aleopardstail

incidentally, in the unlikely event anyone wants this the .stl file is on thingiverse, likely do one that can take microswitches at some point but for now don't actually need it

benefit of owning a 3d printer, Ender 3 V2... simple custom servo mount, servo clips in and out if needed, held securely, frame then screwed in place, actuator wire through a hole that in theory is there but in practice needs drilling (as I set the thing to low quality/fast print) crude but it works installed, takes about an hour to print each one, likely one more tonight (need some extension leads for the servos to arrive anyway to hook them all up) but installs easily enough

Analogue controller now installed complete overkill with a Nano for just driving the L298N board but the clones are cheap so not too bothered, not made the handset module as yet but that can wait. Then cut the remaining 7 servo mounts, the usual 15mm Aluminium C channel, same stuff I've used before but a new length.. someone has decided a slight cost saving can be made by fractionally narrowing the walls of the channels, as a result the servos are no longer a friction fit - this is exceedingly irritating to say the least. experimenting with a 3d print design for a basic servo mount I hope will resolve this, if it works will print a few more.. Grrr meanwhile the underside of Dingleberry is gaining more wires by the day you will be please to know that no wild "cable managements" were harmed during the manufacturing of this mess, the most shocking thing is that it actually works, quite a bit more to go in yet

already have the driver, going for computer and not panel control, I could also drive a relay reasonably easily but prefer the idea of the mechanical interlock (goal is one switch for a frog and another one or two for feedback on the position), issue is more mounting the switches to be operated reliably - first pass was screwing them to the sides of the channel but the servo horn doesn't touch them, may just have to make a custom mounting - I've got a 3d printer so shouldn't be too hard. was more wondering if there was "accepted wisdom" on solving this :)

actually in MERG, hadn't seen that, will have a look cheers :)

Hi, on Dingleberry I'm using the pretty common SG90 servos for point control, jammed in 15mm metal C channelling, and for switching points it works. I was wondering, is there an "accepted" way to piggy back some sort of auxiliary contact switch to this arrangement? don't exactly need it here but say for frog polarity switching? I did think of physically mounting a pair of small microswitches to either side of the channelling but was wondering is there a better way to do this? Obviously I can also use the driver MCU to control switches, or for like frog polarity maybe drive a small relay but it would be nice to have some reasonably self contained way to add a few switches. thoughts?

DC controller modified slightly, this will lurk under the baseboard eventually (though the relays are set so unpowered the layout will run in DCC mode so can run without it). In place of the 10k potentiometer & screwdriver control method the arduino driving it now responds via its serial port to some very basic commands. basically send "f 128" and the thing goes forwards at half speed and similar.. this will allow it to be driven by another arduino eventually via a small 5 pin connector that can sit in a handset, but for now it means can drive via the USB lead... point was to make it easier to oil the two old GF diesels and let them run for a while. discovered the Cl25 isn't happy round curves in one direction which suggests a dodgy connector, fine the other way though so a job for the future. forgot just how noisy old models can be.. the XPS under the track and the glue is limiting the track noise pretty well, just the wheels and the gear sounds from the ancient motors & gear trains.. I think that once the tracks are all done, points etc working correctly I may have to consider buying a newer locomotive of some sort just to see what the newer generation are like. by eck though the smell of running these brings back memories of when I was somewhat smaller and quite a bit lighter. youngest (21, on leave from the army, where the heck did those years go?) said "so presumably this is just mucking about and when do you start the bigger one?"

got bored installed the relays, wired up the L298N module and added some flying leads.. rubbish flash pic but can now drive DC locomotives, ok only one at a time as the whole layout is live but means can test/run in which was the goal. means the #25 (pictured) and #47 could be given a run, old Poole era models but I like them, oh the benefits of all wheel pickup over the even older Lima #86s I have which currently are the only ones DCC fitted.

Slight touch more progress, now the TMD is physically present and the programming track is there the switch gear to toggle it mentioned previously is needed. well today I wasn't doing much else so hence: a pair of DPDT relay boards built (and tested, they have a tick and everything). pretty basic driver circuit, the LED isn't overly needed as these will lurk under a board however nice to have and I have a heck of a lot of LEDs so why not? Next up, get these installed so the default no power to relay mode is "DCC track" & "programming track in running mode". this also then allows the DC running, meaning that controller needs setting up (L289N motor controller and an Arduino)..

new track sections wired in and tested, one exception the small siding towards the front of the new area is not connected yet - this will eventually be the DCC programming track and needs a pair of DPDP relays wiring up - the first will swap it from "track source" to "DCC programming source", the other will be swapping the "track source" from DCC to a DC feed (not yet connected itself) - this so I can use the whole layout to run a single DC locomotive when testing/running in prior to DCC conversion. nice when one of the ancient Lima Cl86 ran first time. also the TMD is wired slightly differently to the rest, its all one "block" as I'm not using automation or block detection on any of it, equally while the pointwork is all set track insufrog I have wired each siding to always be live and made sure in effect to feed all the points backwards, thus at no point in that area does electrical continuity depend on the switch blades and stock rails making contact.. something I seriously wish I had thought about for the rest of the layout as this seriously improves running, not easy to retrofit elsewhere either due to the way most of the points have insulated joiners directly on the common crossing outlet rails for section gaps - we live and learn.

final sections of track have been added.. to be honest if I had planned this better track laying would and probably should have been a single weekend's work.. and it was, just a very long weekend. TMS track now added, inspection pits installed, all the dropper wires soldered, buffer stops glued in (and the beams painted so I can find them on the brown carpet), holes for point servos drilled, the little plastic mask to avoid the entire 8mm holes being visible added etc. bit of cleaning to do, waiting just on some more 0.8mm piano wire and the rest of the servos go in (7 to add) and then thats the actual mechanics done. step after is very likely starting the "concrete pad" for the shed and likely a few bunded bits for a fuel point and off load point and a bit of roadway where is crosses rails in front of the shed - then onwards to painting..

Final few bits of track on order to allow the TMD to be installed, will take a while to throw in all the point servos etc but the track being down means I can start painting it, which will be nice. counted up the power feeds, for a small layout there are a lot - 32 currently, will be 40 eventually. Why so many? why so utterly over complicated? well part of the purpose is to experiment with automation so for each "block" we have at least two feeds, a main bit and an "end bit", the odd bi-directional section has two "end bits" and a middle, sidings typically have two, a main bit and an end bit etc. the TMD will be wired similar though no intention of automating that, but more there because at that point why not? waiting delivery of some header pins and the point controller will be on stripboard, recoded to change the pins used on the Pi Pico for a more sensible layout, but that now works so just needs building and fixing to the underside. 3d printer on order also, not just for the layout, but that means I can start on platforms, retaining walls and similar as well so hopefully some more interesting visual progress over the next few months

that framing looks industrial strength, can't see anything moving, and seriously nicely done for such a complex shape in such a small area. very nice indeed and definitely full marks for wood butchery, makes my "no accurate measurements were harmed, or even seen, during construction" look garbage 😀

Progress! Spent a bit of time with the servo point controller, was having an issue whereby it worked, but there was no feedback to JMRI as the points moved. Bit of a rewrite and the controller now pretends to be a 32 bit in & 32 bit out SUSIC module, 16 outputs driving the 16 servo slots (8 installed), and 32 inputs provide each way sensors for each of them in pairs - so I have "clear", "Throw" or "Inconsistent" while the blades move. took a bit of debugging of the hardware code to stop the thing stalling and jittering, mostly by adding some code so it only attempts to actually change turnouts that need it and doesn't try to set all of them each time one changes. that done and can now drive them all from JMRI which is nice next up is some hardware track fault diagnostics, one of the crossovers and the turnouts for the two long sidings are causing trouble, can run through the crossover in straight, but throw it and trains stall. there is power when checked with a meter so its not a blade contact but a train there stalls, not getting a short circuit cut out but something is causing a power drop. same happening with the siding turnouts, again fine clear but set for the diverging route and train stalls. All the others, installed identically are fine so taking a bit of head scratching. next job is to get the circuit moved from breadboard to stripboard and mounted under the baseboard out of the way, also have found enough servos to power up the remaining installed points which should allow trains to traverse all the laid track and then short of the yet to materialise TMD the layout to start being operated at a basic way. after that its a second Pi Pico to listen to some MERG occupancy sensors and to start the process of sensoring up the track.

I got slightly bored earlier, so decided to hook up the point servos, so far of the nine connected, eight work, the other fails to respond so a bit of debugging needed - its set in the right direction so not too bad.. Just means nothing can drive into the TMD, which since that doesn't yet actually exist is perhaps a good thing. I'm using normal SG90 servos and a PCA9685 control board, a clone of the Adafruit board, to drive them. easy to use and works fine. Thats installed on the underside of the layout with various extension cables to reach the servos, I'm hoping for a layout this small extensions will work fine, anything larger and would likely have a few controllers. This is in turn controlled by a microcontroller, it would have been easy to use an Arduino, however because I was experimenting I've used a Raspberry Pi Pico which makes adding a small screen and keypad quite easy without running out of memory, the controller itself is still on breadboard but it works the blue and yellow wires go to a 16 key touch pad controller, connected via I2C. the wires going off the bottom of the image go under the board to the controller board itself. this is running some custom code that drives the screen, it needs updating as there are two modes, here its in "manual" mode used to adjust the individual settings - the display shows its configuring servo #0, which will go from -25 degrees to +35 degrees from the mid point, and will on power up go to the mid point. it is set to take 1 second to go from one end to the other providing a nice visually appealing and not too noisy operation. ok actually changing the point position via this is cumbersome to say the least but it works, the breadboard also has an RS485 module and the Pi Pico is running a C/MRI driver so once that adaptor is added and wired up this will talk to JMRI to do the actual leg work - just needs a software update so it takes the commands to drive the 16 servos but also reports back all 16 positions using 32 channels (each servo getting 2 bits to show the two settings and also a third state for "in motion") all in happy, and the test train trundled round nicely here you can also see the touch keypad, its labelling isn't exactly ideal, it may get a paper overlay added

Blindsided today, literally jaw dropping. Joined MERG on Tuesday and ordered a couple of block detector kits to experiment with, expected them sometime next week, maybe week after.. They arrived today.. Amazing service assembled them, though without the header pins as I want all screw terminals and they have yet to arrive. The LED replaced with a pair of two pin headers so the LEDs can be mounted where I can see them to try it all out. worked first time, picked up some old coaches with 10k resistors on one axle perfectly as well (once I cleaned the coach wheels a bit) screw terminals due next week and then will hard install the first two boards, only wired up the two longer sidings initially to test things, next up will be adding the board outputs to a C/MRI enabled Arduino to provide some feedback. very happy bunny

*coughs awkwardly* its been a while, however this project is far from dead, more track has gone down, though still not all of it - the loco depot remains a paper project for now. naturally within a few minutes of the board being back on the desk for testing a squatter arrived.. so far the track that is down is actually wired up, the underside being a bit of a mess but for now it will do the light blue and orange blobs are current transformers, these are to be removed (some already have been) as I'm moving over the MERG occupancy detectors to save whats left of my sanity - yes I can make perfectly workable ones myself, indeed I have, but the MERG kits look a lot less trouble and for the price.. so yes.. the next job now that the track is powered and has been tested and it all works (after adding another feed and repairing too loose wires) is to fit the next eight point servos which I hope to start work on later tonight with the brackets being finished and to start installing tomorrow to test over the long weekend. will then get the C/MRI interface stuff re-attached and can actually play trains again.

bit more rolling stock arrived this morning, quartet of GF Mk3 coaches, here nicely demonstrating that the two longer sidings are not long enough for such. to be honest I really only wanted some Mk3 <i>now</i> for gauge clearance testing, at best the layout supports a 2+2 HST which to be fair looks to my eyes a bit silly anyway. still, they are quite nice have also etched boards for more IR sensors, will drill these later, track loop needs three more currently which once tested can see the outer loop fully secured down. also have some conductive paint so plan to try a 10k resistor retro-fit to a coach and see how the detection system handles it (and to try a comparator based current sensor in place of the current ADC versions)

Experimentificating! this blurred mess demonstrates two things, firstly my iPhone is really rubbish at close up photo work, and secondly that it is possible for me to create a small circuit board and hand solder 0603 surface mount LEDs more amazing was that is actually worked, this is a prototype for a 4 aspect colour light signal head. it has shown me a few things, mostly I need a better way of cutting the board for things this size, also that I need to find a slightly different way to actually wire it, may enlarge the upper side of the anode solder pad (top in the pics) so the wire can come out the side to wrap around to the rear of the board where they can be twisted (the wire is enamelled)and then fed down the inside of a narrow brass tube. the PCB is 0.4mm thick, in theory double sided but only using one side here (there is no way I could drill accurately enough to go through the centre of a pad) does however show that in theory at least I can make working colour light signals

both long sidings, and the adjacent track section between the two currently unused turnouts now have the IR proximity sensors installed, they are now all also fully glued down. need to print off, transfer, etch, drill, cut and assemble the next batch of proximity sensors - that will give me the four on what will be the outer loop of track (each of which will eventually have a signal associated with it) and the two for the sidings (so inbound trains know when to stop before going splut). can then start to consider the inner loop, which needs another five such sensors (four blocks but one of them is bi-directional so a sensor towards each end). currently these are just being tested but not actually hooked up - same as the current transducers for the inner blocks will be installed but not fully wired up. need to revise the point motor servos to include the position sensor switches (and to be centred a bit better), means cutting some more channelling and configuring to put the mounting screws into different positions which isn't hard to do. this is the "fiddling" stage but actually quite enjoyable, now the basic techniques are down things can be installed even if the actual operation is all a bit basic

slight delay, waiting for some bits. anyway, first sensor head was built and tested, it worked, so a second was built for a bit more testing in situ this time, looks thus the black painted circuit board is flat with the bottom of the track, wires down through the board, needs painting grey before the track gets actually stuck down properly. have to remember to cover them when painting the track and ballasting. hopefully this shows why I'm avoiding the ebay "IR position sensor" modules with the 5mm LED form factor devices, these should be invisible once ballasted and painted up. testing! 1.48v output without a train over the sensor head. 2.46v with a train. the head has a second ambient light sensor, it feeds into the analogue inputs on an Arduino (Uno currently but will be a Nano with three connected and the I2C pins clear), this means if the light is brighter I can use the ambient to provide a differential sensor spotting either the reflection or the shadow the vehicle casts. only mode to the coach is a bit of white paint under the coupler housing, it will "see" a train without that but the delta is much less so this is a bit more reliable. if this was OO I'd use a PCB sleeper with the sensor and emitter mounted directly to it now I know the sensor head works I can build & install others, the below board gubbins can come later, this has been the hold up for track laying

Sensor V3 prototype board ready for etching easy enough to change it and make another one, so I did, etch likely tomorrow (I hope), build next week, need more 0805 resistors. note the design now has the pull down resistors and the current limiter for the LED on board. Ambient sensor (two sets of squares along the bottom) is a bit further from the rails, the four circle are between sleepers outside the rails, sensor and emitter back adjacent in the same sleeper gap - will also be angling the emitter and sensor slightly to try to get a better reflection

Second version of the sensor head assembled today, this one moves the LED current limiting resistor and photodiode pull down resistor to the head itself, simplifying using it considerably, have also added a second phototransistor that sits adjacent to the track - this one to pick up an ambient light level. Idea being to use the different between them to detect a train. Still feeds into an arduino, results are mixed, this design puts the emitter and detector with a sleeper between them which I don't think is a good idea now. measured voltages back are lower (the sensor isn't getting enough of the light bouncing back) room ambient light and the sensor head (between sleepers) picks up 0.01v, basically its seeing nothing, the ambient sensor seeing 0.03v. move the white painted part of the train into view and as expected ambient doesn't really change, the sensor head is now seeing 0.11v so the delta between them changes nicely. the difference now is when the light is brighter, bringing a desk lamp near and ambient reads 0.13v, with the sensor reading 0.11v being the white bit of the train in and again the sensor sees the same 0.11v, hmmm.. however, as the shadow of the rest of the coach passes the sensor now drops to about 0.04v, so now the readings are the other way around but the variation between them is still enough to detect a train the first version couldn't cope with brighter ambient lighting as it would see a higher voltage and register a train. ok, more work needed on the sensor design to get the emitter and detector closer together in the same sleeper gap, I want the voltages back to where they were before up to about 1.5v, or maybe slightly higher, will also move the ambient sensor to be slightly further out to avoid some of the shadow falling over it. ironically the surface mount stuff may actually be a bit too small, as in shallow profile, could do with being nearer the top of the sleeper than the bottom - may be worth adapting the sleepers with a file to reduce the profile where the circuit board will end up. basic concept works though and once ballasted would be essentially invisible unless you knew where to look

Experimentation updates, parts arrived earlier to assemble the IR sensors, so the first prototype have been assembled and tested. here it is post assembly but before painting, the two 0805 SMD components in place (both from Kingbright, a KP-1202F3C 0805 infrared emitter LED and a KP-2012P3C 0805 infrared phototransistor - the two being designed to work together). they are meant to face each other to act as a photo-interrupter however here being used to project a cone and catch a reflection. initial testing showed light bleed between the two, so the head was painted (it was going to be painted anyway) this left just a small window exposed, when tested with an arduino and a couple of resistors (a 180 ohm for the LED and a 10k pull down on the phototransistor - limits current flow nicely and the extra response speed isn't needed here for a lower resistor), this works, could see a bit of paper up close. this is the basic test setup, the arduino is set to measure the analogue input for the sensor voltage, 0-5v, this to provide an idea of the range we will get - eventually this will go via a comparator and a reference voltage for each sensor, of which more below. initial test, the Arduino on board LED is set to illuminate at about 0.75v (found by experimentation). the 0.5v reading is essentially ambient light being picked up as the sensor is pointing directly upwards. sliding the coach along a little reading goes up as the coach underbody reflects a bit of light - the issue is the underbody is black, which absorbs the infrared light, so the underside of the coupler housing has been painted white, also tried a bit of the actual underframe, which while it can be seen is a bit too far to be easily distinguished from ambient light levels, I suspect it will work better in darkness. so it works, mostly, I was going to give each sensor a fixed reference voltage (well a potentiometer but designed to set & forget between 5v and 0v). I think I may consider a reference voltage that includes a second sensor for ambient light, hence we check for a threshold as a percentage of the ambient light level, not hard to do, just means having a few ambient sensors dotted about, then hooking the potentiometer to the output of them, will have to experiment. anyway, a bit more work with this one and will install it into the actual track to test it in more detail. Not too bothered if I end up putting a bit of white paint on the bottom of coupler housings to make it work reliably.

I did wonder about just having more blocks and working it that way I have to say - when I started out I wasn't too sure how well the block detection would work so thought of the split with both block and IR sensors, I think on a larger layout I'd just use the block ones