Oswestry Works

Locomotive works diorama in 4mm

Category: 3D Printing

The Apprentices

Researching Oswestry Works has taken me down all sorts of interesting avenues of investigation, from 15ft long Cambrian plans in the National Archives to well thumbed notebooks recording repairs and works undertaken.  A few months back a former Boilersmiths Apprentice John Dyke got in touch with me via RMWeb, he was researching his own and his family’s history and whilst looking for information on his time in the works and came across my project.   John had also been in touch with another former Oswestry man, John Morris, a Fitters Apprentice, and he arranged for the three of us to meet up a couple of weeks ago.

John Morris and myself taken by John Dyke

John Morris and myself taken by John Dyke

Being able to talk about the works with these enthusiastic gents was fantastic.  I could barely keep up with their stories and recollections so after an hour they agreed to me recording the meet up on my phone.  I’ve listened back many times already and it was a lovely experience to talk first hand about day to day life.  I took my collection of photographs and plans to help clarify some areas I wasn’t sure about – how were locos moved around the works? where were spares kept? how many people were needed to operate the traverser?  The list is endless and they were only too keen to help answer as many as I could remember.

Albert Jones (L) and John Morris (R) working on the cylinders of a 14xx in Oswestry Works

Albert Jones (L) and John Morris (R) working on the cylinders of a 14xx in Oswestry Works

What originally set me on this path of 3D scanning was seeing a collection of photographs taken by Geoff Charles in the early 1950’s.  The collection is kept at the National Library of Wales and they have very kindly given me permission to share some of the photographs here.  What is different about his photos is that they are of the workmen of the works, forming a small part of a vast body of Geoff’s collection capturing every day life in Wales during and immediately after the Second World War.  These photos brought the works to life and inspired the idea behind 3D scanning people for highly realistic, naturally posed figures.  Better still – scanning the actual people who worked there…!

Left to Right: Arthur Kynaston (Boilersmith), Don 'Wacker' Rees  (Boilersmith apprentice), Tommy Pritchard (Boilersmith Chargeman), Jack  'Kinnerly', Jack Whitby (loco fitter)

Left to Right: Arthur Kynaston (Boilersmith), Don ‘Wacker’ Rees (Boilersmith apprentice), Tommy Pritchard (Boilersmith Chargeman), Jack ‘Kinnerly’, Jack Whitby (loco fitter)

Talking with John Dyke and John Morris has answered a lot of questions I had about the works and how to represent it as a model. It’s also given me much more to think on – small details like where the fitters used to keep their coats, warming them on the heating pipes during the winters; where the erecting shop and boiler shop fitters respectively used to have their snap; the flow of work from the erecting shop and boiler shop to the machine shop and back; how components were taken to the bosh for cleaning; how some works were carried out in the back yard if the weather was nice! And much more. The most resounding feedback from both John’s however was that I must model the Machine shop, it was integral to the works and the model wouldn’t be complete without it!  Thankfully Geoff Charles photographed many of the men who worked in the Machine shop and there is a well detailed plan of equipment in Steam Workshops of the Great Western Railway by Ken Gibbs.

Thomas Marshall (boilershop labourer) cleaning out piston rigngs and Dai 'Swingboat' Wynne (fitters mate) scraping carbon deposits from a front valve chest cover

Walter Ancell (fitters mate) cleaning out piston rigngs and Dai ‘Swingboat’ Wynne (fitters mate) scraping carbon deposits from a front valve chest cover, taken in the traverser shed.  Note the primitive sinks behind!

For the time being however there won’t be much progression on the works model, other than the completion of a few locomotive detailing projects (9000, 9005, 9018 and 7819).  Most of my time now is taken with developing Modelu and its potential, including planning the first stage for a range of highly realistic figures.  It’s likely I’ll start out with loco crews and shed staff as they have quite a wide appeal.  In addition to 3D scanning I’ve had a steady stream of some interesting commissions, from terraced house chimneys to point rodding stools!   More to come on that front soon.

The men of Oswestry Locomotive Works Approx 1954

All photos reproduced with the kind permission of The National Library of Wales.  All loco works names provided by John Morris.

Modelu first steps

Missenden Abbey 3D Scanning Demonstration Sample Prints

It’s been a hectic time since the Missenden Spring weekend with not much time for modelling unfortunately.  On March 31st I officially finished work at my IT job and in effect officially started working self employed with Modelu.  The rest of March was spent finishing the garage/workshop conversion then April so far has been getting down to business working on the scans taken at Missenden Abbey.   I owe the Missenden organisers a huge amount, having the opportunity to demonstrate there has really helping bring things on a pace.

The scans were all edited in Netfabb, having anomalies removed where the person may have moved slightly or the scanner picked up the ceiling or floor.  Printing wise, I’d like to say everything has gone smoothly after using the printer for a few months, but there is always a bit of room for user error to mucky the waters…! After a few days of shenanigans I’ve been getting consistently improving prints, helped by input from jewellers who are the predominant users of the B9 Creator.

Missenden Abbey 3D Scanning Demonstration Sample Prints

I’m really pleased with how these have come out, the printer continues to exceed my expectations – but it does demand some serious time and energy, there is no way I could have got this far with it if I was still working full time.   Scanning takes some patience and imagination; its quite easy to fall into the trap of imagining yourself as the model instead of the real person, posing in ways which are either unrealistic or overemphasised.  There is a definite knack to it and it will come with practice.

These little cameo’s were fun to put together – this is exactly the kind of thing I want to do for Oswestry Works and what ultimately inspired me down this path in the first place.  Being able to have scanned figures of some of the men that worked there and being able to share and recreate some of their memories really appeals to me.  On that note, I’m meeting two former Oswestry men towards the end of the month, one a former boilersmith apprentice and the other a fitters apprentice.  I’m really looking forward to listening to their experiences, it will give the works project a real boost.  That’s if I can find time to work on it!

I did make time this evening to finish off No.9000 however, all its needed was numbering and allocating, so with some help from a fellow Cambrian modeller, Alan Jones I got some new plates.  He sells etched plates and yesterday I put my last minute order in, hoping to get 9000 finished before the weekend.  Alan kindly sorted my order out and got it in the post first class and the numbers arrived today.  Check out his website if you need any GWR and absorbed etched plates, he also has lots of plans to cater for Welsh modelling, Cambrian Railways in particular.  No.9000 ran with a Dean tender so it’s not quite finished yet, it also needs weathering at some point.  Smokebox plate is Pacific Models again and 89C shed code is from the Model Masters range.

P1190180

Tomorrow I’m off to ScaleFour North for the weekend, I’ve been invited to demonstrate 3D scanning so I’ll be there both days with all the kit.  If you want to have a go yourself, or see the samples, or just say hello, you can find me next to the Missenden Abbey stand, opposite Mark Tatlow’s superb Portchullin.

Scalefour North 2015

Saturday 18th April 2015 – 10:00 – 17:00
Sunday 19th April 2015 – 10:00 – 16:00

Queen Elizabeth Grammar School
154 Northgate
Wakefield
WF1 3QX

Admission £7.00
Scalefour Society members £6.00

P1190175

I really must spend some time painting figures…

The Workhorses of Oswestry Yard

The Cambrian system featured relatively few Panniers compared to other corners of the GWR, the exception being Oswestry.   57xx, 74xx and 54xx classes were represented over the years, but the most numerous were the 16xx class.  These little Hawksworth engines were not introduced until 1949 but were pure GWR design, specifically built for light branch work and shunting.  At Oswestry they were employed extensively shunting in the yards, as station pilots or on the branches around the Oswestry area.   Oswestry shed had 10 class members on its books at various times between 1949 and 1962.  Of these, 1602, 1603, 1604 and 1636 had long associations, 1604 spending its whole life of 11 years at Oswestry.  1604 and 1636 are two that I’d like to model in Oswestry Works.

Tom Wright's 16xx and a NuCast 16xx

A while back I picked up a built NuCast 16xx with a view to improving the detailing and adding it to the works roster.  Other than replace it’s incorrect 57xx chimney with a 3D printed one its not had any other attention since I bought it.  In the mean time, Tom Wright has produced a very nice 3D printed 16xx, including frames available via Shapeways.   The design and build of the 3D model is chronicled here on RMWeb.  Tom has kindly modified his original design to exclude some details which I’d prefer to add myself, such as sprung buffers, etched lamp irons and the like.    I ordered it from Shapeway’s on the 17th Feb and today it arrived.

Tom Wright's 16xx

Tom Wright's 16xx

The model includes a body shell, OO gauge spaced frames with nem pockets and a separate smoke box door and roof.  It has a nice solid feel, the lines are crisp and the body shape is captured perfectly – thanks to Tom’s design skills.  The wall thickness is good, the cab sides in particular look better than some RTR offerings.   The main benefit over the B9C is being able to print the loco in one piece, whereas on the B9C due to the restrictions on the build volume at 30 microns, I’d have to print the loco in parts.  However, when it comes to the surface finish the grainy/frosted look needs attention.  The whole body is going to need sanding down, fine for the cab and tank sides but a challenge for anywhere else.  Finer details such as the boiler fittings and smoke box door will be easier to remove and replace with prints from the B9C.

The key point though is that the design itself is great and it will be another nice detailing project to get stuck into.  It’s the material that’s the problem (for me at least!) unfortunately, though I’m confident with some sanding and fine detailed parts it will really lift the model.  It will get the same treatment at 5726 – handrails, boiler fittings, buffer beam detail, pipework, scratch built pull rods and lamp irons etc and will be out shopped as 89A’s No.1604.

It will have to wait a little while though as I’m currently converting our garage into a workshop, I’ve had too many days in there with the temperature hovering just over freezing whilst I’ve been working on the printer!  With the garage/workshop finished in a few weeks time I’ll be ready to get the new 3D business going in earnest.  I’ve still been working my notice so I’ve not as yet been able to devote a lot time into getting things off the ground, but come the end of March 3D design, scanning and printing will become my day job! It’s a bit daunting/exciting/terrifying, but so far so good.  There should also be a surge in progress on the actual building of Oswestry Works in a few weeks, finally…

Being back in Wales now its been great to have Oswestry only an hour away. A couple of weeks ago I had my first experience volunteering with the Cambrian Heritage Railway, the days task being clearing out the cattle pens near the Coney Green.  About 12 of us spent the day tackling the undergrowth as well as clearing old sleepers from the mainline.  Back breaking work but great fun and a nice change to be working on the real thing!

Oswestry Cattle Pens

Dukedog variations – Sandboxes

No.9000 Sandboxes and Top Feed fitted

A number of the Dukedog’s had above footplate sandboxes as determined by the frames of the donating Bulldog.  No.9000 and No.9005 will feature in the Works, both of which had this particular trait.  There is also another slight variation, No.9000 and No.9005 had their pull rods above the filler lids, whereas 9008 had them running along side the sandboxes, in between the wheel splashers and springs (it’s quite hard to make out, but here is a picture courtesy of the GWSR)

This evenings task was seeing how the 3D printed sandboxes for 9000 and 9005 would fit.  I’ve had to tweak them a few times to deal with the slight lip on running plate near the rear wheel splashers.  First job though was to remove the front sandbox lids from the running plate.  These are part of the metal chassis and were not going to budge with a knife, so I threaded a piecing saw blade between the boiler and the running plate in order to cut them away.

Removing the sandbox lids

First complication was that the front grab rails needed removing in order to get a decent cut.  These are also metal and quite hard to remove cleanly, in fact I only successfully pulled one, with the other three snapping off and leaving the peg in the chassis.  After taking off the sandbox lids I redrilled these ready for fitting replacements.

The sandboxes ping off their build supports really easily, but they do need a little cleaning up with a knife or needle file to ensure a flush fit.  The build support penetration can be configured, in this case it was set to 25 microns, so if you remove the supports carefully they just need a quick once over with a file.

I’ve experimented with what angle to build them at quite a bit, the best detail being printing them in the normal orientation, but this does mean the base of them is rather uneven – this is due to the resin dripping away and some of it curing.  I’m sure that can be resolved, once I replace the layer of silicone in the printer I should get even sharper prints.  But for the time being I’ve rotated the prints 45 degrees which gives best of both worlds.

Dukedog Detailing Parts

On the top of the sandboxes I’d added the short spigots for the pull rods to connect to, more as a test if anything to see if the printer would cope with them. It did, but in practice they are useless, there is nothing for the pull rod to attach to and its too fine a point for glue.  These were clipped off and 0.4mm handrail wire used instead, allowing the brass strip pull rods to be soldered to the handrail wire.

Fitting 0.4mm brass spigots

Sandbox pull rod fitting

With that done the next jobs were to bring No.9000 up to the same basic detailed level of No.9018 – replace all the buffer beam fittings, add the ATC electrical conduit etc.  The top feed also needed fixing and the pipework modifying, all tasks which involve a fair bit of handling and not ideal after just fixing the sandbox pull rods!  Next time round I’ll put them on last as the pull rod connection to the spigots is quite delicate.

No.9000 Sandboxes and Top Feed fitted

No.9000 Sandboxes and Top Feed fitted

Dukedog variations part II – No.9000

Tonight’s detailing has been some tentative steps with a revisited top feed design for the Bachmann Dukedog.  It’s still not 100% right but I’ve made a start anyway, by fitting the pipework temporarily I can get some idea of anything else that needs adjusting.

No.9000 has a couple of detailing tasks, the top feed and the next post will be on the sandboxes.  For the top feed I started out with some 24 swg (0.559mm) copper wire, taped to the boiler with some Tamiya masking tape to get the right shape.

P1180166

With the wire tacked in place it can then bent to shape, with a pair of 16 BA washers added to represent the pipe join before it dips behind the wheel splasher.  I’ve cut the tape here in about a 3mm wide strip, thinking that this could be a good way of simulating the thin metal sheet that covered the pipework from the top feed to just before the bend in the pipe.

P1180171

With the top feed temporarily fitted with Tacky Wax it can then all be jiggled around until it looks somewhere near right.  I had to make 2 slight cuts in the Tamiya tape to get it to bend around the shape of the boiler, as well as the wire.  I’m not sure its worked too well, though once I come to fit it permanently some super glue will help it mould to the shape of the boiler and the wire.  I think bend in the pipework where it runs back towards the cab needs to be lower as well.

P1180173

That’s it for the first entry on 9000, the next post will be fitting the top feed permanently and fitting the sandboxes and pull rods.

P1180177

Dukedog variations part I – No.9018

P1180136

With 5726 waiting on some final bits (smokebox number and shed plate) I thought I’d make a start on the trio of Dukedog’s that will feature in the works.  At this stage its detailing I’m concentrating on, converting them to P4 will be a challenge for later when I’ve developed some better skills…! There is a well documented process to convert a Dukedog to EM on Alan Gibson’s website, hopefully this could be a good starting point.

The 3 Dukedog’s will give some opportunity to model some of the slight variations between the class members as detailed by Russ Elliott on the GWR Modelling website.  Another great source is Railway Bylines Vol 16 issues 2 and 3 (Jan & Feb).  Both have 14 page or so articles on the life of the Dukedogs and some great photos. For the works I’ve selected first in class No.9000, No.9004 and No.9018, all 89C:

9000 (great reference photos at Dovey JunctionAberystwyth and Machynlleth)

  • Sandboxes above the footplate (determined by the particular donating Bulldog frame)
  • Top feed
  • No whistle shield
  • Small Dean/Churchward tender
  • Fluted rods
  • Lamp iron on smokebox door
  • Black background number

9004 (On shed at Portmadoc, in store at Wellington)

  • Top Feed
  • Short whistle shield
  • Fluted rods
  • Lamp iron on smokebox door
  • Red background number

9018 (On freight at Aberdovey, not long till the end, Oswestry)

  • Parallel buffers
  • Tall whistle shield
  • Red background number
  • Fish belly rods
  • Faded GWR shirtbutton motif on tender

Other options in the future might be 9005, 9014, 9017… can you tell I like Dukedogs… 🙂  Looking at the works registers for the period I’ll be modelling, there are plenty to chose from.

Works Entries - 9000 Class

9018 spent it’s last days in store out on ‘the batter’ siding at Oswestry.   Along with 9017 and 9004, these were the last of the Dukedog’s to see service in 1960.  I’d picked up a cheap Bachmann GWR Dukedog from Ebay with various parts broken or missing – the perfecting starting point for 9018 as most of the stuff which was missing I was planning to replace anyway!

First up are the buffers, I’ve used Alan Gibson parallel Collett versions.  I can’t say how easy or hard it was to get the original ones off – they weren’t on there in the first place! The remains of the old buffers are part of the chassis casting, so these were filed down slightly before fitting the replacements.

P1180125

Next was the electrical conduit for the ATC equipment – a small pipe routed through the cab front, down through and then along the running plate to the ATC kit under the front buffer beam.  This was a pretty straightforward task with some 0.4mm hand rail wire and some holes drilled in the cab front and the running plate.  Photos from a trip to the Bluebell to see 9017 in store helped with this.

P1180128

Next step, though probably a little early considering what else there is to do, was to fit the fine whistle and mounting bracket.  I’d printed this a few days ago after a couple of days of fine tuning the B9 Creator settings.  This is the finest detail I’ve got so far, though amazingly there is still some room for improvement looking at what others are achieving!   It’s fitted to a Comet Model’s long shield, part of the LS74 GWR detailing etch.

P1180130 P1180131

That’s it for 9018 now until a few orders come in next week – I’m waiting on cabside numbers from 247 Developments and smokebox numbers from Pacific models. For 9000 and 9004 I’ve final samples of the 3D printed top feed and sandbox prints currently curing in the UV oven.  Both these designs needed some revisions as the first round of prints weren’t quite right.  The top feed was a pig to design, not having an diagrams to work from just photos, but I think I’ve got it somewhere near now.  Sandboxes are quite straight forward, there are 2 types, the more common type with the pull rods running along the top, and a less common variation with pull rods running along side the sandbox.

P1180135

Pushing the envelope with a B9 resin mix

GWR 9000 Class Top Feed test print

After what was in retrospect disappointing results with Spot HT – through my own ineptitude I must add, I went back to the tried and tested  1:1 mix of B9 Creator Red and Cherry resin.  This isn’t to say one is any better than the other, but with little life left in the layer of PDMS silicone before it needs replacement I didn’t have the leeway to experiment further with Spot HT.

This last batch of prints has been more experimentation with what level of detail is possible.  Second time round I reduced the size of the whistles and the steam heating cock to be a more prototypical size, even more challenging for the printer.  After quite a few false starts (left the projector lens cap on, build table not calibrated properly, not enough resin, ultrasonic died), I finally got a successful batch of prints of the highest detail so far.

This feels like uncharted territory because I’ve not wanted to let myself believe that it was capable of this kind of detail, until I’d seen it with my own eyes, coming out of my own printer.  I firstly revisited the design of the top feed for the Dukedog, the previous version not having a recess for the pipework.  This was designed to fit 20 swg copper wire, but the printed hole is slightly tight. It was a good exercise to see how well the resin could be worked with and the small hole could be opened up easily twiddling with a small drill bit.  This top feed will end up on 9014.

The whistles came out almost perfectly, considering that some of the detail in the design was under 0.25mm it would be impossible to see with out a macro lens or magnification.  Once cured these fine parts are quite robust, they aren’t indestructible but they did handle tweezers and finger tips, positioning them and trimming away traces of Tacky Wax under a magnifying glass!  It’s the detail presented on these that has really left me speechless.

GWR Standard whistles and mounting bracket test print

GWR Standard whistles and mounting bracket 30 micron xyz print

The beauty of this resin is it’s use to complement traditional materials, in this case matching up the whistles and mount with a Comet whistle shield etch and two strands of 36 swg copper wire to represent the whistle pipework.  The whistle print stood up to the finicky task of glueing everything together well, with impatient tweezer nibs looming, third hands full, trying not to drown the details out in glue.  This is the kind of detail I was hoping to attain after a few abortive attempts last year.

So with that done I think I’m out of excuses as to why 5726 can’t be finally painted and have its correct numbers fitted!

5726 with new whistle assembly

Comet Whistle Shield Etch, 36 SWG copper wire and 3D printed whistle and bracket assembly

5726 with new whistle assembly

5726 with new whistle assembly

Experiments with Spot-HT Resin

Other B9 Creator users have had some impressive results usingSpot-HT resin, in particular for small highly detailed components.  The main difference with this resin to the proprietary B9 resins is that you cannot use it for casting, it is also much less viscous which should mean that it settles quicker.

I’d just about enough life left in the layer of PDMS for another print, so today I tested printing some detailing components using Spot-HT.   The test parts I’ve roughly drawn up in Sketchup, some will need a little more work depending on the results of the print.  The test pieces were:

  • 4x Pannier tank rear steps
  • 1x Top feed for 74xx
  • 12x GWR Whistles pairs and mounting bracket
  • 14x Steam heating hose cocks

Spot-HT Test Print components

First up, the steps have come out really well, with a thickness of 0.5mm they look just right.  The rivet detail has come out nicely but I will probably increase the size next time round.  There is some flash to remove from the edges of the steps, due to the positioning on the build table.  Another thing for next time is printing these small components on a sprue. They were a devil to get off the build table without damaging! That said, this material is pretty tough even before curing under UV light.

The whistles have lost quite a lot of their detail but I think this is down to the clouding of the PDMS.  However, the small nut on top of the whistles has printed really well, considering its 0.25mm wide its pretty amazing!   The steam heating hose cocks also look pretty good, with the addition of a turned brass pipe and handle made from 0.4mm handrail wire.  However both these items are slightly overscale, I’ll have to get the micrometer out next time and make them more accurate!

Spot-HT test prints

5726 with 3D printed steam heating hose cock

For now that’s it for printing.  I’m waiting on my spare vat to return from the USA as its being fitted with a replacement for PDMS called Nuvat, developed by Evert on the B9 Creator forum (details here).  This new material should last a lot longer than PDMS.

A steep learning curve…

The last few days I’ve been trying to get some test prints out of the B9 Creator I bought back in September.  To say its been a challenge is an understatement, however the support and help I’ve received from the B9 community has helped me progress from tomato soup to something physical.

The B9 Creator uses a DLP projector to cure a photopolymer (light initiated) resin.  The 3D print is sliced in the desired Z axis size, in my case 30 microns, each slice is then projected into the resin through a ‘window’ in the bottom of the resin vat.  The z axis build table steps up out of the resin as each layer is cured.  Once I’ve worked out how to calibrate properly and the optimum settings have been fathomed, it should be capable of some pretty mind blowing results.

However, at this point my first prints were a complete failure – nothing curing in the resin at all, nothing floating around.  Experimenting with higher exposure settings resulted in some layers appearing, but they’d detached themselves from the build table and were floating around in the vat.

After some more hair pulling I’d realised I’d calibrated the build table wrong, then retried with the higher exposure settings and voilà! Something was created, just visible on the base of the build table under the excess resin.   There are a number of post print steps that need to be carried out, removing the print from the build table, cleaning with 70% isopropanol, putting into an ultrasonic bath and curing in a UV oven.

Once I have some decent prints to share I’ll explain a little more of the process, in the mean time here is my catalogue of errors!

November activities

Erecting Shop test cuts

Erecting Shop test cuts

Quick update as I’ve not had a lot of time online lately – I’m in the process of moving back to Mid-Wales for a few months so things have been a little hectic of late.  Early in November I was able to spend a few days with Tim Horn getting some preliminary cuts of the works made.  Tim’s expertise in laser cutting is invaluable, his input really has transformed this project.

Here’s a few shaky captures on my phone, these are a few test cuts as Tim perfected the laser cutter configuration.  I’m hoping to be back again soon to get a dry run cut, without brickwork to see how everything will fit together, in particular the raised floor in the works, the traverser and the inspection pits.  In the mean time I’ll be using these test cuts to perfect how I’m going to paint and weather the interior.

The other bit of news is that I’ve finally constructed the B9 Creator which I had delivered in kit form back in October.  With the assistance of a mate we put it together in about 3 hours.  It’s a dream piece of kit, superbly engineered and well thought out.  It will be a little longer again before I have a test print to share, I’ve been slowly building up the additional equipment needed for post processing the prints.  In the next post I’ll be able to share some experiences on the build, calibration and hopefully by then, the first print.

B9 Creator v1.2

B9 Creator v1.2