Remember the old joke – If you want to make a small fortune – first start with a large one.
Or another old joke – this is a non-profit organisation, it was not intended that way.
When I retired in 2000, at the age of 55, Orion had about £100,000 left in the bank at that point (from software development).
Five years later I moved from a 5 bedroom detached house worth about £400,000 to a 3 bedroom detached house costing £300,000. Unfortunately, repairs and rebuilding to the new house took up all the money released by that sale. It also took about five years. Somewhat oddly 21 years later this is now worth just under £1m – not that that actually benefits me any. Various health issues have also knocked a lump out of the last 26 years. Add to that a lot of time spent gardening.
The original objective for Orion models was just to keep my brain active in retirement. If it actually made a profit, that was a bonus. In practice, I have yet to even recover my costs. So that original £100,000 I started with has now diminished to near zero. I currently have orders for 31 locos of 10 different types, so there is the potential to improve that position if I can just find the time to build them all.
Initially I looked at three different types of railway modeling – G1 standard gauge (highly detailed and accurate), 16mm narrow gauge and larger scales. All live steam. I had spent most of the previous 30 years messing around with 16mm NG for the garden and Swiss metre gauge so by this this time I had a lot of drawings for those. Mostly they were ones that appeared in books or the model railway press. I was never very happy about the accuracy and detail of those drawings and had a load of my own photos of the full size stuff (pre-digital unfortunately). From those drawings and photos I had drawn up my own 2D CAD drawings to a standard I was happy with. I decided not to get involved in that market as it was already quite crowded. I looked at the large scale stuff and decided the market was too small – fine for one-offs, but even there the prices were too low to justify the time and expense involved. So I settled on G1 standard gauge. My first effort was a GWR City class built to near-museum standard. Unfortunately the number and cost of castings involved made the cost prohibitive. I bought a Taig CNC mill to make all the masters for the castings from aluminium. This took a very long time to make the masters and the software used to drive the mill was a bit clunky. By the time I had come to the conclusion that the model was not a viable proposition the move to a new house was underway and I had lost my garden railway. I did not get round to building a new one at the new house within the years of rebuilding the house. By that time all the newly planted shrubs had grown to the point where there was no longer any room left to build a new railway. So I no longer have any test track. Fortunately, along came Dick Moger’s DEE locomotive book. which gave me an inspiration for building a cheaper form of model locomotive. I built a very small number of those locos, mostly kits. These were followed by Dick Moger’s ARM1G locomotive as kits with some of my own improvements. Most of my work went into the etches. All this while I was researching other locos that I wanted to build – going to various steam galas to photograph the originals (now, thankfully, digital photos) and obtaining the original works GA drawings wherever possible (or in the case of the Q1 actually measuring it up myself with help from David Halfpenny). Taking inspiration from various mechanical parts from both ARM1G & DEE, I have developed a range of standard parts that will go in all the locos I currently have under development. The greatest part of my time goes into researching the locos, developing the 2D & 3D drawings for them and drawing up the etches.
I could go on at length about the history of the machine tools and software involved in actually producing the models but that would be boring and not relevant to the current position. Suffice to say that If I knew 26 years ago what I know now and had the technology available back then that I have now I would have saved many years of frustration and considerable expense.
So the current setup is as follows – one 13 year old laptop running Ubuntu used purely for emails, one 2-year old tower PC that dual-boots into Windows 11 & Ubuntu (but only one at a time, and Ubuntu is rarely used on this) – this runs in Windows 11 – TurboCAD for 2D CAD, SolidWorks Maker Edition for 3D CAD, Dolphin CAD/CAM for creating gcode for the mill and the lathe – these are the only software left that cannot run on Ubuntu – this also doubles up as the heating source for my back bedroom workshop in winter, one 1-year old mini PC (not even big enough to fit a DVD drive) running Ubuntu on a fast but low power usage laptop processor (almost as fast running Ubuntu as the big tower running Windows) – this runs FreeCAD for 3D CAD and Libre Office for spreadsheets, word processor and databases. and finally one old PC running Ubuntu and LinuxCNC as the CNC controller. The latter is due to be replaced by a much more compact wall mounted CNC controller powered by a Raspberry Pi 5 (I have all the parts – I just need to put it together) later this year. All these PCs and the machine tools are all connected via WiFi.
Now to the machine tools – 24 year old 4-axis Taig CNC mill (still as good as ever, but now updated with a high speed spindle, ER16 colletts and linear glass DROs). I am now considering an upgrade to ball-screws, old HobbyMAT lathe (my own conversion to CNC, but despite my best efforts has too much backlash and not accurate enough). The lathe is about to be replaced by my own design of lathe attachment to go on the front of the mill (Taig does an equivalent but that goes at the back and requires rebuilding the mill/lathe assembly whenever a change is made between lathe and mill operation). This uses the mill to drive the tool and the new lathe bed merely rotates the material and provides a slide for the travelling steady and the tailstock mount. This provides the accuracy of the mill to the turned parts and allows me to pass up to 1″ diameter rod through the headstock – uses ER32 colletts and 7/8″ and 1″ C5 colletts for the larger sizes. Both mill and lathe are driven by a LinuxCNC controller. Putting the lathe on the front of the mill saves me 3′ of workbench, mounting the new CNC controller on the wall saves me another 1.5′.
Elegoo 3D printers - Mars 5 Ultra resin printer and Centauri Carbon (with Canvas upgrade) FDM printer. Three previous Elegoo resin printers have now been retired - Mars 1, Saturn 1, Saturn 3.
Draper Band Saw 250 and Rexon BD150A disc and belt sander.
Evolution F210SMS 210mm Sliding Mitre Saw With TCT Multi-Material Cutting Blade (230v)
Canon LBP631Cw laser printer, MG5750 Inkjet printer and LIDE 400 scanner
12" sheet metal roller
Sievert gas torch with multiple burner heads, refractory blocks and Propane gas tank
Katsu high pressure hand pump for pressure testing boilers and gas tanks
Graskop Resistance soldering unit - sadly now out of production
90W temperature controlled soldering iron - generic type
Loktite AA326 glue and retainer compounds
Unidentified mini pillar drill with added compound table and vice
Metalsmith rivet embossing tool
My design of tapping stand
A load of storage for all my tools, parts awaiting assembly. partly built locos
Masses of tools for lathe and mill - drills, taps, dies, indexable lathe tools & mount
also see my recommended tool list for building model locos (here).
Masses of little used tools for general wood and metal working
Large pillar drill - little used, but essential when really needed
My own 3D printed collett racksAs my design and production techniques evolve some of my older equipment has become redundant –
Multiple airbrushes – now replaced by coloured acid etch aerosol primers and laser printed vinyl wraps.
Small compressor – originally bought for airbrushes, but still useful for testing loco parts.
Wolf WBC1 sandblasting cabinet – now replaced by sanding etches before detaching from fret and assembling.
Large compressor for larger spray gun and sandblaster.
The current (2026) replacement cost for all the currently in use PCs and machinery is around £9000 – maybe a bit more, I may have missed a few bits. I have a detailed breakdown if anyone is interested. Note that excludes all the drills, taps, dies, milling cutters etc. that you actually need to produce anything from metal.
I hate to think how much has been spent on now obsolete machinery, PCs and software.
It is immediately obvious that 3D printing is far cheaper, although resin printers can rapidly rack up costs of resin in failed prints or prototypes that need changes. FDM printers tend to be cheaper on materials and scrap can theoretically be recycled, though maybe not in the case of the reinforced filaments. The advent of cheap low volume printing from China and others via CraftCloud solves the problem of 3D printing in metal, but brass is not an option for those printers.