicely with the L293D amplifier and the EDE1204 IC by E-Labs, about eight bucks plus shipping and waiting, but being the impatient type, I rummaged in the junk box and done the same thing with a couple TTL chips, a 74LS169 up/down counter and a 7438 exclusive OR chip. Either way, the technology is elementary so any electronic hobbyist should be able to keep it running. The postioning optics are mechanically similiar to the original, differing only in some fine points. For the circuitry, I used available TTL rather than the original CMOS because it was there and I already had a 5 Volt regulator for the motor driver. This time our 555 timer controls the step rate rather than the pulse width to the motor, the results being the same. To make the ciruit more friendly to duplicate by others, I put a 10 turn pot' in the photo-transistor divider to better establish a midpoint when they are equally illuminated. Set it for 2.5 volts at the testpoint. You can more easily taylor the 2.1-2.9 Volt "stop window" by varying the value of the given 5.6k in the biasing divider. The other parameter that you can tinker with is the physical positioning of the photo-transistors. It's quite easy to trim things out so you are within a single step of optimum alignment. If you get too close, the stepper motor is often indecisive between two positions .002" apart which would cause annoying noise and eventually damage the friction drive. The fix for this is the 220k feedback resistor from the direction flip-flop to the sensor input.
Warning: This is new stuff, albeit using old technology. Check back in 20 years or so, and I'll be able to attest to its long range performance :-)
02/07/01 Installation time for the new turntable on the PMRHS module. In this pix, setting on the roundhouse module, you see the added simple motor mount at the lower right, the 5 Volt regulator and heat sink above it and the electronic package haywired on a scrap of perf' board in the upper center. The black wire running off to the right goes to a "wall-wart" power supply
02/07/01 Set in place, we are fine positioning tracks. First the table rails are positioned on their PC ties so that end-for-end rotation still has them lined up. Then the opposite radial pairs are fine positioned. Our prototype has 12"x 20" curb timbers around the pit edge for the rail ends to anchor. They are concaved to the pit contour. We made these from PC board and epoxied them in place permitting very accurate rail positioning with the simple touch of the solder iron. With that task done, we'll again use our solder iron underneath on the positioning vanes to fine tune the auto-seeking mechanism
With the easy work complete, it's time to start on the bridge constuction. Another club member is working on that.
You will observe that the sub floor of the 26 stall roundhouse is galvanized sheet metal. A rugged steel frame using 1/8" square stock for the 12"x 12" framing timbers will built. For a structure this big, we're not going to mess around with a fragile framework! We're still researching the wall, window, door and roof modeling techniques. The MEI code 83 track is fastened to the roundhouse floor with a double clad PC tie at the door and another at the rear of the house.
If you would like to know more about making a replica of a BIG roundhouse, go to our club site, Pocatello Model Railroad and Historical Society.
Teton Short Line Home Page