Gauge 1 Diesel Multiple Unit

This page last updated: 17 February 2019.

Having begun the construction of my front garden railway on a postage stamp (intended to reflect the Rhymney Valley line in south Wales from around 1964) at the start of 2018 I needed to resolve how it carries passengers.  The Rhymney Valley line was converted to diesel multiple units in around 1958 and, with the help of Noel on the WRRC forums, the immensely detailed www.railcar.co.uk, and my memories, I determined that I needed a class 116 DMU, gangwayed (i.e. converted to have a corridor running all the way through) in all blue livery except for yellow-painted ends.  However, no-one makes such a beast so I had to construct one myself.  I began work from these drawings:
Stuart Mackay of www.railcar.co.uk helped me with additional pictures and Chris Moxon of preserved.railcar.co.uk helped me get in contact with Alan Pitt of the Great Central Railway near Nottingham where Dave Watts, the co-owner of a class 116 DMU (one of only two that remain) very kindly allowed my to take detailed pictures and measurements.

For the modelling part, I joined the Gauge 1 forums to ask for advice.  David Halfpenny pointed me towards David Leech, a Canadian modeller who has a retirement business scratch-building coaches and has attempted to build a class 121 DMU using 3D printing.  David Leech has given me lots of great advice on the use of 3D printing, the use of bent aluminium sheet and many, many other aspects of Gauge 1 coach building.  I also purchased a copy of "Carriage Modelling Made Easy" by David Jenkinson which describes another method for making coaches.  I decided to 3D print the ends and initially I thought I would make the body of folded aluminium sheet but in the end it turned out that, for my purposes, 3D printing could work for the entire body.  Drive will be provided by a couple of Fosmotors on one bogie, the power and control for which David Leech believes can be hidden underneath the DMU.  Here's what I think I have to do:
Having never done this before lord knows how it is going go turn out.  But I have to start somewhere.

All the Blender/CNC files created during this project can be downloaded from Github. Note that I've chosen to use 1/32 scale to keep things simple, rather than the exact 30.48 factor implied by 10 mm to the foot.

Drawing

The dimensions of the vehicle are based on the drawings from www.railcar.co.uk enhanced by a visit to the Great Central Railway for a detailed photographing/measuring session.  One point of detail to note, which I got wrong first time around: the hinges of a door are always on the left as you face the door, hence the doors on either side open different ways with respect to the direction of travel of the DMU.

Motor
                Open Second Body
Cabin dimensions
Motor Open Second Brake body
Motor chassis
Trailer
                Composite body

Trailer chassis


3D Printing: The Cabin

I spent a month or so designing the 3D printed cabin. It took some considerable time to get my workflow correct in Blender, the de-facto free 3D design package (primarily intended for animation); it is arcane and complex but there is a lot of community help out there which makes it ultimately useable. Since Blender is intended for animation it is not so good at maintaining a "manifold" object, i.e. one which 3D printing software can understand and print; a real closed shape, without one or two-dimensional protuberances, which can very easily appear when creating complex meshes.  Anyway, here is the finished article and the final print from my Prusa 3D printer, along with many of the test prints, all printed in PLA, the default material for 3D printing work (which is long-term biodegradable).

Cabin in Blender
Cabin printed

The "optimal" resolution (balancing print time and quality) of my printer is only 0.15 mm and you can see this very clearly in the rivets, probably the part of the model which took longest to get right.

Rivets in Blender
Rivets printed

I tried increasing the resolution: here is the riveted area, photographed under a microscope, at 0.15 mm, 0.10 mm and 0.05 mm resolution, print times increasing from 3 hours to 12 hours:

Rivets at 0.15 mm
                resolution
Rivets at 0.10 mm
                resolution Rivets at 0.05 mm resolution

0.10 mm layer height seems a sensible compromise. And I need to align those edges along the x-axis to avoid the rastering effect.

Of course, I also need to start 3D printing in my target material, ASA (a UV-safe version of ABS).  I had originally expected to get the final parts printed on a more expensive, higher resolution, 3D printer, so I sent the front section to be printed on a few other 3D printers (all in ASA):
Professionally
                printed
Dremmel print
Printed on my
                printer (in ASA) for comparison
Professionally
                printed, detail Dremmel print,
                detail
My printer
                for comparison, detail

The Fortus 450MC prints are definitely much higher quality, the individual steps being more finely registered.  The Dremel prints have a higher definition around the window area but are otherwise pretty similar to those from my Prusa printer.  But how much all of this matters will depend a lot on how I finish the prints.  The suggested way to remove the "ribbing" of the 3D print is to use sandpaper, which would be really laborious if I'm going to make loads of 3D printed parts, so I thought I'd try my hand at acetone vapour smoothing.  Placing an ASA (or ABS) 3D print in a sealed box containing kitchen-towel soaked in acetone for a few hours effectively melts the surface of the print.  However, you lose definition along the way.  In a test, I found that if I left the print in the acetone long enough to smooth the ribbing (between 3 and 4 hours) then I also lost the detail around the windows and, even if I might prevent that with some form of acetone-resistant coating, the result still wasn't really good enough, too indiscriminate for such a small and detailed print:

Before acetone
After acetone Before acetone,
                rivet detail After acetone,
                rivet detail

So, looks like it's loads of sanding for me.  Here's a test piece after 15 minutes of sanding with sanding sticks, one coat of primer, another 15 minutes of sanding and another coat of primer.  Good enough; the secret is to be bold on the coarse grade (400 grit).  It has somewhat lost the rivets, and in the paint rather than the sanding:

Sanded finish

Having satisfied myself that this was the right approach, I started adding detail.  First, I added a slot to the Blender model of the cabin into which I could insert windows (cut from the 0.5 mm plastic sheet used for dolls house windows); these windows aren't going to be pushed out by mistake when handling the model.  They would have to be masked if I were painting post-assembly but I think I will be able to paint the individual sections before assembly and only then insert the windows.

Window slot
Window slot

Then I added a dashboard...

Dashboard Dashboard

...and some ridges to the cabin and cabin roof so that they would mate together somewhat...

Ridges in cabin
Ridges in cabin
                    roof
Ridges in action
Assembled cabin

..and finally, since such an intricate object tends to cause the slicing program to add random support structures which become difficult to break away, I manually added just the supports I needed and spent 18 hours printing out the final ASA object in 0.05 mm resolution.  Here it is, on the left just as printing completed and on the right the assembled article awaiting painting; the gap between the roof and the cab front is for a guttering strip which will be added later.

Final cabin
                      before finishing Cab completed


Bogies

I purchased from Tenmille an AG140W bogie kit and from Fosworks a pair of MOT100 nose-hung Fosmotors, powered from 12 V on 30 mm diameter wheels, plus a pair of plain-old MOT310 30 mm diameter wheels; I tried 34 mm wheels originally but the aspect ratio of the whole thing looked wrong and the buffers to adjacent cars ended up being a few millimetres too high.  Assembling the bogie with the un-driven wheels was fairly straightforward.  I filed off the "SR" emblem from the axle covers of the casting and left out the optional step across the middle (which is only required for brake vans).

Undriven bogie
                      assembled
The original

To fit the Fosmotors in the other bogie, first drill the 2.25 mm holes as directed in the bogie assembly instructions (2.3 mm is fine) then test-assemble the lot without glueing anything but with everything held together as tightly as the final thing will be.  Check that the wheels run freely; if not, you may need to drill the 3.2 mm holes in the side-plates slightly deeper.  With the wheels visually centred between the side-plates, mark on the bogie centrepiece where the Fosmotor suspension brackets land.  Disassemble everything and now glue the nylon axle bearings into the holes with cyanoacrylate adhesive; be sparing with the cyanoacrylate adhesive, you don't want it collecting in the hole.  Drill 6 mm holes in either side of the bogie centrepiece where you made the marks, also drill the 6 mm hole in the centre/top while you're at it and again test assemble the lot, easing the Fosmotor suspension brackets into the holes and making sure that the wires come out over the top of the bogie.  You will probably need to disassemble and reassemble a few times, using a needle file to ease-out at least one of the holes in the bogie centrepiece sides in order to get everything to assemble nice and square; you can see the shape one of the holes ended up for me in the picture below.

Mark bracket positions Drill holes Eased-out hole
Fitted
Fosmotorised bogie assembled

I originally tried making the steps up to the cab with folded aluminium (you'll see these in some of the videos below) but they stuck out too much so instead I printed them in black ASA. The steps were adjusted as necessary with a sharp knife, ensuring that the portion which hooked-over inside the bogie did not foul the wheels, and glued into place with cyanoacrylate adhesive.

Steps glued into position

Chassis

I decided to CNC cut the base of the chassis on my High-Z/S-400T CNC milling machine from cnc-step.de.  I began by making a few drawings, including castellations that represent the wooden steps up to the doors.

Motor
                open second base
Motor open second brake base Trailer
                composite base

Note: the bogie attachment holes are 6 mm and are drilled beforehand in a stand drill so that I can then use the same holes to bolt the plate to the milling machine while it is being cut (and in fact I had to make an additional 6 mm hole in the centre of the plate for that purpose as well).  I added two slots which will accommodate Velcro straps to hold the battery in place within the chassis of the motor open second.  Here's a preview of the motor open second base, drawn in VCarve, the software which happens to come with my milling machine, and the resulting work.

Base
                  plate preview Milling in
                progress
The milled result

I cut the first version in SWG 10 brass plate (3.2 mm thick, which just about scale-matches the 110 mm depth of the upper part of the visible chassis frame), thinking that I could then braze a length of brass bar to each end to which I could attach buffers etc.  However, this was way too heavy (over 1 kg) so I re-cut it in 3 mm thick aluminium plate (weighing less than 400 gm) and will attach a bar to the end by some other means.  The spindle was run at 14000 RPM and I used 2 mm and 4 mm Only One PM60 end-mills from Cutwel.  A sacrificial strip of 3 mm MDF was placed underneath the job so that the tools could cut through.

The bogies were mounted on the chassis using 40 mm long M6 cheese-head hex bolts.  I used one nut (5 mm high) and a washer then the bogie itself followed by a nice large washer and a lock-nut.  The drawings suggest that the buffer centre should be 180ish mm below the top of the chassis plate, 6 mm when scaled, leaving 5 mm clearance between the tops of the wheels and the chassis base. I will need to make sure that the buffers meet with those of other Gauge 1 rail vehicles such as the standard coal truck employed in the picture below.

Nuts
Buffer height
Clearance
Assembled

I don't have a speed controller yet so I just wired the motors directly to a 7.2 V battery pack I had lying around and made a circle of Peco G45 track to see how it would go.  It certainly went; click on the link below to see how and refresh this page if no YouTube video image appears below this text, sometimes it doesn't load on the first attempt.

 
Having made the main part of the chassis, I spent some time creating and 3D printing the undercarriage, which included a box to house the electronics and some very intricate chassis sides.  Since these were too long for my 3D printer to print in one go, I made them in several parts and created a jigsaw pattern (with 0.2 mm clearance left between the jigsaw edges) to knit them together.

Jigsaw pattern

The key with the intricate chassis sides, even with a high resolution print, was careful finishing: I purchased what became a really indispensable Herzo hand-held rechargeable grinding tool and used it to roughen-out the ridge marks left on the top surfaces of the black ASA by the 3D printing process.  You can see this in the "before and after" picture below where the top part is not yet finished: it has obvious 3D printing horizontal surface zig-zags and stepping, while the lower part has been finished, taking about 15 minutes with the Herzo, usually on its lowest speed setting.

One side finished, one not, with the Herzo hand-held
        grinding tool

Here are all four pieces of the sides of the motor chassis finished and placed in the right order:

Chassis sides (in four parts)

After making sure that the jigsaw edges were nice and clean I was able to carefully squidge the lot together using a wooden-jawed vice, no glue required.

Jigsaw edge of chassis side
Chassis sides assembly
Jigsaws joined Chassis sides assembled

For the means of attaching the 3D printed parts to the aluminium chassis base, see the body section below.  Here's a comparison with the real thing; good enough for me.

Motor Open Second, right
                hand side.
Motor Open Second, right
                hand side, original

Getting there.

Chassis frame with
                  sides fitted Getting there
Getting there

Electronics

For the electronics I went with a Fosworks rig, built by them to order and including the wonderful Legomanbiffo sound.  The sounds are loaded onto a DCC chip and hence I needed a transceiver that spoke DCC, which made for a lot of small boards and required 12 V power.  The boards all fit into a box that is part of the 3D printed undercarriage design.  Of course, I could probably have made more of an effort myself and put together the system more cheaply but, aside from a pause to charge the battery, this professionally pre-wired rig literally worked out of the box, which was very pleasing indeed.

The
        electronics

The patch board allows for connection of forward and rearward lights and an aux output which I can use for internal carriage lights.

Click on the link below to see/hear a control/sound test; refresh this page if no YouTube video image appears below this text, sometimes it doesn't load on the first attempt.

 

Body

While showing the 3D printed cabin off at work one lunchtime I was asked why I don't 3D print the rest of the body. Given what I've learned about 3D printing, and what I don't know about folding aluminium sheet, it was worth considering. The entire body of one coach works out to be about 600 mm long while my printer maximum vertical dimension (the body would need to be printed vertically to avoid a large support structure) is about 200 mm, and I'd want to stay well below that as shapes can become unstable and print a bit "raggedy" as they get higher. So it would need to be split up into 4 or 5 sections. This is quite possible since the coach side is split vertically at each door section (see below).  I could use a guttering strip around the top and covering material of some form on the roof to hide the rest of the joint.  And the bodies have repeating sections which would reduce the amount of design work required.

Door view Body
                  permutations

I had a quick go at a test print of the "A" section above and that worked out pretty well, see below.  The remaining question was that of access. I wanted to be able to get into the body for maintenance but I also wanted it to be rigid and firmly attached to the aluminium base of the chassis.  In the end I decided to add a rim around the bottom edge with the slightest of clips to attach the body to the aluminium base and, by default, for the body to form a closed loop, i.e. including a floor.  This floor is actually only maintained around two sets of holes for more M6 bolts that will hold the chassis detail in place; the remaining portions of the body floor will be cut away (in the Blender model) to allow access.  I created traps in the floor to hold the nuts so that the bolts can come up from below; hence the bolts anchor the body to the base as well as the chassis detail.  The ASA-printed traps had a slight tendency to crack when inserting the nuts but this was nothing a little cyanoacrylate adhesive couldn't fix.  The required holes in the aluminium and the chassis 3D printed parts will be drilled only once the body is assembled to allow for alignment slop.  Note, also, the slots in the Blender image of the body sides for windows to be inserted.

Body test 3D print Closed loop body Trapped nuts

It took several days each to design/test-print the sections before final printing in natural ASA at 0.15 mm resolution.  I abraded the outside of each section pretty vigorously with grit 80 on an abrading stick, while avoiding removing features, the aim being to remove the "3D printed" look.

Body section D
                      abraded Body section D in
                positoin

Before I could complete printing the body I needed to figure out how the corridor joiners should work.  I tried 3D printing one out of Flexible PLA (from rigid.ink) which is really flexible and looks good but, despite being as thin as I could make without causing the printer to leave holes in the shape, the square nature of the structure meant that it wasn't concertina-style side-to-side flexible as I needed it to be.

Flexible PLA corridor
              joiner

However, at this point I had completed all of the 3D printable parts in Blender and, while up to now I had simply been using Blender as a means of creating printable shapes, it occurred to me that I could put all of the parts together, colour them and create a properly animated/rendered video tour.  Here it is; refresh this page if no YouTube video image appears below this text, sometimes it doesn't load on the first attempt.

 

Internal Features

The internal features comprised the seating and partitions; I decided not to attempt any form of luggage rack as it would not be visible from the outside in any case.  The seats were 3D printed in natural ASA at 0.10 mm resolution which, handily, needed no finishing aside from a quick rough filing of the horizontal seat surface; the ribs of the 3D print matched the ribbing of the seat fabric.  Mounting ridges for the partitions were built into the 3D prints of the body sections to which the 3D-printed partitions were glued.  The mid-body partitions were printed in pairs, one glued either side of the ridge to achieve the correct thickness.

Seats Partition

Back to Meades Family Homepage