justin1985

I'm taking a bit of a break from working on the track work on Long Melford (long hours working in the garage seem less appealing now the weather is better), but I've been making a start on planning out the signal box, which will be one of only two "railway" buildings on the layout. Amongst the notes that came with the layout was a rough sketch of Long Melford Junction Signal box, with key dimensions worked out apparently from the known height of the junction signal - I spent quite a while trying to translate this into CAD, before remembering that I had picked up a copy of McLean's 'Pictorial Record of LNER Constituent Signalling' from a second hand shop ages ago (a real page turner ...). Quick check, and it reproduces what seems to be an original contractors' drawing of a GER standard signal box that is extremely similar! 

The only significant differences are that Long Melford was a bit shorter (fewer layers of tongue & groove above the locking room windows) and, like Sudbury Goods just down the line, it had a curious "lean to" extension on the northern side - presumably for better sighting of the junction? 

"Sudbury Goods Junc Looking South" - Tilt Cab Man on Flickr

I've been working at building up blocks of the standard elements like the windows and framing, which should make it simple to build up the elevations as arrays/using blocks. I'm hoping to be able to laser cut the windows and laser engrave the flush tongue and groove/non-corner framework. Thinking of using Paxolin/SRBP - which seems to be the same thing that many continental modellers have success laser cutting as "hartpapier", and similar to the "Polybak" material Americans use as "laserboard". 

The thing I'm least sure about is building a structure with some actual strength to it! The signal box will be close to the front edge of the layout, between viewer and track, and I can imagine it being very vulnerable to swipes and knocks. It is primarily facing away from the viewer, but with the 'extension' having wrap around windows (including one facing the 'back'), it will be quite easy to see in. That also means the rear wall is literally the only solid element. I'm wondering about building the corner framework from 1mmx1mm brass section, and aralditing the panels between them. Does that sound sensible?

Also wondering about what glazing material to use. Ideally, I figure it should be something that adds a bit of strength itself - like perspex / cast acrylic. But the thinnest clear perspex I seem to be able to find is 1mm, which seems much too thick. Looking for 0.5mm clear plastic sheet, most of the stuff available seems to be PET-G (i.e. plastic bottle material) which seems to be both very flexible, and tends to have a blue-ish tint. Any suggestions would be appreciated!

Justin

20 minutes ago, 2mm Andy said:

 

Graham Eason (the man behind this company) is a 2mm Scale Association member (and a fellow NEAG member) but I've not heard anything from him since the start of the first lockdown last March. Have you tried emailing him at the address on the cached website? IIRC the website/company name has changed a couple of times over the years (Vintage Transport in Miniature, etc.).

 

Andy

Thanks for the background, Andy. I did drop a message to the address quoted there today, but as the page is stamped last updated Feb 2020, I thought it might be worth asking if there might have been any news I might have missed.

Actually, now I've noticed there is a Peedie Models credit at the bottom of the cached Vintage page, and if Peedie themselves recently got a new website (looks very flash now!) that might explain the website disappearing, if it had been hosted as part of their old one ... maybe.

I'd been meaning to get around to ordering some of the resin kits sold by "Vintage Miniature Models" - there was a really nice 3D printed traction engine in N that they had produced in collaboration with Peedie Models, as well as some recently launched sailing barges, including a very promising looking kit for a steel Thames barge, which featured in the N Gauge Society Journal quite recently. 

However their website now just says "Page not found". The traction engine doesn't seem to feature on the Peedie Models new website either (I seem to remember it was on their old one, and it is used as a category image!). Google does seem to have cached the mobile version of their website: https://webcache.googleusercontent.com/search?q=cache:7oqs1Xx9VF0J:https://www.vintageminiaturemodels.co.uk/road-vehicles-bt-models/4592512754+&cd=13&hl=en&ct=clnk&gl=uk&client=safari 

Does anyone know if something has happened to this company/range? 

Has anyone built the Gramodels "2mm/N" kit for a Rectank wagon? I don't recall having seen one in the flesh, so I'm curious how it looks, and how it scales for 2mm?

Or has there ever been another kit for the Rectank? The long skinny structure looks like it might really be better suited to etching than resin or 3D printing, perhaps?

I'd love to depict this little scene of the Romney Hythe and Dymchurch's "Green Goddess" being dispatched from Paxman's at Hythe (Colchester). I think a T scale (1:450) A1 would be just about right! 

On a similar note, can anyone comment on the Gramodels "Whippet" tank (late WW1) ? I'm wondering if it would make a good load for the new Farish WD Parrot wagons (rather than using the identically sheeted Mk1 tanks supplied with some other Farish models - which @CF MRC refers to as dog turds ...)

Resurrecting this thread with a query on types of printer ink.

I normally model buildings in plasticard, but I downloaded a Scalescenes bridge for a quick project. My home printer is actually an HP colour laser - the prints look great, but on fold lines I've noticed the printed surface tends to crease and crack.

I guess that makes sense as toner is basically fused on as a surface layer, rather than soaking in like an ink. For now I've tried to touch in the creased white spots with watercolour pencils - which seems to have worked quite well (touched in on left hand buttress, creased on right)

I will be needing to make quite a lot of brick buildings soon as part of bigger projects, and frankly printed brick does look an awful lot better than what I can do with textured plasticard (I think the reverse is true with stone).

So I'm beginning to wonder if it might be worth buying a cheap inkjet printer if it might avoid this "cracking"? i.e. assuming ink that soaks in rather than forming a "crust" like toner does?

Could anyone share experiences of inkjet colour fastness of different brands? Not much seems to be written about this aspect, but I got a sense Epson have the advantage of pigment rather than dye inks? But on the other hand they tend to clog more? (Which might be an issue it's only for occasional modelling use)

Cheers

Justin

Combined with the bordering on nasty comments in response to queries about whether lapsed members would have their orders fulfilled on the NGS Members Email Group , this does start to look a little bad!

Thanks so much Nigel, that did indeed do the trick! The default for bit 1 of CV29 was indeed 0, according to the leaflet. I read back the rest of the CV using JMRI then used your calculator on the 2mm website to add in bit 1 = 1, and hey-presto, the light works flawlessly and also now responds to F0 rather than just working with movement. 

I would never have thought to trace an issue with the lights to speed step settings!

Incidentally, JMRI does read back the decoder type as "Fleischmann Twin Mode", and now I know what to search for, Fleischmann do briefly mention this on one English technical page: https://www.fleischmann.de/en/service/faq/digitaltechnik/index.html 

Quote

When the decoder, no matter if it is a twin-decoder or a DCC decoder, is internally set on 14 speed levels, but is, due to a setting in the TWIN CENTER or in any other DCC control device, controlled with 28 or 128 speed levels, the light is switched on or off with each 2nd speed level in the TWIN CENTER.

Turning the light on and off with every other speed level makes total sense now. It seems like "Twin" refers to also working with an alternative Motorola based standard?

Anyway, it is now a beautifully smooth running and sure footed little shunter - thanks again!

I picked up a Fleischmann N Gauge V60 shunter, which was factory hard-wire DCC decoder fitted, as a non-runner in the last few days. A quick play revealed the running issue was simply the wheels having been covered in oil - with the wheels and contacts cleaned up it runs beautifully now. However I've noticed  the lighting works intermittently, and only ever when the loco is in motion. Sometimes the loco will move off very smoothy, up to a fair speed, and the lights won't come on at all, or flick off when running, but stop and start, and they come on, so it doesn't seem like a pickup issue.

I had wondered if the lights being intermittent was caused by one of the brown wires (from decoder to rear lighting board - another brown wire returns to front lighting board) having been pinched/damaged, and replaced it, which I was hopeful had improved it, but after re-assembly it was still intermittent. (unless it got pinched again when screwing the body back on - although I was careful to route the wires inside their guide pips).

The instruction sheet is minimal, with only a handful of CVs documented for inertia and factory reset (tried this, no help) etc., and extra slip of paper lists a series of CVs, with a note only in German explaining something to the effect that lights will prototypically be on for both sides for shunting with the default value of 3 for CVs of 33 and 34, but with those set to 1 and 2 respectively the lights will change with direction of travel, which at least solves one mystery! I guess this setting also explains why the lights don't respond to any function keys. 

Any ideas what the other CVs listed might do? Is this likely to actually be a custom Fleischmann decoder, or perhaps they re-badge decoders from manufacturers who might be better documented?

Any thoughts on where to focus in fault-finding for the intermittent lights?

I've just noticed that the Farish War Department Parrot tank carrier flat wagons have been released. It doesn't look like they've been mentioned on here since the original announcement here: 

As the original announcement explained, these were built for the War Department during WW1 to carry tanks, then sold to the LNWR, MR, and CR, becoming part of the LMS fleet. 

There are some interesting photos of the Parrot in use in WW1 over on the thread for the Bachmann OO wagon: 

This made me think though - the Farish "landship train" pack of a few years ago came with several resin sheeted Mk1 tanks which sat on green WD livery Bogie Bolster (30t Bogie Bolster C?) wagons. I had imagined that the WD Bogie Bolster was a complete fiction, but according to this site some GWR Macaws (at least some family lineage to what became the Bogie Bolster C, I think?) were reinforced and adapted to carry tanks: 

http://www.railalbum.co.uk/railway-wagons/military/ww1-gwr-macaw.htm

Seems like Farish missed a trick not including the resin tank again with the WD livery Parrots! 

Has anyone got one in the flesh yet, or any ideas for other loads for them?

Justin

Some progress - with aid of a bendy stick! Alignment doesn't seem 100%, but hopefully not too far off.

The crossover dead centre seems to have had an awful lot of rubbing out and redrawing on the plan drawn onto board - bendy stick is very helpful to decipher which is the correct line! (Although of course the button gauge is exaggerating the bulge in the middle as the ply strip is otherwise inboard of the rail - representing check gauge, I guess)

Thanks for the reassurance, chaps! The opposite handed-ness to the curve leading to an S curve seemed like a likely explanation, but worried that was wishful thinking on my part!

I had been aiming for completing the branch through line just to "get one route completed" - but Don's advice to work on the two acute crossings before anything else now does sound very sensible.

Justin

After several months of DIY and garden projects, I've come back to this layout. Some progress over the last few evenings, although I've hit something I'm not quite happy with ...

Having started with the turnout vees approaching each side of the branch diamond crossing, I then filled in the little gap to make the start of the K. Then I worked back and fitted the curved switch to the first turnout (paving the way to gauge the other stock rail, which will feed in to the diamond as well).

However, keeping the turnout vee crossing itself straight, using the angles worked out before from Bills drawing/notes on the board, it didn't seem to align particularly smoothly with the switches. I've ended up with a bit of an S curve type situation where the blades curve normally but then slightly curve back to meet the straight crossing.

Looking from directly above, this siding track seems like it necessarily has to have a slight S curve to it to fit within the available space. Doesn't it?

Or is it a case of having to start again with the crossings on this turnout to different angles? 

On 06/05/2021 at 00:14, CF MRC said:

Bloodstock on CF will soon be improving...

 

Can you reveal who did the design/printing for these, Tim?

I seem to remember hearing at Alan at ModelU was of the opinion "never work with children or animals" ... 

A bit of Googling suggests there's quite a market for photogrammetry of horses though! https://3dprint.com/137010/my-horse-in-3d/ 

I got a copy of this from Titfield the other day (great service, as always!)

I found it interesting, but I couldn't help thinking it was a bit of a missed opportunity to do something broader. Reflective is a good description of it - nothing wrong with that - but it's definitely a personal account. The vast majority of the examples are from the Welsh borders region, for example.

I guess I was hoping for something closer to the lines of Brunskill's "Handbook Vernacular Architecture" but for the kinds of bits and bobs that "Trifles" looks at - e.g. I'd like to have seen a full chronology of phone boxes and post boxes, rather than just one or two interesting examples. (there is a full chronology of petrol pumps in photos, but that seems almost incidental in that all could be found around the Marches). 

Scope for a future project for someone, perhaps?

On 29/03/2021 at 09:42, richbrummitt said:

There are actually some interesting articles on the Chitubox website that I found much more useful than a lot of the forum posts from the 3D printing communities that came up in my search results. 

These are definitely incredibly useful articles! Much better than the confused and repeated advice on Facebook etc forums that seem to be the main communities of resin printer users.

But some of the phrasing is ... spectacular ...

Quote

Uneven quantity of 3D printing material and tumultuous peeling force among layers result in variable shrinkage which is the main cause of misrun and irregularity.

Couldn't have put it better myself!

Tumultuous is a word that I feel is not used anywhere enough in describing finescale modelling!

(from https://www.chitubox.com/en/article/support/howto/chitubox-free/editing/25 )

1 hour ago, Crosland said:

And then the problems started. I want to print what is basically a flat plate about 2.5x10cm and 2mm thick. It has a few holes and some embossed text. I am trying to print it vertically (long edge parallel to the build plate) in order to maximise the number that can be printed at once. The first attempt (with mostly chitubox automatic supports) resulted in only the supports being printed. The remainder was just a blob stuck to the FEP.

If it really is a flat plate, and you don't mind a slight "elephant foot" effect on the sides (which could be sanded off) then simply print it completely flat and flush to the built plate! A razor blade type scraper should be able to get it free once printed. 

If you do need a detailed rear face etc then raising it with supports will be necessary, but the trick will be to avoid dramatic changes in layer surface area, or if that can't be avoided, design supports to compensate for that. The "leading edge" always needs as many supports as you can possibly squeeze in, for example. If more "weight" is passing through some supports than others, that's when you'll get a support failure.

Also, if you're getting not even the supports left on the build plate, then it's worth checking the plate is properly levelled. It's easy to knock out the alignment of the plate from its bracket when scraping off a print (at least it is on the photon - imagine very similar)

J

That looks great Adam!

On 26/03/2021 at 13:03, A. Bastow said:

Things I would do differently; not sure that the roof needs to be separate so in future I would probably revisit this so that the floor is separate from the body shell instead. 

Assuming you're using a resin printer like a Photon or similar, printing without roof or floor is probably actually really helping!

If you include both vertical and horizontal surfaces, however you orient the print you'll get a dramatic change in surface area, which will cause suction issues and kinds of witness marks at that layer in the print. Your current design with a separate roof probably helps avoid that nicely?

J

20 hours ago, queensquare said:

The tie bar on a point has to be physically thrown by something, doesn't matter what, so why not just use that movement to work a microswitch at the same time. All this frog juicer stuff just seems to be a very complicated way of solving a very simple problem. Somebody once told me that the advantage of them is that you can run through a point that set against you without it shorting - presumably the train then falls off the track so I can't see much advantage.

jerry, a great believer in the KISS principle.

But that depends on what "complicated" or "simple" are to you, Jerry!

For me (I'm in the minority amongst 2mm modellers, I know) anything mechanical is "complicated". Cranks, cams, working out throw distances, how much slack or flex in a mechanical linkage, are BAFFLING to me! "Just" use a microswitch? I've wasted HOURS trying to get microswitches aligned to reliably "click" from the tiny throw of a 2mm turnout, while not getting in the way of other things, or stopping the tiebar from moving freely.

But, for me, electronic systems are "simple". Something like a Frog Juicer might be complicated circuitry inside, but if it is a "black box" that can be plugged in and trusted to do its job, then that is "simple", for me. I might not know the precise details of what goes on in the "black box", but if it just does it, then its "simple"!

I totally get it that a (commercial) frog juicer is a "dirty" solution to a problem that can be solved more elegantly in engineering terms, but if it does the job, then surely it is a horses for courses question? Dare I wonder if this might even be a generational difference? Or at least a difference between those who had a proper education in engineering type issues at school, and those (like me) who didn't?

19 hours ago, Donw said:

The MERG circuits for frog juicers include two which are innovative.  Julian Coles designed one that rather than look for an excess current looks instead for a colapse of the voltage across the track when a short occurs. This enables a quicker response. The other by Tim Coombs leaves the frog disconnected and uses a low voltage to spot a connection from the Frog to either stock rail and then connects the Frog to that stock rail should both stock rails get connected to the frog in perhaps a derailment the frog is left disconnected. This circuit can be used with DC.

Boards are usually available from other members or you can order them yourself.

Hi Don, could you clarify how Julian's circuit is available? I don't remember seeing it in the kit locker when I was briefly a member last year. 

One of the things I couldn't get my head around was why MERG kits were only available as kits. All the electronics tinkerers at my local Makerspace seem to go straight from breadboard to ordering ready made surface mount versions of their designs from Chinese "Shapeways" type services (even for one off or small quantities), which seemed cheaper than buying the equivalent through hole components from Farnell etc, let alone getting custom PCBs etched on their own!

J

Today's progress was gluing down the last few sleepers, and making up a 4 degree vee for the furthest turnout on the Bury St Edmunds branch. Making such a sharp vee was ... fun ... it is incredibly fragile!

I've hit a bit of a query though - Bill had marked the 4 degree vee with a few different marks - a biro mark one sleeper right of the 4, and then a pencil mark another sleeper further right. But actually the angle between the diamond and the vee doesn't fit as 4 degrees on either of those sleepers - but it does if the vee moves an extra sleeper left - which is where it is posed in this picture:

So I have a choice of putting the vee further left than Bills mark (i.e. where it is posed in the photo), or making a less sharp vee (about 5 degree) to sit on the sleeper marked with the biro mark. I can see advantages to both - the further left position gives more space for the diamond; the further right position allows a slightly less sharp (and vulnerable) vee. I'm leaning towards the latter - is there anything else I should take into account? 

The "straight" stock rail for this turnout will be first to get soldered anyway - but then I need to work out where the check chairs need to go ... 

J 

On the trap point question, it occurred to me to take look on the NLS collection of OS maps when I was looking at that for something else today. I'd previously been mainly referring to a paper copy that was amongst Bill's notes, which the notes on the back reveal to be a 1927 revision.

I can't believe I hadn't noticed this change before ...

The NLS 1902 survey / 1904 publication map actually shows quite a different arrangement with both sidings having full catch points with stub sidings (in the case of the Cambridge line, more of a full headshunt). More significantly the Cambridge line has a much simpler arrangement with two back to back plain turnouts of much sharper radius.

The Bury line is also only connected to the up platform line, but its branch siding is directly accessible from both platform lines! The branch siding perhaps begins to make a little more sense in this layout: if it is accessible from both platform roads when the branch itself isn't, then it makes a bit more sense as a way of exchanging traffic between the lines.

Reproduced with the permission of the National Library of Scotland. Full map: https://maps.nls.uk/geo/explore/#zoom=19.344505969847457&lat=52.06739&lon=0.71208&layers=168&b=1

The 1927 map, which Bill's planning reflects with some degree of compression:

The diagram on Signalbox.org, attributed to 1912 (but a 1982 re-drawing) shows an intermediate state with the additional diamonds in place on both branches, but the stub siding still in place on the Bury branch siding: https://signalbox.org/~SBdiagram.php?id= 1064 

Short of digging through the relevant GER ledgers to references to expenditure on track layout changes when the Essex Record Office reopens, I don't think it will be possible to pin down the date of the changes any more precisely. No one has ever published a "proper" (Oakwood or Wild Swan type) history of this line - only ever picture books and pamphlets! 

The era I want to focus on modelling is what I'd define as "Georgian" (George V) - c.1910-1935 - so I think the 1912 signal box diagram is my best guide, really.

Upshot of all this? No actual change to my plans! The only change that the earlier track plan would seem to need would be the Bury line stub siding catch point, and I don't have room for that, so single catch points it is. Still, it was interesting to me, and I think it does help to make a little more sense of the sidings and how they might have operated. 

Justin

Another vote for Finetrax if you'll be using exclusively diesel / electric stock and/or commercial N models. I model in 2mm for my pre-grouping steam models, which would usually have to be kit built anyway, and where the wheels and connecting rods are more "obvious", but if I ever get around to building a layout for the sector-era BR models that I've collected, it will be Finetrax/hand-built N. Why go to all the effort and expense of changing all those wheels when they're mainly hidden behind bogie frames?

The Finetrax range is growing, and the straight turnouts etc can be curved using the techniques that Mick Simpson wrote up for the pegged version of 2mm Easitrac turnout kits (which are essentially identical to the Finetrax ones, apart from the gauge). Indeed, you can get beautiful kits for concrete turnouts in Finetrax that aren't available in 2mm!

As well as Blueball Summit, TomE's Ropley comes to mind as an excellent example of hand-built N track using 2mm components. 

On radius I would also echo the comments on couplings being key. If you could rely on everything having effective close coupling mechanisms then you'd probably be fine, but alas even some brand new models like the Farish Mk2 air-cons don't have close coupling, so could prove to be a problem.

11 hours ago, Donw said:

I would think a single blade in the outside rail before the diamond would do the job nicely and throw the vehicle away from the diamond, Any movement in and out of the sidings would be at low speeds.  Unless you see a photo suggesting it was a double bladed trap.

That sounds very do-able! I'll definitely plan on doing this.

11 hours ago, Donw said:

You could consider a megapoint servo controller currently £63 for 12 servo controller, servos about £2 or so .  Unless we can tempt you to join MERG

I did join MERG for one year, and built some of the ServoSet controllers (I concluded an Arduino was a more satisfactory way of doing the same thing) and other little boards (lamp twinkler etc.). However the magazine invariably scared the bejesus out of me, and I concluded that it was very much not for me. I'm happy to tinker with things, but for me I think that is at the level of combining ready made things like Arduino, and tweaking/adapting code. 

54 minutes ago, Nigelcliffe said:

I think the first question is "what sort of control is wanted".   ie. if wanting a screen interface, then JMRI will be fine, the locking rules are relatively simple.     However, if wanting a physical switch interface, then it depends and low-tech may give all that's needed at lower cost and lower hassle. 

Very good question Nigel - I have been giving it some thought, but not really at a firm conclusion yet.

I used to think DCC point control sounded like an unnecessary complication, but then I had always been thinking of "cameo" type layouts! I think these are my real requirements:

• Operate from the front (at home)
• AND Operate from the rear (at exhibitions - IF it does)
• One person operation must be possible (even if two - layout + train table - is preferable) - which by default means that person needs to be at the train table end a lot of the time - but hopefully not all of the time

As far as I can think it through, that means EITHER some kind of complex duplication of physical switches and loads of wiring; OR a plug in control panel that can be plugged in to more than one place OR on a long enough cable to be moved. That seems to push towards some kind of bus arrangement, and seeing as I'm convinced on DCC for train control, it makes sense for that to be the DCC bus ... doesn't it ?

Thinking out loud, absolute ideal would be a physical switch panel of some kind acting as an additional "throttle" for a DCC system, which could be plugged in with a LocoNet type plug at several potential locations. I had imagined this is what the DCC Concepts levers might have done, but it looks like they're actually just fancy analogue switches?

Justin

EDIT - just realised the DCC Concepts Alpha Encoder unit does what I'd been describing!

https://www.dccconcepts.com/product/cobalt-alpha-main-unit/

Many thanks for all these thoughts - really appreciated!

4 hours ago, Nigelcliffe said:

( How the turnouts are interlocked depends on what form of switches and turnout mechanism is to be used.  There are lots of solutions, but  how depends in preferences for types of switch/lever, type of panel, choice of motors, etc.. ).  

The removable trackbed means space is actually quite limited. Bill had started installing PECO Smartswitch Servos on other parts of the layout, which I got rid of, but I had been thinking of some variation of servo with under baseboard tiebar and microswitches. Perhaps run from something like an ESU Switchpilot Servo. I imagine it would be possible to implement the interlocking inside JMRI, but I can't help thinking something at hardware level would be preferable?

3 hours ago, Donw said:

If I still remember my maths tan 4deg or sin 4deg are both about 0.07 or 7 in 100  1:14.3 . Now a D14 is a common sort of size for passenger lines where you do not want a severe speed restriction. There was one in the station throat of Selby. I know that because we used to meet in Pete's shed on the Isle of Wight and he had built a 7mm one to match the actual track plan. You will find plenty of others if you look.  It will need a bit of care to make but will look superb in 2mm. A D blade is planed at 1:48  so about an inch of taper for association rail.

Thanks - I'd been meaning to look up how to translate between angle in degrees and ratios! 

8 hours ago, Izzy said:

 

"I think Bill's only simplification was to remove the headshunt / trap at the end of each siding. "

 

Given the relative complexity of the rest of the track that’s a shame. I think I’d want to include them, even if they were just simple single blade cosmetic jobs for the overall look seeing as how the sidings face on-coming passenger carrying track.

Agreed it needs something. I can't actually find a pre group photo of this end of the station (plenty of the yard end). Maps don't suggest a "proper" stub with bufferstops though (and the cutout trackbed doesn't have enough space!), so hopefully a cosmetic trap point will be enough. I imagine a "double trap"(?) would be most appropriate - i.e. a pair of blades only?

Well I think we're concluded they are not true wyes - the branch track to the right is the less used route, and sure enough it is a sharper divergence. Nonetheless I've out joggles into both rails - just because it seems difficult to get a 2mm point blade reliably held sufficiently flush against the stock rail without them - at least in my limited experience. 

1 hour ago, Donw said:

The the  two sidings do seem odd. I notice that both are trailing connections quite why they have to cross via diamonds is puzzling. The appear to be rather short for lay byes. Now if this had been in hilly country  I might have throught about them holding a banker but even then why two one would seem sufficient. The other throught was it being something to do with interlocking but as they cross the other route on a diamond one would imagine that would be interlocked too. 

Is this based on a real location or something Bill came up with?

 

It is a real location! Signalbox.org shows the track layout essentially as modelled as the 1912 layout: https://signalbox.org/~SBdiagram.php?id= 1064 The bottom siding is labelled as refuge siding, and the top one as branch siding. As you say, both are only accessible as trailing connections from the up platform line - although that is the line with the connection to the yard (extensive, but not likely to be modelled). I think Bill's only simplification was to remove the headshunt / trap at the end of each siding. 

A modeller who has built the station in OO has posted a c.1955-61 signalling diagram (his own?) that shows the Bury branch (top) siding removed, but the lower one labelled "Haverhill Dip Siding" - which might be a clue? (not sure to what, but you mentioned hilly?)

https://www.facebook.com/longmelfordjunction/photos/pcb.593063704737635/593058104738195/

I've traced a track plan from one of Bill's sketches - I've tried to reflect some of the changes as built (such as the simplified hidden return loop and less divergence on the branch siding) - but I think the curves are pretty much just representative. Therefore it is far from a scale plan, but definitely useful as a schematic. 

I've started sketching in gaps and colour coding for feeds - definitely a work in progress (haven't even tackled the lower half yet) - but hopefully looking along the right lines?

The crossing angles are those Bill had written onto the board. 4 deg does sound extremely shallow, but relative to the others, does perhaps look right?