Wayne Roderick, 3rd Division, PNR, NMRA (life)
06/19/98 rev 01/15/07
12/30/02 update: Things are changing., I haven't had to update my home-brew system because it is built to the NMRA standards for DCC and it isn't broken, but it doesn't have all the features that I have found desirable. I leave you with the following dated page so that you may take pleasure in the low price and compactness of DCC decoders today and suggest you look at DCC interface device (DCCID)
The Teton Short line is committed to the installation of DCC, but unlike most of the folks, we like to suffer through the design and construction of something new. After gleaning much data from the 'net, we decided to buy North Coast Engineering (NCE) decoder kits and build the rest from scratch- well not really scratch, we'll buy the little black bugs that contain the magic blue smoke from others.
Why do it the hard way? 'Cause it's fun and that's why at 62 years of age, I'm still playing with trains. Besides, it'll save a bundle, and work with our existing walk-around cab system, and we can do simulations on the computer. The NMRA standards define the CONTROL STATION to LOCO interface, but do not provide for the CAB to CONTROL STATION interface. Some folks are working on this and perhaps somewhere down the road something will emerge.
To make things more difficult, The TSL has a policy that requires us to be as operational as possible on Thursday evening. This long standing policy insures that the construction mess is cleaned up at least weekly whether it be plaster or electrical. It does indeed complicate the effort, but isn't that all part of modeling the real thing? The prototype folks insist that trains keep running during construction.
TSL DCC SYSTEM:
The typical COMMAND STATION will not be found on the TSL. We are using the embedded computer to interface our existing walkaround cabs to a hardware device, the DCC interface device (DCCID) that generates the DCC commands, as a TTL level signal. The TTL output then goes on to feed the power boosters and the programming booster. The DCCID has a RAM chip that holds the current commands for 21 locomotives whether in service or not.
QUESTIONS: Looking at our unorthodox approach to DCC leaves me with some thoughts that you may have some comment on.
Why should a loco ever be "disabled" if I have the capacity to scan the whole roster several times per second?
Why should I use the consist address feature when I can just command each unit every scan?
Why should I fool with setting many of the CV's in the decoder when I have a high power computer to keep track of that stuff and adjust the commands accordingly.
DECODERS: The NCE one amp decoder kit is not the smallest in the world, but its well built, very economical in quantity, and easily assembled, but you don't build it with a $5 solder iron from Radio Shack. A regulated solder station is a must. It has headlight control and output for other device(s), but we choose not to use that option. I want the lights burning on both ends whenever there is power on the track. IMHO, the greatest value of the headlight is to indicate power available and dirty wheel/track.
The NCE-DK15, one amp decoder measures: 2.25" x .650" x 0.25-0.30"
(10/15/99 update) The decoder kits are no longer available. As Jim Scorse explains: "--the street price of our factory built decoders approached the kit price, PLUS the factory built ones have a warranty--".
New decoders that are smaller, with more features per dollar, are showing up from several competing vendors. This was the promise of DCC. Such is progress- Isn't it wonderful?
LOCO LIGHTS: A comment on loco lights is in order. Once upon a time, long ago, we became offended at the bright light glowing in the cabs and through the walls of our diesels. Our solution then and now is to provide small lamps for the headlamp (and number boards sometimes). The TSL has had a high-frequency lighting system since 1967 and it still works. It puts 8 volts at 20 kHz everywhere on the layout tracks. We now use #682 t-1 lamps (1/8" diameter) acquired at a surplus store, but still available new for about $2. This lamp is rated 5 Volts at .060A with a life of 40,000-60,000 hours and it has a small base that can be carefully soldered. The old Athearn GP-9's have a tiny piece of tin can stock in a Vee form soldered on as a reflector and the set right behind the plastic lens. We limit the high frequency current and keep the DC out with a capacitor for true constant lighting. The capacitor does not heat like a resistor. We use a similiar scheme, with battery backup, for caboose marker lamps (need link here).
The high frequency lighting system will be abandoned, being replaced with the DCC power of 13-14 volts at 8kHz. Vector summing the voltage drops of two lamps and 0.1ufd capacitor comes out just right and the 190 ohms capacitive reactance limits inrush surge to preserve the lamps. We always use surge limiting on our incandescent lamps to extend life!
DCC DESIGN & CONSTRUCTION STATUS:
A progress, or lack of progress report:
02/16/98: We have eleven of the twenty-one decoder kits built, with nine of them installed and preliminary checked out. We have enough hardware and software built to do simple decoder programming and operate them on the baseline addresses using the fundamental 14 speed steps. Cab to Loco assignment and consist makeup is working well. The signal generator is built and interfaced to the computer, One of four power boosters is built and the programming booster is in service.
02/22/98: A couple of painful lessons and the extended addressing works with 28 speed steps! The problem with the extended addressing was frequency drift in the DCCID. According to the NMRA Standard S-9, it was barely in tolerance, which was OK for the shorter baseline command but wouldn't let the longer extended addressing work. The problem with the 28 speed addressing was a pitfall, surely laid by those software nerds and authors of the Recommended Practices (RP), to slay us do-it-yourself nerds. The five bits of data for the 28 speeds are rotated so the LSB appears before the MSB! I admit, its documented deep in the fine print of RP9.2.1. I've posted the QBASIC source code of my most recent test program dcctest8.zip in case you're interested.
D-DAY 02/26/98 We simultaneously moved two consists, one loco turned backwards, up and down the track today, each consist under control ot it own TSL standard cab!!!! Using baseline addressing and 28 speed steps- abandoning extended addressing while we work on the simulation software.
Now, the job remains to incorporate the test routines into the embedded program and we should be up and running- uh- only about nine months behind schedule..
03/18/98 Malfunction Junction (MFJ) has a it's own throttle/cab system, not part of the walkaround mainline control. We like the feel of the analog control knob for slow speed switching with lots of reversing. See our novel DCC adaptation: MALFUNCTION JUNCTION CAB
03/28/98 Sometimes you have to smack a mule twice to get his attention. I feel that way today. Burned up the second decoder due to problems with the transition from conventional DC power to DCC. The danger comes from using common rail wiring which you probably won't do if you start off fresh with DCC. The larger, and more complex model railroads usually used common rail to simplify the wiring. As your loco crosses a boundry gap between the old and new power, it is possible to sum the voltages and overload the decoder. The first time it happened between the DCC mainline and the conventional yard cab at MFJ where we had an old MRC power pack. They made those things with voltages far exceeding the NMRA standards, so the children can run trains at 150+ mph. We fixed that one, but didn't get the Termite Timber Line (TTL) branch fixed before a visitor innocently ran into it. This time, the overheated decoder melted a hole in a Kato RS unit!
04/14/98 OVERVOLTAGE PROBLEMS: After an extended discussion and monitoring of the DCC SIG group, I'm scrapping the COMMON rail concept for the BALANCED feed system called "home wiring" by some. Both rails are gapped and fed with a pair of wires- I use 16 guage zip cord. GROUND/COMMON is found only prior to the H-bridge in the DCC booster. Any rail in the system can only be at ZERO to +14 volts and so no combination across a gap can exceed 14 volts. The old conventional cabs serving local yards are now rebuilt to not exceed 13 volts and will be wired so the negative terminal goes to GROUND before the direction switch. The overvoltage risks to the decoders should be gone. Look at Common Wiring for DCC and other systems for a good pictorial explanation.
DCC DECODER INSTALLATION EXPERIENCE:
Listed below are excepts from the official Teton Short Line files:
Kato RS-3, RS-4/5 and RS-11 ALCO's
WEAK COUPLER PROBLEM: The plastic frame extension that the coupler must mount on is too flexible. When pulling hard it bends, permitting the coupler to lift out of its mate. With a single engine, this is not a problem, but when ran at the module club MU'ed together, it failed repeatedly. Problem repaired with tin can stock.
SHELL REMOVAL: Remove handrails from the cab to prevent breakage! The body pulls straight up with a little wiggling. No need to remove couplers as the walkways, pilots and couplers remain. ====>CAUTION====> short connections with #30 wire to the lights. Decoder stays with the motor, just unsolder the light wires to remove.
DCC: RS-3, 4/5. Removed small amount of weight, sawing off the small extensions that project to the cab center to reduce the risk of electrical shorts. In this small body, the vertical clearance was a major problem! Scrapped the grey plastic lamp and buss board over the motor, and cut the mounting lugs off. The plastic light guides had to go also. Mill the lumps off of the ceiling where the decoder needs clearance. The decoder is mounted solder side up with carpet tape separating the motor and components. This allows the slightly thicker end to extend beyond the motor and hang over the flywheel. Two brass buss wires cemented on the motor to carry the power between trucks. This was a time consuming installation!
DCC: RS-11's 1/5/98. Removed a small amount of weight as above. With more room to work, kept the grey plastic mounting board, mounted the decoder solder side down with a piece of sheet plastic for separation. The plastic light guides had to go.
LIGHTS: Two #682 in series with 0.1uf capacitor. The cap is between the lamps, cemented to the cab ceiling where it doesn't interfere with the decoder. Plastic lite pipes are cut off leaving about 1/2" stub. The stub is filed and rounded a bit, so it can be attached to the lamps with silicon model aircraft fuel tubing. Two #30 "wire-wrap" wires are soldered to the lamps and a 1/4" piece of shrink tube insulates them from the lead weights. The stubs (headlight lens) are ACC'ed in place in the smaller bodies. They snap in tight enough in the RS-11's
COUPLERS: Takes a Kadee #5 in its box with the ears trimmed- clean!
SHELL REMOVE: Remove the handrails from the cab to prevent breakage! The body comes off leaving the running boards, and lower details. Put a little pressure on the tabs above the inner axles.
DCC: Unscrew and discard the extra weight (2oz) above the motor and lighting system. Mill out the roof space a tiny bit so the width will take the decoder snugly. Can also file (very slightly) the decoder to reduce its width. Press it in with solder side against the roof. Trim cylindrical extensions from the grey plastic light housing. Leave Org & Slate wires full length. Cut back the motor connection tabs from the brass busses and roll a small loop in them. Solder the Org & Slate wires to the motor tabs. The Red and Black wires solder to the mid-point of the power busses, using a bit of cardstock to prevent melting the plastic.
LIGHTS: Salvage the bulbs from the tiny pc board, and discard the board. Restore them to the original locations, wire them in series and connect to the adjacent power busses. The plastic light pipes are not disturbed. Cut a piece of card stock about 3/4" x 3" and cover the decoder by slipping the ends between the plastic light pipes and the ceiling. An easy install.
Proto2000 GP-9 ph 3
Beautiful body work nearly completely assembled, except for optional all weather windows and footboards that must be snapped on after assembly.
Complicated weight casting fills all available space with a " too shallow & too short " recess in the top where the diode lighting circuit board is. Six diodes directionally power 1.5v lamps The motor was found to be insulated from the metal chassis with a pigtail attached to the lower brush and brought up the the circuit board. This is a heavy loco and can afford some weight loss.
COUPLERS: Comes with McHenry style knuckle couplers. Will work OK on the TSL, where no more than two units will MU'ed, but we've found McHenry's pulling apart when trains get up around 60 cars on our module clubs water level trackage.
SHELL REMOVE: Footboards interfere with couplers when removing the shell requiring that one or the other be removed. The fuel tank must be pulled off to release the shell tabs. To remove the weight casting, requires unsoldering the non-cab end truck and the motor wires. The decoder, lamps and weight stay together.
DCC: milled out space in the top. 1/4" deep and longer plus an additional.010x3/8 at the non-cab end where a .010 piece of plastic was inserted to insulate the diodes. Glued decoder, component side down with ACC on the transistors to the casting- should make a good heatsink! Milled a wider and deeper a wire channel under the cab end light. Milled a rabbet into a small removable weight at the cab end sufficient to make space for a pc board solder terminal area. 5 solder pads serve the source, motor and light. This is much tidier than splices and can be easily undone. Chassis connection was made by squeezing a stranded wire under this piece.
03/15/02:Recently there has been some discussion on the SIGlists about poor electrical pickup with this loco. Mine have been OK but a friend suffered. His units have Digitrax decoders that fit much easier than mine and he suspected them- Perhaps that decoder is more sensitive to brief power interruptions- I really don't know. The electrical problems were traced to the greased mechanical interface where the truck bolster pivots on the body bolster. It just isn't good enough- Yes, I know that Athearn has used that technique forever and maybe we could just clean up the grease and fix it but I'd gone deep into it and wasn't going to do a half-way patch. I cut of the tab that brings the other connection up from the truck and soldered wires to the truck. Dressed them so they would not be abraded and Used a short piece of heat shrink and ACC to lash the wires to the remainder of the cutoff tab to keep them in place. Here is a picture. Made solder connections on pcb pads for easy truck removal.
LIGHTS: A pair of #682 in series with 0.1uf. grounded to the casting at non-cab end. 'cap is soldered to the edge of term' board. power from non-cab end truck comes up thru the weight and snakes along the top, under the decoder. #30 wire between the lamps snakes along here too. Heat shrink tube around the lamps made them a snug fit.
This was a time consuming installation!
Return to Teton Short Line Home Page