Dunedin

It's interesting how explaining the difficulties I'm having to someone else helps clarify what the issues are to me too and assists with coming up with a solution. I've realised that what I need for my incorrect signal aspects is a timing circuit. This would be arranged such that when the route selection button is pressed, it grounds the Minipanel input to set up the route and starts the timer, but doesn't ground the input for the signal. After a pre-determined time to allow the route to have set, the timer expires and grounds the input to set the signal. As this will occur after the route has set, it will select the correct signal aspect. Simples!

Thanks Andy and please do come and see us.

One example of an anomaly is where a train is sitting in a section beyond the signal, so in theory it shouldn't be possible for that signal to display any aspect other than red. I'm using block occupancy detectors to detect the presence of the train, which they do. These can be linked to an input on the NCE Minipanel to revert the signal to danger when it sees the input from the block detector. If a train enters that section, the block detector switches on and the Minipanel detects this and runs the macro to revert the signal. If the train is already just sitting there, although the block detector detects it, the Minipanel doesn't react and it is possible to select that route again and clear the signal. This is because the inputs are being scanned for changes of state - the processor is looking to detect a falling edge as the input is grounded. If it's already grounded, there's no change of state, so the input is ignored. My solution at the moment is to have the block detector drive a relay and the state of its contacts determine which input on the Minipanel is grounded when you attempt to select the route. If the relay is energised (normally open contacts closed), you get a proceed aspect. If it is de-energised, the normally closed contacts are closed and you can only get a red. Those relay contacts are connected to different inputs and those inputs determine which macro gets run when you press the route button. This works ok until I start interlocking the signals with the points. The signal aspect depends on the way the points are set and until the points have changed, and the appropriate contacts which detect their position have changed state, it's possible for the system to decide to display the incorrect aspect on the signal. Once the points have changed, if you press the route setting button a second time, because the point detection contacts are now in the correct position, the system will now display the correct aspect. Once an operator knows the trick is to press the route buttons twice, I suppose it's ok, but it isn't intuitive and it's a fudge and I don't like fudges. For me, this is a fairly significant area for improvement. For exhibitions, operation needs to be as mistake-proof as possible. Whilst at the moment it's good, it isn't where I want it to be, so watch this space. There's more to run on this topic.

Wow, thanks Andy! I had thought this might be content for an article in some publication at some point. To be honest, I hadn't thought about MERG, mainly because this is mostly using proprietary parts and where I'm having to use logic for decision-making (i.e. which aspects get displayed for which route conditions), I'm doing it using relays rather than electronics because I understand relays. I know it could be done much more neatly with some integrated circuits and AND or OR gates. Unfortunately, the NCE Minipanel can't do "if-then-else logic". If it sees an input, it just runs the appropriate macro that corresponds to that input. You have to make the decisions about which inputs to give it elsewhere. I have some friends and former colleagues who are electronics engineers and who I know would help me with this, so there's much more to do to iron out some of the anomalies that I have at the moment.

How does all this work in practice? Here is a video showing A57, with the route set for the transition from Down Main to Up Main, with approach control. This is all set up to run automatically, when that route is selected from one button on the panel:

The Yamorc has 16 switchable outputs. It needed 6 for signal A57, so that left plenty over to do something else with. I used the next three to control AK60, the position light ground signal for the move from the Up Main into the depot. Signals A57 and AK60: That left the banner repeater, A65BR. Originally, I thought I would just site this on the curve, somewhere before the bridge like this: The more I thought about it, the less happy I was with this arrangement. It wasn't a good angle for seeing when viewing the layout from the public viewing side and it was a little too close to A57, so moving it nearer to the bridge would be better. The problem was that when I built the bridge - in a hurry back in 2017, trying to get the layout finished in time for the Derby Roundhouse exhibition that May, I'd put a retaining wall there, with no room for any signal. That would mean remodelling the cutting, which also meant remodelling the bridge. More of that in a subsequent post, because that has become another project in itself! The banner repeater looked much better with the reprofiled cutting and sitting in front of what would become the bridge abutment wall: Once again I found that the Traintech decoders were unsuitable for the banner repeater: it was too dim, despite it also having no resistors when I specified it. Unfortunately, these can only be supplied wired common negative at the present time, so I couldn't drive it directly from the Yamorc switch. What the switch can do though is provide an on/off output. This is intended for controlling lights on a layout, but there is sufficient current to drive a relay, which in turn is used to control the banner repeater which is then fed from the 12V dc supply on that baseboard, via appropriate dropper resistors.

How hard? Well the first thing to note is that these signals have multipin connectors that need to pass through a rectangular hole in the baseboard to mate with the sockets on the baseplates which mount under the baseboard, whereas the Berko signals just had a flat plastic base that you pined to the baseboard surface. All that was necessary was a hole in the board to allow the wires to pass through. Cutting the rectangular hole is a bit of a nuisance, but it's ok except where there's something below the board surface. In the case of A62, there was: When we added the mainline, we widened the original boards by adding a 9 inch extension to the rear of each board. The mainline runs along the new board, but A62 was right above the join between the new and the old boards. With only a few wires to pass through, it was possible to fudge it by drilling a hole at an angle. That wouldn't work for a rectangular slot. I ended up buying multitool to cut out the original gusset below the board to make way for the mounting circuit board. Eventually it all went in. A62 installed, just reballasting to do to complete: For A57, the installation was more straightforward, but getting it to work was another matter entirely... A57 installed, ballasting still to be completed: With the original Berko signals, I had used Traintech signal decoders because they were relatively cheap and simple to install and set up. They are designed to work anything up to a 4-aspect signal and have two DCC bus addresses, each one on or off, giving four possible outputs. They can therefore be used for a 4-aspect signal, 3-aspect, 3-aspect with route indicator etc. I was able to set up the decoder for A62 without much trouble. I knew that for A57, with it being 4-aspect also with a route indicator and position lights things would be different, but I thought a second decoder would work: one would be used for the four aspects, the other for the position lights and the feather. Wrong! Traintech don't design their decoders to work like that it seems, so I was getting all sorts of strange combinations. After a bit of research and help from Digitrains, I ended up buying a Yamorc YD8116 switch: This is a very configurable decoder which can be set to operate just about any signal from any country and one of its configs is UK signalling and it will drive a UK 4-aspect signal with route indicator and sub'. You set all this up using a Windows pc interface, so it makes the configuration process relatively simple. Unfortunately the instructions don't tell you a great deal more beyond that and the videos on YouTube manage to leave out some key aspects of the set up which if you don't know about them, you don't know about them and will take you some considerable time to deduce. I had situations where when I wanted a certain aspect, I had alternately flashing yellows and feathers which is not a valid aspect in the UK, if anywhere. This is where taking a break and getting two brains on the job is better than trying to continue and beating your one head against the wall. Cue arranging for Graham to join me the following day and heading out for a walk in the meantime. What the walk did was allow my brain to work out that I had two problems: one with the signal and one with the Yamorc switch. The signal problem was partly due to what I had asked Absolute Aspects to design for me: Normally their signals come with the circuit boards fitted with the appropriate dropper resistors to enable the signal to operate from 12V dc, so each LED has a 2.2k Ohm resistor in circuit. The Traintech decoders don't require resistors because all that is done internally within the decoder so the additional resistors would cause the aspects to be too dim. I had therefore asked Absolute Aspects to exclude the resistors from my circuit boards. Also their signals are also normally wired common negative whereas most decoders are common positive, so I had asked that they wire mine common positive, which they did. Additionally though, Absolute Aspects also design their circuit boards such that each aspect has its own dedicated terminal, so instead of a double yellow requiring connections to both yellows, it only requires one. The subsidiary calling-on lights also illuminate the red aspect. Absolute Aspects achieve this by using a series of diodes. The configuration for the Yamorc assumes that these are all wired separately, so that was one problem identified. The only difference is that it's possible to get the route indicator with any aspect - in theory (even red). In reality, I wasn't getting a route indicator at all with any other aspect. By removing the diodes I managed to get the arrangement to what the Yamorc design assumes is the starting point. The Yamorc provides a 12V dc output at its terminals, so I had needed to put a 2.2k resistor back in again, which I had done on the common positive wire, so all aspects had the same resistance. Some of the aspects the signal will provide via the Yamorc is flashing yellow, flashing double yellow and the above with route indicator. When I had them flashing alternately, this was because the yellow LEDs require less current to illuminate them than the white ones. When the yellow was switched on, most of the current passed through them and not enough went down the white path to illuminate the whites - there are more whites. When the yellow(s) went out, there was no alternative path so the current flowed through the whites and illuminated them. Once I'd understood this, I put different values in each leg until I got the brightnesses where I wanted them. But why was I getting flashing yellows at a all when I expected a steady aspect? That was where the next day with Graham came in: There are so many possible outputs from the switch to drive the signal that it needs two accessory addresses to be called up for each aspect. This is easy enough when it's done via a controller such as the NCE Minipanel which I'm using for all of this, but you need to know what addresses and states (on or off) to command. It turns out that the order in which you send these commands to the switch is critical - and in some cases it's the opposite to what the instructions in the Windows configuration table imply. Once again, if you don't know this, you don't know it and will only find out if you follow a very structured process of trial and error - which is where having two of us on the job was invaluable. Once I understood what was going on, the rest was reasonably straightforward. Big thanks to Graham here for his help with this, working out how many different combinations we could have and for double checking my keypad entries on the cab when we were testing it. His maths degree came in useful after all 🤣😂

Berko and Eckon don't make the types if signals we need, but I came across Absolute Aspects at one of the Warley NEC shows a few years ago: https://absoluteaspects.com You can specify more or less any type of UK mainline signal and they'll build it for you - so I did. In reality, A57 is a 4-aspect signal, so that's what I specified, with route indicator and sub' signal. A62 was spec'd as a three aspect with sub' and I also ordered the banner repeater for A65 which would be A65BR. After a month or so a very nice box arrived in the post: With some even nicer contents. The signals themselves: and the circuit boards and connectors necessary to install them: All that was needed then was to install them. How hard could that possibly be?

It's known as the giraffe house due to it being tall and narrow 🤣

When we rebuilt the layout in 2016/2017, we added some signals on the mainline and around the depot exits, but initially they were non-working. The mainline signals were three-aspect Berko models and for the depot exits we had three-aspect Berko ground signals, similar to those I remember on the Craigentinny departure roads. I had also bought some Eckon ground position light signals for shunt moves. I located the signals where I thought was reasonable and didn't think much more about it for several years. It was one of those projects that I would get around to eventually. "Eventually" occurred a couple of years ago now, and I got the signals working at the same time as I upgraded the auto-operation and route setting, such that when a route was set, it also set the signals and they then reverted to red after the passage of a train. At the time, I thought that was pretty cool and added an extra dimension of realism to the operation of the layout. That was until we had a visit by one of our friends who works for Network Rail and also another who is a driver. They pointed out everything that was wrong with the signalling scheme. Anyone who knows me and also understands that one of the objectives of Kirkhill is to make the operation as realistic as possible (if you forget the fact that it's fictitious and over-developed for a Depot in Aberdeen!), then you would also know that this would niggle away at me (which it did), until eventually I decided that something had to be done about it. To explain, it's first necessary to understand the actual layout at Craiginches and the signalling scheme there as it is (or was) in real life. Graham supplied me with the following screenshot: On Kirkhill, we don't have the Down Sidings, so A63 signal and 107 points don't exist, but we do have representations of signals A57, A62 and A58. A65 is assumed to be beyond Wellington Road Bridge due to space constraints. Our layout plan and interpretation of this is shown below: The north end depot departure signals AK09 and AK10 replace A13 signal on the real-life plan. In our original scheme, AK60 and A65BR didn't exist. Those who have seen Kirkhill in operation will know that when we run the overnight part of our sequence and also on Sundays, we close the Down Main for a civil engineering possession and run trains bidirectionally on the Up Main. In order to facilitate this, our A57 signal has a route indicator on it to indicate that points 25A & 25B are reversed. So far so good, but as our friends in the knowledge pointed out, part of the reason for having a second means of access to a depot in real life is to enable continued access/egress in the event that the main connection becomes unusable for any reason. We included the south end connection because we have a train that departs to the south each morning. We rarely brought anything onto the depot via that connection and when we did, the setup was such that A57 Signal again had its route indictor illuminated, which in reality it wouldn't. In the real world, what would happen is that to allow northbound trains from the Down Main to access the depot, A57 Signal would have subsidiary calling on signal beneath the main aspects - two white lights or cats eyes. These would indicate authority to pass the main aspect at red for a move onto the depot which then becomes permissive, under control of the depot yard staff and the driver must be prepared to stop and obey their commands. The move then takes place via points 25A & 25B and points 28A & 28B into the depot. In the event of the north end depot connection being unavailable, trains would also be brought into the depot from the north on the Up Main via the south end depot connection. To do this, A62 Signal would also have a sub' signal below its main aspect. This would give the driver authority to pass the main aspect at red and to proceed as far as a Limit of Shunt board, positioned far enough beyond the points such that the rear end of the train would clear them. The signal I've denoted as AK60 because it doesn't exist in real life, would then allow the train to set back into the depot via points 28A & 28B, once again under control of the depot yard staff. I said above that A65 Signal is off-scene in our representation, the other side of Wellington Road Bridge, but the line curves through the bridge, so signal sighting would be poor. It therefore seemed reasonable that a banner repeater (A65BR) might be provided to give the driver an indication of whether the signal is off or on, thus reducing SPAD risk, hence A65BR being shown on our plan, before the bridge.

Thanks Andy - much appreciated as always. Graham built the access platform and steps from a kit a few years ago and it was in the original shed for ages - he can probably remember where it came from. It wouldn't have been very practical in the main shed in reality, because that has side pits so the access platform couldn't have reached the side of any rolling stock before it fell into the side pit! It's much better where it is now, so it justifies all the work of incorporating the new shed.

I've been progressing a few improvement projects for the layout recently. The first one I'll tell you about, because it's independent of the other two, is the new HST power car repair/maintenance shed. When I say new, it isn't really; it was a lockdown project from April 2020, when I started to build a new shed for something to do. The shed during construction back in 2020: Cladding applied, still to be primed and painted: It's been sat on the worktop in my project room ever since, without a plan to utilise it. Towards the end of last year, it occurred to me that we could add further interest to the operation of the layout by swapping power cars on the HST sets that arrive on the depot for servicing. Back in the days when I worked at Craigentinny Depot in Edinburgh, power car changes were a nightly occurrence, with a target of five per week off exam or repair (usually engine repairs), plus additional swaps between sets for positioning moves, so they finished at the required location the next night (or subsequent nights) to be taken off there for exam etc. Although power cars were allocated to a home depot, Maintenance Control at York would balance out the workload between depots, so you would occasionally get a foreign power car for repair if we had one that was approaching completion and there wasn't a home power car in need of attention or in the right place. Keeping the service running reliably was a balancing act between getting power cars to their scheduled exams on time and removing power cars with defects from service before they became too much of a liability. Given that Kirkhill is largely fictitious and would not have been the size that it is anyway in a location such as Aberdeen - it has a large maintenance shed and a wheel lathe - it wouldn't have had a power car repair shed either, but so what? It's our layout and it would add interest, so on that basis it seemed worth a shot. I realised that if I shortened my lockdown shed, it would fit on a dead-end road at the north end of the depot and these roads don't tend to be utilised that much anyway. The idea was to have the shed, with an inspection pit, with lighting and jacks and other equipment such as would be found in a typical power car maintenance facility. As I said, as originally built, it was one bay too long, so I had to remove some of the cladding and glazing, cut sections of the plywood out and then glue it back together again and reapply the cladding etc. The following photos show how it's turned out thus far. Trying it for size at the north end of the layout: Preparing the ground, Peco inspection pit cut into the baseboard: Ground works nearing completion: View from the public viewing side, power car in position between the lifting jacks (Bachmann Scenecraft): Interior with overhead and pit lighting, but without interior fittings etc.: View from the entrance to the shed: External view: Future improvements will be the addition of workbenches, lockers and general equipment - all from West Hill Wagon Works Ltd, who make some excellent 3D printed items. I also intend to have a flatbed articulated Swift's truck delivering overhauled bogies from Crewe, as used to happen following the move to component exchange maintenance in the late 1980s. A friend and I are also investigating the possibility of fitting a loco lifting system to simulate a power car being lifted on the jacks for underframe attention / brake disc renewal - which was a frequent overnight activity before the introduction of BSI one-piece brake discs. More of this in the future.

I haven't posted on here for a while, but I'm hoping we will have some exciting news to share with you in the near future. I've been asked to submit costings to take Kirkhill to a major show this autumn. The organisers have a lot of planning to do and so I'm not going to steal their thunder at this stage, but it's good to know that we're in the melting pot for what will be the layout's first outing since Taunton in October 2019.

Picking up where we left off with the route setting and automation: I decided that if we were going to be able to make use of the automation, the system had to have a means of performing a self-check between setting a route and starting the next train, so I came up with a scheme for route proving. This began as a simple concept: each point in the fiddle yard on the ladder has a corresponding point at the other end of the ladder and they have to be set in pairs (one at each end) for the route to be set because trains leave from and return to the same road on the ladder. Each point has a Peco switch attached to the point motor in order to change the frog polarity. When originally built, we had Peco double microswitches on the points at one end and single sliding switches on those at the other. By changing the single switches to double microswitches, there was a spare changeover switch on each point that could be used to detect its position. By daisy-chaining them together in pairs, end to end, it was possible to create a proving circuit. This could then feed into the spare 8th input on the Minipanel. An extra line in the macro for each part of the sequence would instruct the Minipanel to wait until it saw Input 8 grounded before starting the next train. The circuit for the fiddle yard route proving is shown below: To explain the way it works: Wire 2700 would be wired to Input 8 on the Minipanel; wire 2200 would go to Minipanel Ground. With all points set correctly, Input 8 will be connected to ground and the route would be proved. (Ignore the references to a relay panel for now if you will - this comes later- the purpose of this is to get the route proving concept across.) There are five roads in the fiddle yard ladder controlled by the DCC with points numbered 1-5 at the left hand (south) end and 18-22 at the right hand (north) end. For convention, if a point is set straight, it is normal (N), if it is set for the diverging route, it is reversed (R). Points at the south end are managed by Decoder SMD84 (1) and those at the north end by SMD84 (2). The diagram below shows the point numbers: For the first road on the ladder to be set therefore, points 1 and 22 must both be normal – i.e. 1N + 22N. In decoder speak, this is Route 101R – this is the nomenclature given to the first route described by the Team Digital SMD84 route CV list. Remember, each decoder only sets half the route the way I’ve done this – either the north half or the south half. In the proving circuit, the switches represent the microswitches on the point motors that I’ve used for the route proving. The frog microswitches aren’t shown in the diagram because they are irrelevant here. For the route to be proved, the switches on points 1 & 22 must agree and current flows down wire 101, as shown below. For the second road on the ladder to be set, points 1, 2, 21 & 22 must all be reversed – i.e. 1R + 2R + 21R + 22R (Route 101N). In this case, for the route to be proved, the switches on points 1, 2, 21 & 22 must agree and current then flows down wire 201 as shown below. It works in the same way all the way down the ladder, so for the fifth route, 103R it requires 1R + 2N + 3N + 4N + 5R + 18R + 19N + 20N + 21N + 22R. Provided they all set correctly, the route is proved if current flows down wire 501 as shown below: If any point fails to set correctly, the route won’t be proved and the Minipanel won’t start the next train. It is possible that two corresponding points will fail to set and a false route will be proved, but this is unlikely because the points at each end are being set by different decoders working independently of each other and experience shows that point mechanical failure is random.

It works exactly as per the video - and I got it to work with the entire operation from one function with a Zimo decoder, but the slow running with the 08 just wasn't as good as it is with the Lenz Gold plus Power 1. Where I've fitted servos to both ends, I've used two different function pins and mapped them to functions 1 & 2, but you could run both from one function. It can be a little fiddly tying the thread off, but I've got it to work now on three locos and it's so much easier than having to stop on the uncoupler every time; you can just drop vehicles where you want. It also assists with coupling on curves too.

Glad it helps - there is more to come. There is another solution to your uncoupling issue in terms of having to stop accurately over the magnet: I fitted my Class 08s with onboard remote uncoupling controlled by a function key on the cab handset. See Precimodels: https://precimodels.com/en/ These use a small servo to open the knuckle on the Kadee via a short length of thread. It looks a little Heath Robinson at first, but I can vouch for it and say it works really well: With some decoders such as Zimo, there is a group of CVs that control the whole uncoupling sequence (squeeze in, operate the servo, pull away, stop and release the servo) - all from one function key. That would save you several lines of macro. Having used them successfully on the Class 08s, I'm now thinking of fitting them to the HST power cars so I can do power car changes without the need to stop over an uncoupler. On the HST, I reckon the servo can be concealed within the air reservoirs. I don't use Zimo decoders on the 08s because I favour the excellent slow speed running you get from Lenz Gold decoders plus the Power 1 stay-alive units, so I just wire the servos to the Function 1 output. Lenz don't support auto uncoupling unfortunately (or not on my decoder versions), but using the appropriate decoders could be an option in your case.

I considered it briefly and in fact did wonder about using the Raspberry Pi for this, but the Minipanel seemed to offer a much easier way to go, which was more compact and meant hauling less kit around to exhibitions, plus it's cheaper. I think one day, if I ever get endless time, the computer based approach would be an interesting project, but for now, life is too short! I have now managed to get the Minipanel to do exactly what I want it to do. It just needed some help, that's all. That is the subject of my next post.

I know the consisting trick will work because I've tested it and I might even use it on Kirkhill. I think your proposal is certainly more ambitious than mine, but I understand the concept and it should be possible to get it to work. The challenge I can see you having is getting consistent performance from different locos and therefore getting them all to stop accurately in the same place. I'm sure you can get the same loco to stop accurately with some trial and error - repeatability, but I think the challenge is achieving reproducibility with different locos. I find that even what appear to be identical locos, with apparently identical decoders, with the same CV settings still behave differently. Finding decoders with the same version number is also a struggle - we use Lenz Standards on our Heljan Class 47s and there are umpteen different versions of Standard decoder and they all behave differently, not to mention differences in performance of individual locos. I've started using Zimo decoders on the HST power cars so I can consist them reproducibly - CV95 is a reverse trim that assists with speed matching, but even then, accelerations can vary through the speed range. I've tried playing with the CV for constant braking distance but unsuccessfully so far. Another option which is similar to analogue would be to use a relay to cut the track power to the section where you want to stop. That would stop your train dead and probably always in the same place, whatever the loco. If it's in a fiddle yard, a dead stop shouldn't matter provided its at low speed because you're after accuracy, not realism. Something to consider is that you have quite a few sensors - 15 if I counted correctly? Each of these will use up a Minipanel input and there are 30 inputs, so that's half of them. If you program using continuous memory, you will be starting from Input 16 - maybe 17 if you need an input to trigger the start of the sequence like I do, so you would have 14 inputs left for programming, that's only 56 lines of macro. Selecting and starting a loco uses two lines of macro; stopping it uses a third. I think you may find that to control your whole fiddle yard is beyond the capacity of one Minipanel. You can of course add another and possibly a third. NCE Powercab or Smart Boosters will drive up to six cabs (addresses 2-7) and three accessories such as USB interface, Minipanels etc. (addresses 8-10). This means the maximum number of Minipanels in use at any one time is three, provided you aren't using a USB interface or other accessory on the cab bus. I use a second Minipanel to control the signalling - to be the subject of a later post and I use the USB interface so I can drive locos using my phone via JMRI. There is another trick you can use - which I did to get around lack of capacity in the Minipanel, which is to use macros in the NCE cab/booster system to control accessories and then use the Minipanel to call up the macros. This will save you some lines of code in the Minipanel, but the number of macros is limited to 16 and each macro can control up to eight accessories. I'll cover that in a later post regarding signalling and route setting too, but hopefully it gives you an idea of how you can manage a series of trade-offs between decoder route setting, Minipanel macros and system macros to achieve what you're trying to do. Is it easier than using an Arduino? I don't know to be honest because I haven't tried that. I have a Raspberry Pi which I got for the signalling project, but that seemed too much like hard work. Certainly the programming interface for the NCE bits is easier because it's all menu-driven, but it probably has more limitations. Incidentally, I use the Pi along with the USB interface to run JMRI so I can use smartphones as wireless cabs. Hope this helps.

Andy, I'm pleased that our lessons learned are of use. It didn't feel good at the time, but as they say: every day is a school day. I've been giving some thought to your question about needing dedicated locos. The way we have things set up, there is a day time and a night time sequence and each has its own dedicated locos and a dedicated Minipanel. They are housed in boxes with an RS232 connection to link them with the control panel. The one in the blue box is for day time, whilst the one in the black box is for night time. Nicely poke yoke, but it's important to get the right loco on the right road. The thing about DCC is that wherever the loco is on the layout, if the Minipanel commands it to move, it will move whether you want it to or not! With DCC as most people know, you drive a loco by its number, so in our sequence, this is programmed into the macro that the Minipanel runs. There is another way though and I've just been and tested it and it works: Within the Minipanel macro, it would be possible to set up dummy loco numbers - 1, 2, 3, etc. These can be the loco numbers that get coded into the program. You would then pick which loco you wanted where in the sequence and then consist it with the appropriate dummy loco. With advanced consisting, the consisting data is written into the loco decoders, so the DCC control system neither knows nor cares if all the locos in a consist are there or not. This means that the dummy loco is invisible to the system. DCC systems allocate a unique number to each consist. With the NCE SB5 booster (and Powercab) this is in the range 112 - 127, so 16 consists. With some systems, you have to drive using the consist number, but with NCE you can drive using either of the loco numbers. This means you can drive using the dummy number regardless of whether it really exists or not. If the Minipanel is programmed with the dummy loco numbers, this would work. You would have to remember to cancel the consist if you wanted to use the loco elsewhere on the layout and replace it with another in the consist, but this would avoid having to have dedicated locos to avoid reprogramming the Minipanel. There may be another way of doing this, but I can't think of it just now. Hope it helps.

Thank you Andy! It's not often someone hangs on my every word - hopefully I can continue to share useful experience with you and others.

To continue the story: The set up for the Shildon show went well on the Friday, so we were looking forward to the Saturday and our chance to show off with more trains being run by fewer operators. What is it they say about pride coming before a fall? The layout set up at Shildon on the Friday night, ready for its big day: Within the first hour, disaster struck! One of the points failed to set during one of the route setting parts of the macro, so a train set off and derailed. It soon became clear that one of our fancy SMD82 switch machine drivers had got its CVs corrupted and we were unable to change most of the points at one end of the fiddle yard, either using DCC commands or via the pushbuttons. The pushbuttons still need the decoders to be working. I spent Saturday afternoon with the control panel in bits, soldering iron in hand, re-wiring the panel to work directly by the pushbuttons and the CDU, which had been retained for those points that weren’t to be part of the automated sequence. Not the best way to run a model railway at an exhibition. Thanks must go to Graham and Alex for keeping some things running at least, whilst I was trying to fix the fiddle yard. Quite surprisingly, we actually won the trophy for Best in Show on Sunday afternoon, so what seemed to me to have been a disaster can’t have been as obvious as I thought it was! Once we got the layout back in storage, I set about trying to understand what had happened. Trying to unpick the rats nest I had made on the Saturday afternoon: I got in touch with Team Digital to ask advice on the reliability of the SMD82s. They told me that there was an issue with the software versions on both of mine. One was V16, the other V17. Apparently, anything below V18 was vulnerable to corruption during power-down. Interestingly, we had had a short circuit caused by an operator error, just before the problem occurred. During a short circuit, the DCC circuit breaker operates and shuts down the track power; the decoders were connected to the track for power and commands, so they lost power as a result. The protection in the circuit attempts to reset and re-establish power again after about a second. If the short is still there, which usually it will be whilst the operator works out what has happened, the system will power down again. A second later, the process repeats, so the decoders experience repeated shutdowns and if they are susceptible to crashes during a power down cycle, a short circuit is almost guaranteed to cause one. Shorts are also more likely during exhibitions due to operators working under pressure, so we had the conditions for the perfect storm, which of course, we got! Team Digital offered to provide new chips loaded with new, reliable software (for a price), but given that one of the decoders had a failed output, I decided upgrading to the later version – the SMD84 – was a better option. I bought two from Tony’s Train Exchange in the US. We had another show coming up at Taunton later the same October and I just managed to replace the decoder which had failed with one of the new ones in time – and get the CVs programmed in it. The SMD84s have even more CVs to programme – 256 this time. I decided to take a chance on the other SMD82, which hadn’t failed at Shildon, lasting through the Taunton show. Off we went, hoping once again to be able to use our automated system. Testing the fiddle yard in the kitchen, prior to Taunton. Prior to extending the garage, this was the extent of testing that I could do at home: At least this time, it had the decency to fail whilst we were testing it on the Friday evening after setting up. This time, it was the V16 software that went belly up – again after an operator error induced a short. This time I was ready for it though! I had taken the precaution to wire in a couple of changeover switches and an extra couple of Sub-D connectors which meant I could quickly revert to manual CDU control. Modification made prior to Taunton. By splitting the sub-D connector and changing two switches, the pushbuttons worked through the original CDU and connected to the board via the right hand half of the connector pair: Taunton was the last show we went to before Covid-19 turned the world on its head, so nothing changed until I got the layout set up in its new home last November. At that point, I got the second new SMD84 decoder installed and programmed, so once again, I could test it out and try to break it under more relaxed conditions. I also used my spare NCE SB5 booster as a dumb booster to provide a dedicated separate power supply to the decoders, so a short on the tracks, doesn’t shut down the SMD84s. I still managed to break it! What I found was that occasionally, a point doesn’t throw due to mechanical resistance. If you’re running the layout manually, you notice and press the button again. Nine times out of ten that will fix the issue, but when it was running automatically, the system didn’t have the ability to check that the route had set before starting the next train. There was just a delay in the macro to give the route time to set, but the assumption was that it would do so. Clearly that assumption was unreliable.

Agreed. There were three manuals that I found which were useful for my purposes, but still not comprehensive enough: The Minipanel manual The technical reference manual (but after 12 years, this is still in draft! Come on NCE; get it issued properly!) The Minipanel examples or Minipanel Autoprogram for the benefit of others following this, references are here: https://ncedcc.zendesk.com/hc/en-us/categories/200170015-Automation-Detection-and-Software I found the autoprogram or examples reference was essential to understand what the manuals themselves were attempting to explain. Like I said, if you understand what the ting does, or if you have someone to explain the basics to you, then maybe the manuals are self-explanatory. I didn't find that easy to follow myself though, but I eventually got it along with some trial and error. To be fair to NCE, this is a little similar to someone buying a computer which can do an infinite number of things and then moaning that the manual that came with the computer doesn't tell you how to use the CAD package with it that you also bought to run on it! Having said that, a little more information on exactly what the configuration values really mean would have been useful, but maybe that's just me!

True Graham, it does, however to get the best out of DCC, you really need to wire droppers from every rail to one of the DCC power bus wires (as we did on Kirkhill), so again it's more wires, but over shorter distances. As you say though, it does avoid all that panel wiring and having to run the wires to the panel.

On to the next instalment on layout automation: I wanted trains to start and finish on one of the ladder roads in the fiddle yard. This meant the fiddle yard had to be fitted with block occupation detectors, so the Minipanel could detect when a train had completed its journey round the layout. There were sufficient outputs on the SMD-82s to control five points each on the fiddle yard, plus the points which control access/egress from the up or down mainlines. This gave five fiddle yard roads on the ladder, so two trains in each direction, plus one that would run bidirectionally, so six block detectors were required. The SMD-82s each have 112 configuration variables (CVs) which can be programmed to define point control, pushbutton inputs and routes, although only 34 of these on each decoder were required for the application on Kirkhill, it was still a fair amount of programming to have to do. Programming the Minipanel is straightforward enough, which is just as well because the instructions in the manuals are really quite poor. As with many things, if you know what you’re doing, it’s quite easy but if you’re starting from scratch it’s a case of trial and error, looking through examples online and several versions of the manual. The impression I was left with was that the manuals were written by someone who assumes you’re already an expert! In its simplest form, each input will run a macro containing 4 lines or steps of command – one command for each line, but that wouldn’t work for my application because it needed more commands. The manuals tell you that it’s possible to have “continuous memory” from a “certain input”, but they don’t tell you which input or how to define it. You can also link from one input to another to continue the commands, but each link uses up a command so that isn’t very efficient. Eventually I worked out that there is a config set up option, where you define The cab bus address (everything on the bus has to have a unique cab bus address to avoid conflicts with other cabs/controllers - the address ranges for different devices are defined in the NCE PowerCab or booster manuals, so these are quire easy to determine) Where the continuous memory starts from - i.e. which input (the bit I couldn't figure out from the manual) Whether or not you want the grounding of particular inputs to cause the Minipanel to execute the command line associated with that input (makes for more efficient programming – more of this later) Whether you want the Minipanel to execute a series of actions at start-up – i.e. when it is first connected to the cab bus or when you turn the layout on (again, more later). The diagram below shows the set up in the fiddle yard for the sequence, which trains and locos/power cars are used, their place in the sequence, the directions of travel and the route numbers to be called up to set the routes before the trains move. This gives a sequence of six trains – 3 in each direction – and we timed the sequence and it takes about 5 minutes to complete. This gives the operator time to take a break, or sort out the next train to run to or from the depot etc. To start the sequence requires two push buttons to be pressed – these are wired in series, so both must be pressed together to begin the sequence. This avoids accidental initiation of the sequence. These are connected to Input 1 on the Minipanel and this is the only wired input that actually has a command associated with it. The six block detectors are wired to the next 6 inputs so only the first 7 inputs have anything connected to them. Continuous memory starts from Input 9 which is where the programme itself begins; I left Input 8 spare in case I needed to connect something else to it. The command for Input 1 simply says jump to Input 9, so when the buttons are pressed they initiate the programme. The screenshot below shows the first few lines of the programme (macro) as an illustration, with comments to explain. I decided to record it in a spreadsheet for simplicity, but the programming itself is done via a cab handset, plugged into the programming port on the Minipanel. You can also programme via JMRI, but I haven’t tried this yet. I set all this up during the summer of 2019 and we then tested it out by hiring Mickleover Community Centre in Derby. As always, there were other jobs that needed doing on the layout, so the testing of the Minipanel was left until last, but everything appeared to work satisfactorily, so we were good to go for the next show we had booked at Shildon in early October, or so we thought!

Absolutely. I remember Hornby's Zero 1 system which I think boasted just two wires - and of course the saying that there's no such thing as a free lunch is as true in model railways as it is in any other walk of life. You either have "just a few wires" and use DCC but have to press loads of buttons to get one point to change, or you have loads of wires in order to just press one button to produce the same effect. There aren't any short cuts if you want the system to work properly. I think that whilst DCC doesn't really reduce the numbers of wires that you need, it can reduce the lengths of the wires, it enables you to do more and changes can be easier to make afterwards if you can do them by changes in software rather than having to change hard wiring. In the end, it's a case of horses for courses; each to their own.