SAR Class 26 - Red Devil


Some years ago I came across Jim Nolan's website on which he is describing the construction of a South African Railways Class 25NC locomotive in 7.25" gauge:

Northern Steam

Up until that point I had not had much interest in locomotives outside of the UK but this soon changed after I spent many hours on Youtube watching videos of SAR steam locomotives in action. I really like the look of them now - sort of a cross between British and American locomotives. More 'blobs and gadgets' than the British locos but not as 'busy' as the American.

The 25NC class (class 25 - Non Condensing) is basically an American 4-8-4 Hudson running on 3' 6" gauge track but I've decided to go one further and build a 2½" gauge version of the famous class 25 loco number 3450 which David Wardale converted into the sole member of the class 26 - Red Devil. The conversion and following testing of Red Devil is described in David's book 'The Red Devil and other tales from the age of steam' and is a must read for anyone who is interested in the design of modern steam locomotives. The book had been out of print for many years and copies exchanged hands for large sums of money. Fortunately, Camden Miniature Steam Services obtained permission to reprint the book and it is now available once again.

So far as I know there are no drawings as such for Red Devil itself but the drawings for the class 25NC were available until recently from South African Steam run by John Young so I bought a cd containing all the full size drawings a couple of years ago. Unfortunately, John has decided to close down his website but had a 'closing down' sale at the end of last year and offered all the drawings that he had as a 2 DVD set so I decided to buy that as well.

Armed with the 25NC drawings and David Wardale's book I hope to be able to produce a reasonable copy of the full size Red Devil which I believe is now languishing in South Africa somewhere awaiting it's fate. Hopefully, someone will set about restoring this historic locomotive and run it again.

At this point I should say that for the time being this is a computer only project as I have quite a few other projects that I would like to complete first. There is going to be an awful lot of design work involved as all I've got to work on are the full size drawings. Fortunately, the full size drawings are virtually complete and of very good quality and easy to read, unlike some!

Even in 2½" gauge this locomotive will be big! The scale works out at 0.7143 inches to the foot, slightly less than that for 3½" gauge, but the loco is similar in size to a small 5" gauge. The overall length will be about 65", cylinders scale to 1.43" bore x 1.67" stroke, and the boiler barrel will be 4.75" diameter. I printed out the loco and tender GA at full size for 2½" gauge and that's when you realise just how big it is!

I made a start on the frames some time ago and have done a bit more recently. The full size locos had the frames and cylinders as a mind blowing one piece casting which isn't going to be practical to copy, even in 2½" gauge, as they are just too complicated. Instead, I am going to use separate bar frames for the main part and then fabricate the front buffer beam and rear cradle sections to make the whole look something like the original from the outside. The cylinders and smokebox saddle will be separate castings or fabrications as well that bolt onto the frames.

At the moment I am just doing rough 3D drawings to get some idea of how things will fit together and then the individual parts will be refined into the finished items.

Work so far -


I couldn't resist modelling the smokebox at this stage to give some idea of the size. The one shown is actually for the single chimney 25NC but Red Devil was fitted with a twin lempor system so this will be replaced eventually. The scale diameter is just under 5".

I decided to fit the 'top plates' to the bar frames to give the appearance of the full size casting. This will strengthen the frames considerably and add extra weight. It will also hide any extra ballast if I decide to add some to increase the tractive effort at the rail.

I've had to reduce the diameter of the driving wheels very slightly as with scale diameter wheels the flanges touched. There is very little clearance on the full size loco (about 0.75") and with the oversized flanges we use this decreased to zero. I'm actually wondering whether to increase the tread width from the standard 2½" width, maybe to that of the 3½" gauge standard. The 2½" standard gives an almost scale width tyre but the extra width may help traction.

One problem I discovered today is that the drawings of the coupling rods are for the original versions that gave a lot of trouble. They were eventually replaced with ones with conventional knuckle joints but there are no drawings for the later version. Looks like I will have to do some scaling from photographs to get the size and shape.


Spent a quick hour this morning and modelled the Red Devil smokebox. It's nearly 1.5" longer than the standard 25NC smokebox and was presumably extended to accomodate the twin Lempor exhaust. That's unfortunate as a friend at the club had a length of 5" diameter brass tube that I could have used but now it will be too short.

The cutout in the smokebox barrel between the holes for the chimneys is for the feed water heater.


I've roughly drawn up the coupling rods by 'guestimating' size and shape from various photographs and drawn up a simplified cannon axlebox. These will be fitted with drawn cup needle roller bearings running directly on the axles as there isn't really enough room to fit a hardened inner sleeve without reducing the axle diameter more than I want to. At the moment I'm going for an axle diameter of 14mm. The smallest outer diameter of the bearing is then 20mm and I can fit one 16mm long.

Cannon Axlebox

Rough Coupling Rods 'fitted'

I've also drawn up and 'fitted' the shoes that fit at the front and back of the hornslots in the frame (the 'bronze' bits in the photo above). These will increase the bearing surface of the horn slots but it might be tricky to fasten them. I originally thought of using countersink head screws through the shoe into the frame but then realised I wouldn't be able to get a drill in to drill and tap the sides of the horn slots!

I was originally going to fit axle pumps (I like them, many don't!) but the one piece axleboxes make it awkward to fit eccentrics to the axle so I will probably go with injectors. Red Devil had two high pressure steam turbine driven centrifugal feed pumps fitted which were originally used on the Class 25 condensing locos but retained one injector for use in emergencies. The pumps proved pretty unreliable in practice but it seems this was due to lack of facilities left to repair and test them. I don't fancy trying to make a working centrifugal pump in this size but I could always fit an electric pump I suppose?

I also started to draw up a proper version of the injector as it would be interesting to make one of the correct vertical type. The injector comes out at about 1.5" (38mm) tall so shouldn't be too fiddly to make. This will be an ideal use of lost wax casting.

Start of the injector body


I've been very busy the last couple of weeks doing some CAD drawings for someone so they can get some parts laser cut for their Doncaster build. Also the water pump on the car had to be changed as the bearings were shot. My brother and I spent two days in gales, rain, and sleet doing that job!

I decided last night that I need to get the rear cradle done for the chassis so I've made a start on that. This will probably be the most complicated part of the whole locomotive which is why I've been putting it off! As I've been working on designing fabricated parts for Doncaster using slot and tab construction this seems the ideal way to fabricate the cradle. It would be nice to be able to TIG weld the parts together but I'll probably finish up silver soldering them.

The scale thicknesses for the various parts of the cradle are quite thin - often only 1/16" (1.5mm) thick so I think I'll beef it up a bit to be on the safe side and use at least 2mm steel where possible. It should be very strong when it's all soldered together and the thicker steel will add weight as well.

The cradle will consist of a top and bottom plate joined by vertical webs, all of which will need bending to shape before joining. Everything else will be tacked onto this basic structure.

The basic cradle so far

The cradle will have to be joined onto the main frames by a suitable means when I get that far. There shouldn't really be any need for it to be removed again so the fixings can be more of less permanent. I'll probably use countersink socket head screws which can then be filled and painted over to make the whole chassis appear to be in one piece.

Red Devil so far!

You can see from the assembly drawing above just how long the cradle is compared to the rest of the frames.

I must find a better way to export images from the 3D package. Cubify Design doesn't have direct export to image (unlike the more expensive versions) and I'm having to use a virtual printer to produce them. I've tried a few packages and they all seem to produce quite blurred images that are nowhere near as good as the direct export versions.


I had a quick look at the boiler and started modelling the shell. It's not too complicated with it having a round top firebox so no awkward throatplate to make as on a Belpaire type. The overall length is 24.65" and the grate area about 35 square inches so there should be no shortage of steam.

Boiler shell in place

The front parallel section of the barrel is 4.5" diameter so hopefully I will be able to get enough tubes in to try and match the free gas area through the tubes to the massive grate area but we'll see when I start drawing up the tube layout. It's always very difficult to get enough tube area with a large grate area such as this. Aiming for the tube area to be 15% of the grate area means I will need 5.25 square inches of gas area through the tubes. That's something like 27 off 0.5" or 35 off 0.4375" bore tubes which I doubt will be possible.

At the moment I've drawn up the barrel in three sections as per full size but I'll probably dispense with the short rear parallel section and take the middle tapered section right to the firebox. That would also enable me to increase the diameter of the front section of the barrel slightly if necessary.


I've redrawn the boiler shell without the rear parallel section of the barrel and that looks neater as well as removing one joint. It's also enabled me to increase the diameter of the front section of the barrel by 0.125" to 4.625" diameter. Actually, the whole of the barrel could be made as just one long tapered section and dispensing with the front parallel section altogether. It will all be covered in cladding so no-one would know! I could then possibly increase the diameter of the smokebox end of the barrel by another 0.125".


I've also been playing with tube layouts and, as I expected, it's all a bit of a tight squeeze to get enough tubes in. At the moment I'm looking at 24 off 0.4375" dia. firetubes and 5 off 0.625" superheater flues which gives a free gas area of about 11.4% of the grate area (Free gas area of tubes = 3.67 square inches. Grate area = 32.15 square inches with a 0.375" square foundation ring). That's not too bad and better than I expected. I've allowed a spacing between the fire tubes of 0.09375". That's slightly less than the Australian code which states 3mm (0.118") but I'm designing the boiler for me and I'm happy with the smaller spacing. It may be possible to increase the spacing slightly to conform and if I can then I will alter it. I doubt if anyone else will be daft enough to build one of these in 2½" gauge so I doubt I will be publishing the design!

The full size boilers have a combustion chamber but in a model they severely restrict the firebox tubeplate area for fitting tubes. At the moment I have got a very short combustion chamber, more just a bit sticking out from the front of the firebox, but I could probably dispense with it altogether and just have a flat firebox tubeplate. I might be able to squeeze another tube or two in then and it makes the firebox much easier to construct.

With not having the combustion chamber, the tubes are very long in ratio to their diameter and this upsets the Keiller ratio. At the moment the Keiller ratio for the tubes is 103 which is a lot higher than his recommended 70 (Jim Ewins said 80). However, I've reached the conclusion that the Keiller ratio is pretty meaningless, at least in our size boilers. The ratio was originally devised to get the best heat transfer from a tube of a given diameter and was based on studying full size boilers, but it's now pretty much accepted that in our small boilers only the first few inches of the tube actually contribute anything towards steam production, so the length is pretty immaterial. To me. it doesn't matter if the tube is 6 inches long or 20 inches long, the amount of heat it will contribute will be pretty much the same.

I've noticed that GLR Kennions still list 0.4375" dia. copper tubing in 22swg as opposed to the more common 20swg and the slightly larger bore of the 22swg tube does help to increase the gas area by about 0.7%. Every little helps!

One 'problem' with the boiler is that there is not a lot of distance between the firebox crown and the top of the outer wrapper and I can't afford to lower the firebox crown without loosing valuable tubes. It's adequate I think (1.375") but the top connection for the water gauges will have to go a lot higher than the scale positions in order to get a decent length for the gauge glass. The only way to do this will be to fit a boss on the outside of the wrapper as is done on a lot of models.

I was reading (on the 5AT website) about feed water heaters and came across a method that involves fitting a baffle in the boiler barrel towards the front tubeplate (known as an economiser). The baffle is made a clearance fit around all the tubes etc. and separates the incoming cold feed water from the main volume of water in the barrel. The cold water then gets preheated in the front of the barrel and slowly mixes with the hotter water by percolating through the gaps around the tubes. Sounds like a good idea to me and I might give it a try. I think I will also experiment with a stainless steel arch in the firebox. The bottom tubes are quite close to the firebed and an arch should help prevent the tubes getting blocked with ash and small lumps of coal.


Done a bit more doodling on the boiler and modelled the steam turrets of which there are two. On the full sized loco these are fed by long pipes from isolating valves fitted to the top of the boiler barrel just behind the 4 safety valves. You can just see the holes for these about halfway along the top of the boiler in the photo below.

On the model the turrets could be fed directly through bushes in the top of the firebox wrapper and the feed pipes made as dummies, although the scale takeoff point is quite a bit higher up and the steam would be drier. My only concern is that the feed pipes are over 7 inches long and condensation may be a problem. I could put an internal pipe inside the boiler to collect the steam to avoid that I suppose.

I've also modelled a bit more on the rear cradle and added the side arms.

I can't help thinking that these side arms would be easier to machine out of the solid rather than try to fabricate them. There's still a lot of brackets and other bits to add yet!

I had a quick look on the internet yesterday and found a supplier of 127mm (5 inch) diameter steel pipe with a wall thickness of 3mm at very reasonable prices. That would do for the smokebox. It's oversize in diameter by 0.104" but I can cut a strip out of it and then either weld it or silver solder it again. 3mm thick wall is a bit thin to machine down by that amount although it could be done. Maybe I can find some thicker wall stuff when the time comes.


I've about finished drawing the rear cradle at last. It's been very fiddly adding all the various brackets and getting them in the right place.


All I've got to do now is come up with the best way to fabricate it all and then finalise the individual pieces.

The two vertical brackets at the rear of the cradle are for the rocking grate cylinders but I don't think I'll be fitting a working version of that!

The complete frame assembly so far:

The overall length of the frames is 33.25" (844.55mm).



I started looking at the tender the other day and the best way to build up the frame. On the full size loco the frame forms the bottom of the tender water tank and is a hollow casting but my version will be much simpler but look something like from the outside.

The first idea is to make the top of the frame from a thick plate with rectangular holes to reduce the weight, add a centre spine underneath this, and then plate the underside so it resembles the full size casting. So far I've just drawn up the top plate which is 8mm steel. The size is approximately 27.5"(699mm) long by 7"(178mm) wide.


Originally, the idea was to get most of the parts either laser or water jet cut but I think that that will increase the cost considerably, possibly treble it, so I'm going off the idea for the larger and simpler parts such as this. Now I'm not working, funds are very limited so costs are important. To be honest, parts like this are pretty easy to produce by hand and the milling machine. It just a question of spending time rather than money.

With this in mind, I've ordered some 50mm x 8mm black steel bar for the loco frames and the steel tube for the smokebox. I also ordered a laser cut steel disc from the same supplier for the smokebox front. I may as well start collecting materials for the project even if I don't start it for some time!



Following a plea on the SAR Live Steam group on Facebook I now have copies of the drawings for the later type coupling rods fitted to the 25NCs. Many thanks to Andrew Matthews from Australia. I couldn't resist having a go at redrawing them last night and here are the rods roughly drawn out.



When I compared these to the ones I had drawn up before, I wasn't very far out with the dimensions that I had 'guesstimated' from photos. The main difference was in the position of the knuckle joints which I had got too far from the axles. These rods are going to be very complicated to machine, especially the fluting on the edge of the rods. They would be an ideal candidate for CNC if I had that available!


< Return to 2-1/2" Gauge Locos ......Page 2 >