jamespetts

Goodness me, that is very impressive.

Interesting. Those are quieter than I had imagined. The footprint seems similar to - perhaps slightly greater than - a servo motor with the Dingo mount.

Interesting indeed - although the problem with using the DR4018 is that this is DCC control. This either requires using the track supply DCC bus, which can introduce unreliability on all but small layouts, or having a separate DCC accessory bus on top of the LocoNet bus, which adds unnecessary work and complexity. One would probably be better off using a LocoNet circuit card designed for stall motors to power these.

I have heard of these MTB MP1 motors; they use, I think, an ordinary rotary motor rather than a stall or servo motor. They are a little louder than servo motors, I think, but that may be of more importance to some than others. They cost £15.50 each, which is a little more than an inexpensive servo plus a good servo mount. They probably take less work to set up than a servo, which requires the construction and installation of a mount. That they take no current when idle appears to be an advantage over stall motors, although, looking at the specification, they appear to have only one output switch, which would make it difficult (without added electronics) to switch both frog and feedback simultaneously. Also, I have not fully parsed the control wiring, but it looks as though it needs 12v applied on one of two terminals to determine which way to go; this would appear to be the same wiring scheme as traditional stall motors such as the Tortoise or Cobolt IP Analogue. There are, I believe, circuit cards that can switch analogue stall motors from a LocoNet input (RR Cirkits, I think, from memory), although I have no experience with these. I am not sure how the footprint compares to a servo motor, nor whether the output pin can be cranked to allow fitting in especially tight places (this can be done successfully with servos and also, I believe, with stall motors), but I can believe that these have a smaller footprint than a stall motor. On balance, these appear to be potentially viable for use in a computer controlled layout, albeit with some disadvantages (albeit also some advantages) as compared with servo motors.

I use the Digikeijs occupancy detectors (the DR5088RC); I have not used the Roco detectors, but I presume that these work in largely the same way. The RailCom signal can indicate what locomotive is in a given section. (There is also RailCom channel 2, which can in theory transmit data such as speed, etc. but there is no standard protocol for this and I have not seen any application that uses it in any way; I have seen reports that it is unreliable). The RailCom detectors have inputs from each section (17 in the case of the DR5088RC) and then transmit occupancy data and RailCom information down the LocoNet bus to the command station. (Other RailCom capable occupancy sensors may use another bidirectional bus such as XPress net to do this). What these occupancy detectors cannot do is encode the section that the locomotive is in and send this together with the datum comprising the locomotive's DCC address back to the command station using the DCC bus: they need to use a bus specifically built for bidirectional communication, such as the LocoNet bus, to do this. Also, as a secondary point, actual occupancy sensing is generally kept separate from RailCom; the DR5088RC uses a standard current draw sensor in addition to RailCom sensing. This is for two reasons: (1) if RailCom were used as an occupancy sensor, the occupancy detection would not work with vehicles with decoders not equipped with RailCom decoders; and (2) RailCom is very fragile because it is trying to be bidirectional on a bus (the DCC bus) not designed for this, so it is much less reliable than standard occupancy sensing. It would be theoretically possible to build occupancy sensing that uses RailCom instead of current sensors to determine occupancy of individual sections of track (albeit it would have the disadvantages set out above), but it would still need an additional properly bidirectional bus, such as LocoNet, to send the occupancy information back to the command station.

I have just come upon this thread; I notice that Wordsmith is planning something similar to what I am working on for my own home layout; the test layout is also very similar to the test layout that I put together in 2018 for the same purposes. I thought that it might be helpful to share some of the knowledge that I have acquired in that time. First of all, Nigel Cliffe, who has posted in this thread, gives very good advice. In case of conflict between that and any other advice, I suggest taking Nigel's advice unless there is very strong evidence that he is mistaken in the case in question. Conflicting advice - sometimes given very stridently - is sadly common, so it is useful to be able to find reliable people. Secondly, automation software: JMRI is good control panel software, but extremely limited in automation. I looked into this in some detail. JMRI lacks many of the abstractions necessary for automation to work with built-in code (e.g. there is no abstraction for a schedule), so having full automation in JMRI essentially means writing a complete, full-featured piece of software more or less from scratch in JMRI's scripting language, the only parts being done for you being the GUI and the hardware interfacing. Scripting languages are far from ideal for such a large undertaking and a big problem is that, as JMRI changes over time, the scripting language calls change such that a very large piece of software written in it becomes very fragile. I do not know how experienced that you are with coding, but you will need to understand how to write software and how to make software robust and maintainable using appropriate abstractions in order to make a workable automation system that is not very fragile (i.e., one that does not need rewriting from the ground up if you make any small changes to your layout or its operation). Some people report finding it easier to write an entirely independent piece of bespoke software than using JMRI's scripting language to achieve full automation. Be prepared, therefore, to invest the sort of time and effort necessary to write an entire computer program from the ground up if you want full automation in JMRI. For my own purposes, I chose TrainController. iTrain, which is cheaper and has a better interface and better support, is slowly catching it up, but it still lacks the ability to use variables in macros, which is important for sophisticated customisation. Third of all, turnout motors: stall motors (such as Tortoise motors or Cobolt motors) have the disadvantage of having a large footprint. This can be a problem in N gauge, as it may not be possible to use them to motorise many turnouts in close proximity to each other, although there may be some scope to work around this using cranking of the rods. I ended up using servo motors in the end, which take more setting up but can be mounted in a smaller footprint and are also much cheaper. Finally, stall motors draw a significant amount of current when idle, whereas servos draw none. This can be a problem when you have a very large layout, as the idle current draw may exceed your power supply's capacity. My own layout uses a 1A 12v auxiliary supply. The more amperage that your power supply puts out, the thicker the wires that you will need so as not to create a serious fire hazard in the case of overload: generally, your wires need to be rated to carry the maximum current output of your power supply unless protected by a circuit breaker, fuse or similar (e.g. District Cut-Out), or else there is a risk of overheating and fire. 10A (or even 5A) rated wires are impractically thick. Even 3A rated wires are quite thick. Do be careful what sort of servo motors that you buy: digital servos are more widely available, but they cannot be turned off at the end of their travel easily, meaning that they are very prone to buzzing, creating irritating noises, and, more importantly, constantly drawing currant, which will, if enough are affected, have the same problem as with the stall motors (and possibly worse). Constantly being under power may also overheat the servos. However, analogue servos have their own problem in that they will twitch when first turned on in many cases. If you have a great many of these, you will find that them all twitching at once will draw too much current from your power supply, tripping the breaker, blowing the fuse or simply causing the power supply's voltage to drop to very low levels, which would then deactivate the servos, allow the voltage to be restored, causing the servos to twitch again and cause the whole cycle to repeat indefinitely. What I do to avoid this problem is use analogue servos with power on delay cards connected many of the servo cards to stagger the turning on of the servos and thus not draw too much current at once. As to the Digikeijs unit, I have one of those. There are problems with these units in that they sometimes become unresponsive when connected via USB. They can also be connected by ethernet, but TrainController is not compatible with these units when not connected via USB (but iTrain and JMRI will work with these connected via ethernet). It is possible that a recent firmware update has corrected this, as I have not noticed this problem when I have used this recently, but I cannot be sure of this. I use a self-soldered Hans Deloof command station, albeit that has its own disadvantages, including the inability to use it for programming using the computer (and JMRI's DecoderPro, which is the best DCC programming software that I know of). As to buses, you will need a LocoNet bus for your feedback units. This is because DCC is, with the exception of RailCom, which cannot be used for occupancy detection, a one way bus: it can be used to communicate commands from the command station to the locomotives or accessories, but it cannot be used to communicate data back from the locomotives or accessories to the command station. Since automation requires occupancy detection so that the computer knows where a train is at any given time and therefore how to drive it, automation requires a bus capable of feedback. The LocoNet bus is not the only bus capable of doing this: other buses, such as the MERG CBus, the new LCC and Lenz's XPress Net are all capable of feedback for occupancy detection, but the Digikeijs occupancy detection units such as the DR5088RC work with LocoNet, and, because it is a bus with an open protocol, many other devices also work with this. I have found that this is the most flexible of the buses to use. Using LocoNet with servo motors for points/semaphore signals and LocoNet servo control cards (and LocoNet IO boards for colour light signals), you can dispense entirely with the DCC accessory bus. You will then have: (1) a DCC bus for your track; (2) a LocoNet bus; and (3) an auxiliary power bus. These are easy to manage with cable colours, especially as the LocoNet cables are obviously different to plain wire and need no further distinction. For my own part, I use red/black for the DCC bus and yellow/blue for the 12v accessory bus. Note that there are two different varieties of LocoNet bus: LocoNet-B and LocoNet-T. They both use the same sort of cable, but two of the wires in the cable do different things in each case (I recommend using different coloured cables for each; I use white for Loconet-B and black for Loconet-T). LocoNet-B is the default type of LocoNet and using this type is necessary to connect boosters as it carries the railsync signal. Loconet-T is used for throttles, as the pair used for the railsync signal instead carries 12v DC. Many of the Digikeijs units are powered by the 12v on the LocoNet-T bus, so need no independent 12v bus (although the circuit cards to control servos from LocoNet do require a separate 12v bus; these take a LocoNet-B connexion (I think that they just pass through the pair that is railsync in LocoNet-B and 12v DC in LocoNet-T so it is possible that they will take either connexion, but I am wary of the current rating if these have to pass power)). Do not underestimate the number of feedback sections that you need, especially if you want to do any coupling/uncoupling automatically. If you want, for example, a locomotive automatically to back onto a waiting train and stop in the right place, then the automation software needs to know that the locomotive has entered the section so that it can slow it to stop at just the right point that it couples with the carriages without pushing them back. However, if your platform is all one section, the computer will not be able to detect the locomotive entering the section, as it is already occupied by the carriages. Thus, you will need an extra, short, section in advance of the place where the carriages will stop to use as a detection section for the locomotive backing onto the carriages. (In theory, you can work around this by using the exit time from the previous section, often a turnout; but most feedback units have a delay in reporting the clearing of an occupancy sensor that they do not have in reporting an occupancy sensor becoming activated (deliberately to avoid noise caused by imperfect connexions), so this will be far less reliable). Also, I recommend using District Cut-Outs. I use the MERG type (and I also recommend joining MERG). This will prevent a short circuit on one section of track disabling your whole layout. It will also allow you to use 1.5A rated droppers even if your power supply is 3A (set the DCOs to a 1.5A rating). Make sure, however, if you want RailCom functionality, to use the 1.2 version of the PIC's firmware for these, which I believe that you still need to download and flash onto the PICs yourself at present. You will probably not need these for your test layout, however. As to RailCom, if using iTrain or TrainController, the software will keep track of a train's position automatically using only basic occupancy sensors. Where RailCom is useful is in seeding the information as to what locomotive is where when you first put the locomotive on the track. Thus, you do not need RailCom feedbacks for all sections: only those where you are likely to be manually placing locomotives on the track. Also for occupancy detection, you will need the rear of any given train to draw current. For a multiple unit, this will not need any additional work, as most modern multiple units have rear lighting that will do this. For hauled carriages, however, intervention is needed. There are one of two things that you can do: either fit carriage lighting, which will draw current (a tail lamp for a goods wagon will suffice), or fit a resistive wheelset by putting a 10k ohm resistor across each axle of the rear carriage of a train (or both front and rear carriages if the train may run in either direction). If you later fit carriage lighting, do not forget to remove the resistors. For N gauge, you will need to use surface mount resistors; use glue and conductive paint. Avoid resistive paint: I have found this to be too unreliable to use. For automation, you will need high reliability. This means following the dropper per section of track advice, using live frog points and, especially in N gauge, fitting stay-alive capacitors to all your locomotives (or alternatively, for multiple units, using little pin header plugs to communicate track power all down the train). Without these measures, there is a high chance of any given locomotive stalling after a time. For N gauge, you will need to use tantalum capacitors. I find 4x 470uF capacitors to be ample for N gauge locomotives. Using Zimo decoders, which seem to work well, and the SACC16 charging circuit, this can quite easily be achieved (although note that my N gauge layout is diesel - others will have to comment on fitting these to steam locomotives). One quirk; the capacitors that are small enough and hold enough charge to be useful in N gauge are all 16v. This means that the SACC16 unit must step down the voltage if the track voltage is >16v. They do in fact do this, but only down to ~14v on an 18v input. This is not ideal, as capacitors will age quickly run near their rated voltage. Also, any failure in the SACC16 that passes full track voltage to the capacitors could cause them to explode, damaging your locomotive, although the chance of this occurring in normal use is very small: see here for details. The Digikeijs command station and boosters determine the output voltage using the input voltage. For N gauge, I recommend selecting a low input voltage (perhaps 13-14v) to give a correspondingly low output voltage so that the voltage passed to the capacitors does not exceed their rated voltage. N gauge does not need higher voltages in any event. For uncoupling, I recommend the Dapol Easifit couplings. They are much more unobtrusive than the default N gauge couplers, and can be uncoupled using magnets beneath the baseboard and thus entirely invisible to the scenic section. The magnets sold by Dapol are ceramic magnets (i.e., magnets which are not very strong) and are intended to sit between the rails. This is fine for a fiddle yard, but looks awful in the scenic section. These are also permanent magnets. If there is ever a place that you might want to uncouple sometimes but not others (e.g. a platform where locomotives change), then you will need a controllable uncoupler, and a permanent magnet will not suffice. Although the Easifit couplings are supposed not to uncouple when passing over a magnet when in tension, in reality, the slightest wobble or hesitancy of the locomotive (or even a slight judder caused by running at slow speed) can cause a train to uncouple over a permanent magnet, so I advise using permanent magnets only where you can be quite sure that you will always want uncoupling to happen, such as at the end of a terminus platform. For carriages that will remain permanently uncoupled, I recommend the Dapol NEMCoup basic knuckle couplers: they are cheaper and more unobtrusive than the Easifit couplings and cannot accidentally be uncoupled by a magnet. For places in the scenic section where you want permanent uncoupling, some rare earth magnets buried in the cork beneath the track can work well. Where you want optional uncoupling, the best solution that I have found is to use more powerful permanent magnets under the baseboard raised and lowered by a servo. Dingo Servo now sell a mount for this with easy assembly instructions (you will need to buy the magnets and servo separately), based on a design that I and some others at the Model Railway Club helped to create last year. I also find that the Dingo Servo Micro 10 mounts are good for mounting servo point motors (on which, see above). Plan your final layout for a good sized minimum radius. This will allow you to use shorter arm couplers, which will, in turn, make your trains look a lot better (and take less space, possibly allowing longer trains). The lowest minimum to aim for is 305mm, although an even larger minimum radius would be better, especially as it is easy for a slight error in track laying to make a part of the curve less than the desired radius. In any event, automation is a joy and can add so much to a model railway. Very best wishes with your projects, and I shall be interested to be kept up to date. Do have a look at my video of how I am progressing with this so far and what one can do with it.

Best wishes for a speedy recovery - in the meantime, the grilles on the 47 look splendid.

I use TrainController and find that Zimo decoders woek well for me.

I have tried re-printing these using a black resin material ordered from 3dCompare. The results are below: the locomotive in the foreground is shown with the new deflectors, and that in the background with the old. Lord Nelson smoke deflectors by James Petts, on Flickr Lord Nelson smoke deflectors by James Petts, on Flickr Lord Nelson smoke deflectors by James Petts, on Flickr These are the old deflectors. Lord Nelson smoke deflectors by James Petts, on Flickr As can be seen, the black resin material gives more detail and a smoother finish.

I approve of Craven tanks. Craven seemed to be of the same mindset as many a railway modeller, seemingly wanting just one of every varied type of locomotive.

Thank you! I shall only be able to have it once the Peco Bullhead slips and crossings become available at some unknown time in the future. Also, I approve of your approval of Terriers and push-pull stock.

This past year, I have been focussing on the N gauge layout, whose track laying needs to be completed before work on this layout starts. The pandemic and consequent delays in the availability of Peco Bullhead slips and crossings also means that the start of this layout has been delayed. However, to have a change from the vast amounts of code 40 track building for the N gauge layout, which is happily nearing a close, I have returned briefly to some planning for this layout, in particular, planning the necessary rakes of carriages and corresponding services, and then making some consequential amendments to the fiddle yards in the track plan to increase flexibility of operation and to add capacity. Here are the plans for the various rakes for the Bournehampton guise: The scheme of these plans is to allow as much as possible to be ready to run stock initially, but to allow greater variety and accuracy to be added with kit built stock as necessary slowly over time. Under the locomotives column in each sheet, the red types are those not currently available ready to run. The idea is that some longer rakes can be formed by combining some of the shorter rakes (as described in the appropriate columns above). This will require a fiddle yard pilot engine to do this automatically. In researching the Chichester slow trains from Portsmouth on which the Chichester services on this layout would be based, it came to be apparent that these would have been likely to have been a motor train. For that reason, I added a further motor train specific siding in the fiddle yard to accommodate this service, and thereby reducing the amount of space in the Brighton side fiddle yards that would need to be occupied by rakes. Therefore, I amended one of the roads on the Brighton side to be accessible also from the South-Western lines, increasing the number of rakes that could access the layout from the South-Western lines. This, in turn, allows me to add a relief train using Maunsell 6-set no. 329 plus one PLV, which is the rake that comes with the Hornby "Southern Suburban 1938" train packs when both are combined. Research suggests that this rake was set aside for "special traffic", of which holiday reliefs would very likely be included, and these services would have been important on Saturdays in the summer in south coast towns. Also, T9 no. 312, which also comes in the set, is a very useful locomotive for the layout, as photographs from the 1930s show this particular locomotive hauling London Bridge to Portsmouth trains. The inclusion of relief services will also give a reason to use an S15 class, which no doubt would have hauled such services from time to time. Here is the amended track plan, showing added motor train siding (no. 23) and additional crossover into the Brighton side fiddle yard:

I approve of terriers.

I have checked whether the appearance of the colour is caused by the lighting conditions, but these photographs taken in different lighting conditions suggest a similar lightness: Dapol class 50 by James Petts, on Flickr 50036 by James Petts, on Flickr 50033 by James Petts, on Flickr 50040 by James Petts, on Flickr Dapol Class 50 by James Petts, on Flickr

One does not solder to the inside of the socket itself. That would not be possible. Rather, one solders to somewhere on the locomotive's PCB with a trace to the right pins on the socket. This needs to be confirmed using a multimeter. It is important to get this right, or else the decoder might be irreversibly damaged. If soldered to well chosen places, there will be no problem with plugging or unplugging the decoder. I should note that I have not actually confirmed that the Dapol A1/A1x models do not have solder pads for stay-alives; I know that the N gauge class 50s do not, and these also may not, but this should be confirmed by somebody with access to a model. It should in principle be straightforward to include these solder pads on the PCB, and not having them is an unfortunate omission, as it then makes fitting a stay-alive charging circuit more difficult than it needs to be, and a stay-alive system will be very useful for an 0-6-0 locomotive.

That is an impressive garter.

The manual does not, unless I have missed it, in which case I should be grateful if you could point me to the page in question as this would make connecting stay-alives easier for me in future, indicate solder points for connecting the SACC16 charging circuit to the Zimo MX618N18 Next18 decoder, although does do this for a wide variety of other decoders, including 6 pin models such as the MX617 series. In principle, one needs to connect the SACC16 between ground and common positive; on most Zimo chips, there are solder pads for both of these, but not on the MX618N18 chips. The idea is, I believe, that there should be solder pads on the locomotive's PCB, but this is in the manufacturer's discretion, and Dapol unfortunately do not seem to provide these for their Next18 fitted models. Thus, one needs to find traces on the circuit board that correspond to these terminals and solder to these. Ideally, one would do it from the locomotive's own PCB, but it can also be done from the decoder itself. I found a way of doing this from the locomotive's PCB on the N gauge class 50s, but only after I had connected a few with one of the wires soldered to the decoder itself.

I do not have one of these, although I am considering getting one. Fitting a stay alive charging circuit and capacitors to a locomotive with a Next18 decoder requires some very careful soldering. I have done it with Dapol N gauge class 50s and a Zimo Next18 decoder and Zimo charging circuit, but I needed to use a multimeter to find precisely where to solder for the ground and common positive and then solder very carefully to surface mount components. However, this was successful and I have fitted quite a few N gauge class 50s with stay alives in this way. The Dapol terrier has a built-in speaker. For those who do not wish to use this speaker for sound, this could be a good place to fit the charging circuit and capacitors (use at least 5x 470uF 16v surface mount tantalums with the Zimo SACC16). Having seen the size of the speaker on one of Jennifer Kirk's videos, this volume should be sufficient for a Zimo SACC16 and circa 5x 16v 470uF surface mount tantalum capacitors if I have parsed the image correctly. If using sound, I am not sure whether there is sufficient space. Having seen videos of the inside of the Hornby terrier, which uses a 6 pin chip, that appears to have no more space than the Dapol version, and even less if one is not removing the speaker.

The roof would appear to me to be the right colour: here is a photograph of 50040 in 1989: berks - 50040 arrives reading 3-1989 JL by John Law, on Flickr here is one with the roof cleaner: 50040 London Paddington by Adrian Hancock, on Flickr and here is a photograph of my model of it: Dapol Class 50 by James Petts, on Flickr The lightness looks quite similar to me.

009 Micro Modeller is correct - the idea was two separate layouts sharing a baseboard rather than a single location with both narrow gauge and underground. A waterworks railway is an interesting idea, but I do not think that this fits with any of the proposed plans at present. I am currently minded to build the version of the Underground layout that leaves no room for a narrow gauge, so this may become academic; but I am not at a stage where I need to start building anything yet, so that may change. I am focussing more or less exclusively on the modern N gauge layout this year, and have progressed significantly with that, so these 1:76 projects are currently in abeyance.

Gosh, poor chap. That does not sound fun. Get well soon!

Splendid to see good use being made of the picture that I took at the NRM back in 2011 - it is always lovely to see people making use of Creative Commons licensed pictures. Also, there's much to be said for a nice LBSCR 4 wheeler.

Happy cakeday for to-morrow! May there be much baked becandled merriment.

Aha, I have been looking forward to the revised NSE liveried variants. As to 50046 and 50033, I am already somewhat ahead of Dapol: Dapol Class 50 by James Petts, on Flickr Dapol Class 50 by James Petts, on Flickr (albeit my 50033 is in late 1980s condition, whereas the new Dapol model will be as preserved).

This is definitely interesting - it is good to see more London Underground availability. One odd thing that I notice is that, despite having working interior lights, the end lights seem to be painted on and non-working. I imagine that users can, with effort, add their own, however.