Tapiola Parish Model Railway Club

TMWDCC digital command station, test setup

Vero boards from left to right: Processor Board TMW1 (connects to PC printer port), Power Supply TMW6 (2A), and Midibooster Board TMW5 (1.5A) (under construction are The Ackdetector Board TMW3 and The Relay Board TMW4). All stations will be fitted with Power supply and Midiboosters.
[24/05/99] Ack detector hooked, capable of reading Probst decoder CV's with no problem. Have not tested yet with Roco decoder
[25/05/99] Connected to layout, operated two locos, one with the Probst decoder and other with Roco decoder. When ended the session and switched off the computer, Roco receiver switched to DC and was running at full speed with no control. This is said to be because of the PC writing "unsolicited material" to LPT port during switch-off. Searching for a remedy. Emergency break will not halt un-assigned locos.

Olli installed DCC-connectors for feeding the DCC underneath the baseboards.

We used crimp connectors and 15-pin D-connectors: The 15-pin D-plug will short all sidings etc. feeders into one, so that the power routing turnouts will be bridged. If we want to use conventional power (no DCC), removal of the plugs will get us back to old sysytem (power routing turnouts).

Opa (one of our members) has fitted old Arnold decoders we have got (Thanks Lars!) to our Dm7 rail cars (Swedish Y6 models repainted by Tomppa, another member of the club).

[31/10/99] Exhibited the setup at Turku exhibition. Operated the TMW for two six-hour sessions with very little problems. The feature we were not aware of was that the throttles are inoperative while keying in assignment commands. We also experienced a peculiar problem due to too weak a power supply (Roco starter set transformer!): when the internal overload cut-out started to intermittedly cut out power of the transformer we succed in brown-out'ing the PIC in such a state that all locos switched to dc. This kind of situation should not happen if the TMWDCC is assembled correctly and fed directly with a power supply of correct capacity. If we would have operated the TMWDCC with the latest version of software the emergency halt would have halted the trains as it now also operates the relays of the relay card!

As the above incident happened at the same time as one of the operators had his mobile phone ringing, we for a while suspected a radio frequency error, but that was clearly ruled out as the operators did make calls to each other with the mobile phones kept delibrately close to layout and moving trains. As the local newspaper mentioned our stand to have the most advanced technology in the exhibition (old 8086 PC?), the boys got an idea while I was away having a meal (they told me this after the exhibition was colsed):

While another operator was running a loco from behind, the other operator came close to the loco and, with his Nokia Communicator open, acted as if he was operating the loco with it -- the first TMWDCC-WAP connection ;)

[07/11/99] In the morning I went to the Old Toy Swap (arranged by Marklin Club of Finland) and bought two old Märklin transformers for the TMWDCC, went to the club and retuned the Ackdetector with resistor in relay card and relying on the LED's installed to the Ack detector output. I'm now able to read all the decoders we have: Old Arnold, Roco, Lenz and Probst! I also managed to install the latest (1.31) version of TMW software. All looks great now!
[07/12/99] Helsinki MRC Xmas exhibition is behind. All in all TMWDCC operated well. We had some problems with our emergency buttons, but those will be solved. We are also still having a cold joint somewhere, as we are unable to connect throttle #4 to the system (this is clearly our fault, not the system fault!)
[09/12/99] Finally found the misconnected wires of throttle#4! Noted that Emergency halt is also inoperative during command keying. Must get the Emergency halt to directly operate the main relay!

We still wish to get the disconnected throttle loco assignments visible on screen, and possibility of assigning locos to disconnected throttles.

[25/04/00] The Model-Expo-2000 exhibition lasted four days, total of 30 hours of running. TMWDCC was operational all the time, and didn't need any booting or special attention during exhibition.

As it was impossible to change functions or see the assignments of disconnected throttles the operators got somewhat irritated. Occasionally someone went to change the assignments during running without warning others and this caused nuisance (but these are features of TMWDCC).

The younger boys enjoyed the freedom of DCC too much: we had two full speed head on collisions within five minutes! The lack of forced dicipline of the DCC in general caused the older members to long for the old system of no DCC!
[05/07/00] Got ver 1.33 beta from Lars for trial run: Ah what Joy!
The screen now displays the loco assignments of disconnected throttles. A faster method for typing in assignments has also being introduced (not exactly to my liking but close enough).

We tried the "reverse DC" operation with one loco [LE103XF: CV29.2=0 and CV50.2=1]. We installed a DPDT switch to the wires between layout and booster so that with the switch we could switch the layout into ordinary DC power pack.
The DC power pack was set into full reverse and the switch was flipped while the loco was travelling under DCC: The loco in came into halt according to the braking rate. We then switched back to DCC and the loco speeded at acceleration rate set into decoder.
I understand that this feature only works with decoders built strictly according to RP 9.2.4. I'm being adviced that those decoders that don't follow this rule [and that are therefore not fully NMRA-DCC compatible] are designed to give better dc operation at low speed and pulsed power, as the processor need not take any action in dc mode. As Tapiola presently uses solely Lenz decoders (home made decoders have not proven to be satisfactory or worth the trouble!) we could exploit this feature.

TMWDCC throttles

This is the first test throttle: 100k potentiometer and a switch connected directly to D15 connector. Four throttles should be possible to be connected to a single game controller card (speed control potentiometers in place of joystick potentiometers and direction switches in place of fire buttons). Maximum of 8 throttles should be possible to be connected using two game controller cards (the other card must be modified to change the port address). A separate bus will be fitted with RJ-45 receptacles for throttles.
[24/05/99] There is a ghost throttle #5 although only throttle #1 is connected. This should fix when the other game port card is installed
[25/05/99] Altered the above throttle from type-A to be a type-C throttle. Works ok. To select loco number needs some practice! Capable of switching lights on and off from Roco receiver (five pushes of button while loco is in motion -- fun!)
[01/06/99] Un-assigning a loco from (any) throttle and then re-assigning it to throttle #9 (keyboard) causes the loco to halt. If throttle #9 speed is adjusted the loco jumps to correct speed. Re-assignment to other throttles works normally.

[10/08/99] Olli and Teppo (our members) got the trottle bus panels made and Lauri (yet another member of our club) got four throttles assembled! After some errors with soldering the cables (my mistakes!) we were able to get the first RJ-connected throttle (cab) to be connected to throttle positions 1...4. We added an emergency halt button to RJ-socket panels just to be on the safe side, but as the boxed TMWDCC sitill has the throttle bus connectors not internally connected the emergency button is inoperative!!!.
[07/11/99] Unassigned trains will now surely halt at emergency too, as the latest version (1.31) of TMW software also operates the relay after three---five seconds after activating emergency halt, so there will be no runaway trains!
[01/12/99] Olli got the throttle bus connectors inside the TMWDCC box connected so that we have the emergency halt buttons of the throttle panels operating. The throttle position #4 is still unoperating, propably due to a cold joint somewhere.
[09/12/99] Found the misconnected wires of throttle #4! Noted that Emergency halt is also inoperative during command keying. Must get the Emergency halt to directly operate the main relay! Throttle #9 is now with 1.32 software reset to zero at Emegency -- great!
[22/12/99] Finally found a "matching pair" of gameport adapters: After the boys manage to solder the missing wires of the second quad RJ-panels of the throttle panels, we'll be able to use all eight sockets. Presently we have a slight problem with selecting of the functions with type-C throttle, but no doubt that will be settled soon.
[27/07/99] By changing the throttle bus capacitors from 200uF to 10uF we got the functions working with the present joystick cards! The screen now displays the loco assignments of disconnected throttles.
[05/07/00] Got ver 1.33 beta from Lars for trial run: Ah what Joy! The screen now displays the loco assignments of disconnected throttles.
[09/04/02] First test drive with a selfmade IR throttle! IR-throttles are under development/construction. We seem to be in a need of wireless throttles as during exhibitions all throttle cables are usually tied into a spagetti knot as operators move between front station and fiddle yard at the rear of the layout.
Development strted a few months ago, and first actual throttle is complete, the second one waits for transfer from patchboard to Vero board. I hope to have two IR-throttles in use at Model-Expo 2002 in Helsinki Icehall in a few weeks time. The project is based on RC5 type bit formation but bits are shorter and message presently only consists of single start bit, 8 bit payload and parity bit. Throttles and receiver are built around PIC16F84 at 4 MHz, transmitters are LD271 at approx. 1 ampere and receivers are TSOP1738 modules. This is a very barebones system, designed for max. four throttles. The development material (not final!) may be found at here.
[27/04/02] We have now used the first IR-throttle for two days in Model-Expo. The throttle performs as expected. I have today assembled the second throttle and we'll test the use of two throttles tomorrow at last day of the exhibition.

[28/04/02] Exhibition over, the system worked well, and will be developed further.
[20/08/02] Got the first proto of MkII IR-throttle! The club is about to attend a show were we were given unlimited space, so we try to put all our modules in a row (something like 30 metres (100ft) of modules. It would be quite unpractical to walk to the middle of the layout just to change the loco you wish to operate, so a mechanism was deviced that allows change of RJ-45 patch cables electronically through the throttle. Software's now patched but not polished.
[09/09/02] Have now assembled most of the hardware to do the socket change. The vero board is fully populated, and I need to make some piggyback chips to save space and to avoid chaining dozens of chips on the vero board. This is no more fun...
[01/10/02] Have returned from holiday, and have come up with an idea of combining the IR receiver and PC into a single PIC chip. During holiday tried to figure out how to combine the DCC signal timing (needs attention every 58us) and the IR signal timing (needs attention at unpredictable moment and then at 416us interval). The code seems to become too complicated for me, so opted to use the TMWDCC hardware, so that the DCC data is sent in 8 bit chunks at approx 900us rate -- yes!
[04/10/02] First successful test run of "PC-eliminator" — i.e. the MkIII IR-throttle!

[08/02/03] The PC-eliminator was running during the XX National Model Railway Days 2002 16th-17th Nov.2002 We wished a location by the wall, but instead we were placed in the middle of the hall with walls and roof far away. We used 8 IR-receivers and got fair operation. The Helsinki MRC's Xmas exhibition was better, as the layout perimeter was smaller. We had two Roco cranes in operation. One of the IR-throttles developed a fault during the first day of the show (It refused to change address). The fault presisted allthough the processor was changed. We have now erected the layout to the club and I hope to be able to search for the remedy the problem of this one throttle.

[28/04/03] Last weekend was the Model-Expo show in the old Helsinki Ice Hall. We operated the three day show with PC-Eliminator and TMWDCC hardware. I havent yet sent the faulty Lenz LE103FX decoders to Lenz for repair (need the extra capacitor next to zener), so we were guaranteed to have some runaway trains. the above mentioned throttle not changing address was fixed (adjusted the speed pot trimmer). It seems that the system is forgetting the addresses of locos or something as we quite often had to reset the locos by keeping them off the metals (tilt) for a few seconds. I plan to modify the PC-eliminator code so that it will display the addresses of cabs in sequence if nothing more pressing is to be displayed so as to make sure the addresses aren't lost during running!

[24/11/03] I have now upgraded the PC-eliminator software so that is not allowing loco (address) capture in case the previous owner has the speed in non-zero position (diplay shows "Err!").
Apparently this is not enough, as we are still experiencing the need to re-press the address selector due to throttle loosing control. I have now included a double packet check into software, but commented it out as it seemed too slow (ther must be two identical commands from individual throttle before the action is taken). I'll investigate further, probably it should be selective so that only direction change and address change would be double checked.
I also realized a design fault: The speed pot is read so that the greater the resistance the faster the loco goes. As the tracks of pots get worn out we get (or got!) an uncontrollable situation. Not nice!
Antti has some rotary encoders, and I'm going to look into possibility of replacing the pots with encoders. Also, I think we'll need button type throttles as well, as those are excellent in trying to get the youngsters into driving more carefully.
The display now shows in sequence the address, speed and functions settings of throttles in case no-one is sending any IR-packets. As the amount of packets during operating session is so great the display is of no use during ordinary running. I think that the display should in fact only show the sequential data and not try to display everything (the eye cannot catch individual commands from flashing display).

See Tillorps Mekaniska Werkstad for complete description of TMWDCC by Lars Lundgren.

This is the first decoder built, it uses PIC16F84/10 controller (programmed with David Tait's programmer), and L272M power op-amp. The original Probst design includes functions outputs and is designed to be assembled using surface mount devices (SMD). This is built on a 23 x 21 mm (9 x 8 holes) piece of Vero board and connected from below with wire-wrap wire.

The components on the lower left corner are: 5V regulator (top), diode bridge (4 1N4148 diodes under black and red wires), 33K resistor between track and PIC input (bottom left), two 10K resistors forming a voltage divider reference to power op-amp (bottom right), 10 uF tantalum capacitor to filter 5V to PIC (middle) and 8 MHz crystal for PIC (right).

[14/05/99] No power to decoder for 0.5 -- 1 second may cause complete loss of control to decoder until another short cut of power is applied. Short loss of power seems to cause no ill effects (still not fitted into a loco!)
[25/05/99] Fitted a brown-out circuit and installed to a Roco Ae6/6 by removing the "cab" from other end of the loco, lights still directly from rails (via diodes) so they are all on if there is power!). Works OK until loco finds a dirty spot. The loco halts, lights on, needs to be tilted to get another power cut and then proceeds normally, or will not start at all until CV's reprogrammed.
[26/05/99] Found an error in brown-out circuit (so simple yet still I managed to assemble it wrong!). Have not repaired yet!
[31/05/99] Replaced the voltage regulator to 6 volt version, re-assembled the brown-out circuit (according to Microchip PIC16F84 data sheet but using 4,7 volt zener). Works well! Sensitive to (quite) dirty track, halts more easily at dirty spots than Roco receiver but will recover and continue running. Does not produce errors! I'm suspecting my programming of one of the PICs: the Probst material does not state should the Power-On-Timer (PWRT) be used? I belive I've programmed some of my PICs with it and some without! Must check!
[01/06/99] Added two diodes and 20 uF tantalum capacitor so that the plus side of DC to the voltage regulator is isolated from the DC that feeds the op-amp (i.e. made a duplicate of the positive side of rectifier bridge). This reduced the tendency of halting, as the newly added capacitor is charged to rectified track voltage and is only discharged to PIC (and voltage divider of the op-amp input) via regulator. Less frequent brown-outs. The loco seems to travel although the lamps clearly flicker due to dirty trackwork (if compared with Roco receiver, the updated Probst receiver is not significantly worse in behavior at dirty spots, rather the opposite!).

[16/06/99] Assembled the second version of Probst decoder, this time I cut one row of pins from both ends of the PIC to save space and tried the DS1818 Reset chip (Thanks Lars!) to replace the Microchip reset. As DS1818 is designed for max. 5.5 volts I replaced the voltage regulator to 5 volts. The reset chip trips at too low a voltage (3.3 V) so an additional diode in Vin pin of the reset chip was added to make the chip reset earlier. This helped, but the operating margin is narrow. The voltage regulator could now be changed back to 6 volts to give more margin as the above mentioned diode will lower the input voltage to reset chip to approx. 5.5 volts.

The Mk II receiver is to be fitted into a small diesel loco (Bjurström), an etched kit from Kaj Wikberg.
[29/10/99] Had the Mk II fitted into the Bjurström, but must have done something wrong as the brown-out circuit didn't function.
[30/10/99] Bjurström was all dead at the exhibition (too shy...). The Ae 6/6 with Mk I didin't go around our tight curves (we are not able to run trains properly at the club, only at the exhibitions!): the wheel flanges shorted to the frame. I removed some material off the frame with Dremel at the exhibiton, and messed the mechanical parts with metal chippings and lost one of the universal joint shafts to the exhibition hall floor, hence no more running with that loco until gears are completely cleaned and the joint fixed.
[1/11/99] While typing in the above, it crossed my mind that why I didn't turn the loco around and run it the other way round, as then the flanges would not have sorted the thing (the layout was a big loop, so the tight curves were always pointing to the same direction. the wheels touching the frame would have been on the same polarity as the frame!)
[07/11/99] Took the Bjurström mechanically apart, readjusted the pick-up wires and got it running again. I still need to check the brown-out, as the loco tends to jump occasionally after dirty spot of trcak. Lubricated the gears and watched the loco to creep a few seconds from tie to tie (sleeper to sleeper). I must install flywheels (or larger tantalum capacitors).
[24/11/99] Opa cleaned the Ae 6/6 and replaced the broken boggie of the DCC-equipped loco with another from another similarily broken Ae 6/6 (both were some ten years ago driven to the concrete floor so that the boggie fixing pins were broken from one boggie!)
[07/12/99] Tried to operate the Ae 6/6 but had constantly problems with either dirt or non-functional brown-out circuit. The Burström took once or twice a 6 vagon train around the layout, but was also very sensitive to dirt. Both locos were mostly off the layout as they were too unreliable performers. Must investigate further!

[25/04/00] Bjurströn now has the Mk II decoder with Mk I decoders brown-out circuit. At Model-Expo-2000 Bjurström was operated for a little time, but as the small wheels got dirty so quickly it was taken off the layout. Otherewise no problems with brown-out (didn't go crazy!).

[30/11/01] I've decided not to investigate the present decoder design any further, as I belive I've achieved my goals of proofing to myself that it can be done ;) The design is not so good (halts for so long after dirty spot) that modification to the PIC software and other measures would be needed. I suggest those willing to investigate further to look into Mr Bolton's design at Model Electronics Railway Group's (MERG) website. There are lot's of others active on this matter as well, some of those are listed at the Tapiola's Control link page.

See Dean Probst homepage [WayBack Machine] for original design and SMD artwork and PIC software and David Tait's PIC pages [WayBack Machine] for further information about programming PIC controllers.. See Tillorps Mekaniska Werkstad for complete description of TMWDCC by Lars Lundgren.