Coronation class with vertical movement DCC controlled AJ's

The latest from the retrofitting workshop; a High Level Coronation class 0-4-0ST with vertical movement AJ's.

I fitted vertical movement rather than the rotary used in the Armstrong and the 2mm scale 04 because of various space constraints. Primarily, the relatively low footplate height, plus existing underframe components would have made a "through buffer beam" rotating coupling difficult to fit. In addition, I wanted to try out a vertical (normal) movement AJ.

The plan with the vertical movement had been to use the iron core of the coil to hold a permenant magnet in the "closed" position, and energising the coil would repell the magnet to the "open" position.



The main lessons :-

• magnetic attraction back to rest is possible, but didn't quite work in this case due to lack of clearance around an existing brake shaft on the model. The iron core in the coil is too far from the magnets on the flap to pull the coupling back to rest, hence the need to add a counterweight (swings in space near gearbox).


• the power required to pull an AJ down vertically is quite large, there is considerable drag as the tail of the coupling disengages due to the horizontal spring strength in the couplings. So, a more powerful mechanism is required than rotating. This required bigger magnets, and thus created the clearance problems in the first point above.


• operationally the movement is better than rotation. The loco can be programmed to backup and uncouple, but does not need to have "move forward" in the programmed sequence. Therefore the movement is more prototypical, giving the crew time to disconnect the coupling before pulling away.

The DCC chip is a CT Elektronik DCX75, mounted in the bottom of the boiler underneath the motor. The coil is about 35ohms, and a further 33ohm series resistor is used to limit the current to the coil. The coil is wired to "half wave" (ie. to a track pickup rather than decoder blue), the DCX75 does not have a blue wire. Movement control is via various CV's which CT provide for couplings, the JMRI definitions file for this is available from me (and in time will be in a future JMRI release).



I am likely to rebuild this coupling in a few weeks with a further modification; a hollow electromagnet will allow the fixed magnet to move within the core of the coil, this should mean more power with a smaller fixed magnet. It will also make the magnetic "return to rest" more likely to work as the fixed magnet will be nearer the iron component.

Smaller (!) DCC controlled couplers

I've had the 4mm scale loco with DCC controlled AJ couplings since the spring. But my aim has always been at "really small". 2mm scale "small". So, the last few months have seen various ideas tried, rejected, tried again.
The latest is for a 2mm scale Farish 04. Its a standard N item which has had the coupling pocket removed from the plastic keeper plate, plus 2mm finescale wheels (yes I know the spoke pattern is wrong for an 04!).



Picture shows coil on bench held on bit of bluetac. The coil wires are 45swg (0.07mm dia), or about the same as hair. The scale behind is a 1mm divisions.

The movie below shows it in operation. I will turn down the movement both forwards and backwards for real operations (various DCC CV settings) as they are currently a bit too large.

The chip driving the 04 is a CT DCX74. The DCX74 and DCX75 both support movement and coupling control from a single function press. It is intended for Krois uncouplers and I am exploiting the control here. The manuals from CT are fairly cryptic, but I think I have untangled most of the uncoupler issues. (The Zimo MX63 and MX620 are a bit larger, and also have functions for Krois couplers, though fractionally different in the way they operate. The Zimo documentation is a lot easier to follow). For those using JMRI (Decoder Pro), I have put my revised decoder file for the CT DCX chips in the JMRIUsers Yahoo group files area, when its had some more testing, I'll submit it for a future JMRI release.

I can see scope for a slightly larger version of this design (parts easier to handle!), with the frame and swing arm etched in nickel silver for 4mm use. I have a sketch design for the parts, and they should work with either a rotating AJ (like my earlier Armstrong Diesel) or the standard "pivoting" AJ with the pivot perpendicular to the coupling axis. I need to talk to someone who can put the etch bits on the side of a test-sheet, plus try out a few more electromagnet coil size options.

DCC Wagon Coupler finished

Video of the finished wagon, right way up !

Goofy blows a decoder

It had to happen, though its taken a while. But I blew up a DCC decoder yesterday.

I hard-wired the wagon with AJ couplings, and tested with a multimeter. All seemed fine. So I pushed some connectors onto the wheels and the decoder malfunctioned. Much prodding with multimeter and the DCC programmer followed, but the chip was dead.

Head scratching for several hours followed, until I realised a potential short circuit route; if the suspension of the wagon is compressed on one side only, a wheel rim can touch a W-iron. The W-iron base is used to solder the AJ wire. The AJ wire is pushed by the operating cam, and the operating cam is connected to the function output. BANG!!. The answer, by touching the wheel against the W-iron I could have put track current onto a function output wire.

Lessons learned:
a) Should have put the blue common onto the "dangerously exposed" bits of the mechanism, rather than the function output. Better still, should have wired it without a blue, but used half-wave back to the track (red/black) as in a NEM651 socketed decoder. (Check decoder supports this wiring before doing it!).
b) Should insulate EVERYTHING so it cannot short.
c) When fitting AJ couplings, should ensure that the coupling is not directly soldered to a metal W-iron, otherwise a short can go the length of a train. Same must apply to metal buffers attached to metal underframes.


Changes to the test wagon:
1) Lots of insulation added; the operating cam for the uncoupler is now rubber coated, the cam bracket has insulating tape between it and the W-iron. Insulating tape added where memory wire might short against W-iron it crosses. Plasticard added to prevent AJ wire shorting across.
2) Checking whether chip will support half-wave wiring (ie. don't use blue, instead connect to either red or black).


And finally, I will be sending the FL2 back for exchange under TCS' very generous Goof-Proof replacement scheme. (Replacement received in 48 Hours, fitted into wagon, all now working well done Bromsgrove Models for the prompt service)


(signed) Goofy.

Wagon uncoupler and TCS Decoder Documentation Shortcomings

The under wagon uncoupler now works. Essentially a piece of "reversing" memory wire pulls a cam which in turn deflects the Alex Jackson wire downwards.

The memory wire comes from MERG, and is different to the more common one-way wire available from other UK sources. "Reversing" wire does not require a big spring to return it to normal length, it returns to length when the heat & current is removed. Thus, only the lightest of spring action is needed to keep the wire reasonably straight. In my case that spring is the Alex Jackson wire.

(click picture for enlargement)
A = cam which pivots in bracket glued to wagon floor.
B = actuating arm of cam, fitted with insulating sleeve (shrink wrap). This presses down on the Alex Jackson coupling, and opens the coupling on the right of wagon.
C = Crimp tag around memory wire. Memory wire is just visible leaving crimp heading towards (D)
D = bolt and washer into soldered nut. Memory wire is looped around this and pulled almost taught before bolt is pinched tight.
E = Phos-bronze wiper pickup onto wheel behind brake gear, all four wheels have similar pickups.
F = bus point for "red" side of wagon pickups. Matching bus on opposite side for "black".

The above photo was taken after blowing up the first DCC chip due to a short circuit, and thus has more insulation than the version in the video below. I also plan to change the wiring of the function output; I will move the output (purple) wire to the series resistor (56 ohm, 0.5W, behind bottom right wheel. 0.5W is under-spec, the circuit dumps 2W through it and it gets HOT, but for intermittant use it should be OK). I will try using a half-wave supply from one of the track buses rather than the blue full-wave supply, this might mean changing the series resistor value.


The video below shows the test wagon working on the bench. The chip is still remote from the wagon (hence no track and some crocodial clips showing). Also, sorry for the slightly shakey camera, my tripod was put away when decorating and I haven't dug it out again!




BUT, there is always a bug somewhere. And this case its TCS decoders. I have a FL2, bought because of the "momentary action" described in their instruction manuals (set a CV for momentary action, and another for the duration of up to 4 seconds). That CV does not work !! Further, digging deep through the manuals on the TCS website, I eventually found http://www.tcsdcc.com/pdf/CVguide.pdf which says that CV62 is no longer supported ! So, a feature described in their main decoder manuals isn't there any more. Email exchange with TCS confirmed that the function has been withdrawn on decoders and they haven't updated the online manuals. GRRRRR!!! I needed the time limit to prevent the current limiting resistor from burning out (its running way over current specification, but a larger one has a space problem in fitting below the wagon). Work around for me is probably to map the output onto Function Button 2, which is non-latching on my Digitrax throttles.

I will inspect Zimo and CT decoders to see whether I can get one at a not too high price which has a function output with a time control (both do have such features, but a £30 loco decoder seems a bit excessive for a 4-wheeled wagon with only one uncoupler !!). The alternative is DIY decoders, either the MERG version, or those developed by Paul Harman.

NCE command station wierdness

I ran my Armstrong loco with DCC controlled uncouplers on an NCE PowerCab at Scaleforum. This is a different command station to the ones I have at home (Sprog, Bachmann EZ and Digitrax Zephyr).

Under the NCE, the behaviour was slightly odd; on most occaisions (but not quite all!) the loco would not stop at the end of the uncoupling sequence, but carried on driving along the track at constant low speed. The only way I found to stop it was the "Stop" key on the NCE, then turn off the uncoupler function. On getting home, I checked the loco on my Zephyr, and it behaved normally, so I conclude its something about the NCE and Zimo chip pairing.

I'd appreciate an explanation as to why this happened.


Also, having played with the NCE, I am correct in deciding that it's not the right controller for me, too big/heavy in the hand, too many buttons (well grouped compared to many, but its still far too many for driving a train), the number keys are badly arranged so I routinely hit "6" when aiming for "5" (have they never seen the square plan of calculators/keyboards!). I can see why its a popular command station and I will still leave it on my list of suggestions to others looking for a DCC starter system, but the standard recommendation of "try several different systems before buying" applies !

CT chip vagaries solved

In an earlier posting I made mention of some problems with CV values in CT Elektronik DCX74 chips. I've now got much further towards a solution. The main lessons being:

1. check carefully which decoder files are used in DecoderPro, some of the CT files are for very old versions of the chips which have different CV layouts. Using those can make a newer chip go haywire.
2. the mapping of half-speed mode to a function key isn't well documented for the DCX74. It is in the DCX75 (German) manual, and there are DecoderPro files available on their JMRI users Yahoo group. With luck the files will find their way to a release version of JMRI/DecoderPro in due course.
3. the movement behaviour for uncoupling is different to that by Zimo. Loco movement seems to be triggered on "function off" rather than "function on". I haven't found a setting which will trigger the operation of a coil at the correct place in the operating sequence. Consequently, I think it will end up as a two function uncoupling sequence; the first to "open" the coupler, the second to move the locomotive. ( Update, December 2008 ; I have now deciphered the CT movement behaviour, and it does all the tricks ! See the 2mm scale 04 elsewhere in the blog. I have updated the JMRI/DecoderPro definitions file to make this considerably less cryptic. ).

Sensibly priced close up photography.

Taking closeups does not require expensive kit, just a bit of thought.

These two pictures are of a 2mm scale wagon with 3 link chain couplers. To give an idea of scale, the wheels are 6mm diameter, the track gauge 9.42mm.

Camera is a Canon A710is, which is a typical Canon compact camera. The additional "magic" bits are:

• Lighting box from Jessops; essentially a foldup white fabric box with a blue curved backcloth. Bright lamps are placed outside the box, and the fabric ensures a uniform light within the box.
• Closeup lens. Canon accessory lens holder for A-series camera (about £17 from Canon). x4 screw-in closeup filter (about £4 from 7-day-shop). I also have a x2 filter lens. These allow the camera to focus closely whilst on maximum zoom and stood back over 30cm from the models. In turn, this means there is minimal barrel distortion caused by having the camera in "macro" mode a couple of cm from the subject.
• Tripod to hold camera.
• Camera setup: Manually set the whitebalance. Set film type to slowest available (typically 80 ISA). Set aperture to minimum, and let camera choose speed. Focus manually (at least check where any auto focus is locking!). Use timer-release option on camera to avoid any shake. If not sure about exposure, its worth trying with +/- a small amount of "EV" compensation.

Auto coupler now complete

I received my upgraded (version 7) MX620's today, and fitted one back into the Armstrong shunter. It works !

Press the function key allocated to "uncouple" and the loco backs up enough to take up the slack, the coupler activates, and loco drives forward to clear train before coupler releases. Total movement is perhaps 1.5 sleepers, all at a slow crawl.

See earlier posting for pictures underneath and description of mechanical components.

There is scope for fiddling with the uncoupling movement time, speed, and the general speed table and acceleration parameters. These all influence the exact speed and distance moved. But priorities are 2mm models for the annual Expo at the end of June.

The video below shows the loco come to a stop under normal throttle control, then the function key is pressed which causes the loco to back-up, uncouple, and pull clear. This is an improved video, with better lighting and some closeups

(Next back to why the 2mm scale CT DCX 74's sometimes loose half-speed mode, and why one of them is reading back wierd values in its primary CV table. The alternative CV option is proving the way out, but this bug is wierd...).

Zimo mx620 upgraded

Earlier in the blog I mentioned that the Zimo MX620 has a small bug in CV116; the shuffle back and pull away after uncoupling feature. This week Zimo emailed me to say there is a new release of software to Version 7, and this fixes the bug.

You need a MXDECUP programming box to upgrade the Zimo decoder chips; either purchase or find a dealer who has the box and can do the upgrade for you.

I'll report back when my chip has been upgraded.

Pictures of DCC AJ Coupling

In an earlier blog entry, I described the DCC Alex Jackson Coupling which I have been developing.

I've now taken some pictures of the first version, and present them here with some annotations. Clicking the pictures should cause them to appear much larger.

The first photograph is labelled to show the main elements.

• (A) is the Alex Jackson coupling, formed from 0.010 guitar string.
• (B) is a nickel silver u-shaped bracket which forms the support for the pivoting wire (A).
• (C) is a steel counterweight. 2mm diameter, with a 12BA thread at one end. The thread is cross-drilled 0.3mm below the brass nut. When installed, the nut is wound down to clamp the counterweight onto the AJ wire (A). I have a better design for the counterweight which will be much easier to install than this first version.
• (D) is the electromagnet coil, squeezed in behind a gearwheel which drives the locomotive jackshaft.

Because the locomotive is upside-down, the counterweight (C) falls to one side under gravity. When the right way up, it hangs vertically downwards.

This second photograph shows the same elements from the other side. The wires from the coil can be seen running over the jackshaft axle before heading inside the locomotive. There is a small piece of PCB mounted below the motor where the coupling wires are connected to the DCC chip wires.

The third picture has a propelling pencil holding the counter-weight approximately vertically. This shows the (approximate) rest position of the system if the loco was the correct way up.

The final picture shows the counterweight pulled over by the electromagnet. The counterweight strikes the side of the coil, not the iron core; this avoids it "sticking" when the electromagnet is turned off. An alternative to avoid "stick" is to cover one of the moving parts with something non-magnetic, such as a tiny piece of brass, thin paper, or even blob of glue.

There is massive scope to improve the engineering; the pivot would benefit from an etched design, the counterweight can be made much easier to install, and the coil pole pieces could be designed to deliver a more effective magnetic field. Even the use of Blu-Tac to fix the coil in place should be avoided in later version !

The locomotive is a High Level kit of a small Armstrong 0-4-0 diesel shunter (original at Shildon museum). Prior to fitting the couplings, it was assembled according to the instructions, with P4 wheels, simple beam compensation on one axle. The next loco to have a similar coupling fitted will be built up with the coupling in mind; its a lot easier to design space for parts before assembly !

Photographs taken with a Canon Powershot A710, with addition of a x4 and x2 close-up lenses fitted to a Canon accessory lense holder. The camera is on tripod, and taken with timer release. Its a pretty cheap setup for closeup work.

Ski Pictures from Norway 2008

Photographs from Geilo, Norway, Feb/March 2008.

Nigel's Photos

Sarah's Photos


There were also photos from Chris and Janet, though they require site registrations/invitations, so consult your personal email if you were on the trip.

4mm Scale Digital Alex Jackson (AJ) coupling

As a break from 2mm scale modelling, I make some items to Scalefour standards. A while back, I started looking at DCC control of model trains and started to think of the control options a chip inside a model might offer.

One such is digitally controlled couplers; a function button on the controller can activate the uncoupler. This is not particularly new; European makers have offered such uncouplers for years on HO trainset models. The only real challenge is to make the system work with the uncouplers in common use in UK finescale model making.

My initial idea came from seeing the latest version of the 2mm scale Electra coupler. This is a small rotary acting coupling whose origins dates back to the 1970's. Its somewhat simpler to make than a 2mm scale Alex Jackson (AJ). There is a similarity in the coupling action of the AJ and Electra, and, as I have 8 times the volume inside a 4mm scale model, I decided to experimented digitally controlled uncouplers in 4mm scale as a route to 2mm scale operation.

The main shank of the coupler is a standard AJ. It rotates in two small holes arranged behind the buffer beam. A steel bar weight is fitted to the coupler shank, and ensures the correct coupler rest position. A home made electromagnet coil can attract the steel weight away from vertical, rotating the coupling head to the "uncouple" position.

The electromagnet is made from 45SWG enamelled wire (from the Scientific Wire Company), wound onto a ~1mm iron core. The core is a piece of garden wire with the green plastic coating removed. I have found that approximately 12m of wire works well (around 50 ohm resistance), though the exact amount will depend on the available space. There are various aspects of coil and magnetic field design which need experimentation in the future, such as turning one end of the core round to be near the opposite pole, or fitting a metal disc to one end of the coil.

Winding was done on my lathe in lowest back-gear, using a close-up TV camera to watch the wire was laying down tight against preceeding turns. Temporary end-cheeks from PTFE rod controls the length of the wound core. Care in winding makes a huge difference to final coil diameter - sloppy windings have lots of air spaces which make the coil bigger. My early coils were secured with occaisional dabs of super-glue, though I'll change to shellac for future ones. The PTFE end-cheeks are carefully pulled away once the coil is complete.


DCC control uses a function output from the chip. Both Zimo and CT Elektronik chips have support for uncouplers within their firmware, with CV values to set the time for the full power applied to the electromagnet (to initiate the pull) and then a lower power to hold the coupler open. Zimo have a function which can, optionally, back the loco up a few scale inches, uncouple, then drive away about a scale foot (though not yet working on the MX620, the manual is wrong on CV116, Zimo promise a fix during spring/summer 2008). I would expect that almost any maker's chip would work using their normal function outputs (lights, fans, etc).

Due to the rotary action, if the loco is not driven away and the coupler closed (magnet off), the train remains coupled. One function key can operate both uncouplers simultaneously. This simplifies the selection of function key on the DCC control panel. Selection of which function key is down to chip programming and individual choice; one or two handset makers suggest F3, but other makers use F3 for "half speed" mode. At the moment I am using F1.


With two coils (two couplings) in series on a single function wire, I measure about 90mA at full power from a Zimo chip, well within the 300mA limit for function output of the smallest Zimo (MX620). Its also just within the limit for the tiny CT DCX74/75 (100mA), though I've not yet tried these chips with the uncoupler; I would be tempted to slightly increase the resistance (small resistor in series) in the circuit with a CT to drop the current a little.

The more current through a given coil, the stronger the magnetic field. On a bench test, I found my coils were getting rather warm with 20 seconds at 150mA, but could not detect any heating at 100mA.



There is a video of its operation on the Google Video site and now duplicated on YouTube


The loco fitted with the digitally controlled AJ will work alongside conventional under-baseboard magnets. The loco acts as any other loco with a fixed coupling, and the wagon uncouples in the conventional manner.


It would be possible to put the uncoupler under rolling stock, though the cost of fitting a function decoder (a DCC chip which can respond to function commands but lacks motor control) under every wagon could get quite high, and having to address a wagon before uncoupling would be tedious on most DCC systems. I can conceive of ways of making it very easy to operate with minic diagrams, though the computer complexity and price of assembling such a system would be very high. There is a half-way house where some key items of stock could have an uncoupler fitted; brake vans, vehicles in trains which are regularly split, etc.


Having retro-fitted the coupler to an existing locomotive, I think it would be a lot easier to design the coupler components when building rather than afterwards. My next loco will be done that way round; design the coupling components before building. It should result in better shape of magnetic field and a smaller air gap for the moving part to move across, that means less current (or smaller coil) to operate. I'm still thinking about the 2mm scale version !


Comments welcome, pictures and drawings will follow when I have some time to add them.