Britannia 70013 in 5” gauge – rebuild story 2018 - 2020


by Norm Lorton








Go to Latest Update





This series of write-ups will describe the rebuilding of a Modelworks 5” gauge Britannia. It will not be a detailed account of machining or tooling used but will discuss design and rebuild issues. It will hopefully be of interest to some, and of use to anyone undertaking a similar task. The model will retain its identity as Oliver Cromwell, partly because in full size it is one of only two in existence and this one is based near me at Loughborough, and also in recognition of what Modelworks tried to create.


I purchased this Modelworks locomotive in November 2018 with the idea that I would sort out its problems, tidy it up a bit and get it running by following an intense three months winter work. I have a Black 5 in build so the idea was to delay that slightly and have this big engine to enjoy in 2019. By following that fancy I soon discovered the law of 'Octuple Time Growth' - a bit like Murphy's Law but relating to the time taken for jobs in model engineering by hobbyists. You estimate a job at 3 months, then discover there is twice the amount to fix that you first thought so it is now 6 months, then you get picky and perfectionist and want to make it a better job so it is 12 months, and now everything goes much slower than you thought and there are Summer holidays to have and other things to fix in the house, and it has all doubled again to 24 months. Thus 3 x 8 = 24 months. So, will I finish it in November 2020?


This engine was one of the last of the Modelworks kits with a 2006 dated boiler.  It had steamed a few times but had never achieved its potential, and spent the last few years in a damp shed. Most of the known kit faults were still to be looked at.  Back in my workshop I stripped it down the last nut and bolt and Octuple Rule part one raised its head – the cylinder and valve bores were corroded and it would need a rebore.


I like many of Doug Hewson's thoughts on subjects. He wrote a good series of 18 articles published in Model Engineer between August 2012 and July 2016 that featured the rebuilding of two or three (I think) early Winson kit models. His writings will be the definitive version, but I hope I can add just a few personal views, successes and failures of mine to his story.  Doug says that “the introduction of machined locomotive kits was the best thing that happened to model engineering” and I agree with him. I also have been motivated to turn one of these kit engines into something to be pleased with.






The frames were stripped to bare metal and I was happy with them. They were strong and well braced.  The large top stretcher was made from 6mm plate rather than 1/16” sheet. The Modelworks design came straight from the drawings of the failed Winson company, but Modelworks did change several features. The motion and weighshaft brackets were improved and brass castings replaced aluminium. The various dimensions are more than 99% identical to the Perrier design. However, I cannot find out much about Perrier's work. One of his drawing copies I have seen was all hand written and undated and I can find no reference to his design in the Model Engineer archive. Norman Spink, who made the castings for early Perrier builds, published the drawings in the 1980s and he must have obtained the copyright. The copies supplied by Blackgates are redrawn by C.J.Gilbert and dated 1984, but no mention of Perrier is given.


One problem with the frames was obvious in that the weighshaft bracket holes were slots! Longer on one side than the other. Hewson mentions finding this on his Winson kits but he does not describe any investigation of valve timing. I can only guess that Winson made an error in the drilling of one of the brackets. Unfortunately, this error was copied through to the Modelworks kits and either the frames came with slots on one side or builders were filing slots if they noticed that the weighshaft was not square to the frames.


Clearly I had to look into this. I had previously been interested in the concept of valve timing and Walschaerts motion but it was not time I wanted to commit. Now Octuple Rule part 2 arrived so I read Don Ashton's book and others about valve timing. I could write several chapters about what followed but I will just summarise. I measured the frames by mounting them on my Bridgeport using the DROs and a laser pointer to obtain all the critical dimensions for the left and right hand sides. I put these numbers into a CAD drawing and also into Don's spreadsheet. I accepted that the return cranks would have to be remade but I wanted to keep as many of the valve motion parts as I could to minimise work.  I moved the weighshaft brackets forward by 0.050” RHS and 0.017” LHS to keep it all parallel. I now had reasonable figures from Don's spreadsheet with a return crank of 1.379” (Perrier 1.401”).


Don's spreadsheet cannot give a full analysis and it provides numbers to type into Bill Hall's simulator. This simulator program shows the valve events in graphical form letting you see the steam admission points at all weighshaft settings from full forward to full back gear.  The results gave me good news in that I was getting good matched events in forward and back gear and as cut off moved from 75% to 45% the forward and backward events remained balanced. The simulator also gave me the same dimensions as the spreadsheet for motions parts once we had solved a puzzle. Don's spreadsheet unknowingly gave different dimensions for eccentric backset to those required by the Hall simulator (degrees vs. linear offset complication) and this was unknown to me until the puzzle was solved by Duncan Webster.  I was put in e-mail touch with him by Simon Bowditch to whom I had been referred by Don Ashton as having taken on much of Don's previous interest.


With the weighshaft brackets clamped in the calculated positions I drilled each one in three places, reamed the holes and inserted dowel pins. Once it was bolted up I put the frames back on the Bridgeport and measured again to see if all was ok.


This all led to another mystery. The spreadsheet and simulator take note of lap and lead. These are important numbers that describe the movement of the valve piston head past its steam transfer port. Lap+lead is the maximum piston valve travel (divided by 2) and is fixed by the main cylinder stroke having its effect through the combination lever. The extra movement imparted through the expansion link is all designed to reduce and reverse this valve movement as the die block is moved by the weighshaft. The Perrier design set the lap+lead at 0.168” but because of the piston head length (0.312”) and transfer port drillings (0.187”) the lead worked out at 0.043” (0.168” + 0.187” = 0.355” and 0.355” - 0.312” = 0.043”). This lead value of 0.043” is excessive. It is there in full size engines to start the steam flow before the piston has reached full distance and finished the exhaust stroke, and the early admission of steam helps cushion the piston mass and ready it for its return. These are features not required in a model. Some designers will agree that lead is not necessary at all and others put 0.010”, no more, in the design. Personally, I think that excessive lead will show itself as a reluctance for the engine to pull away smoothly on small regulator openings. I have therefore set the valve bobbin heads at a full 0.355” to give zero lead and we will see how it runs!


All of the above took up the remainder of 2018. Today the frames are fully fitted with motion and have been air tested. I show them in this next photo for interest. In the next posting in a week's time I will describe work on the motion, pistons and valves leading to that air test.





Part 2: Pistons, Valves and Motion.


It is well reported by others that the Winson and Modelworks valves and piston rings do not seal. I expected to have to replace these but what I found was a bit worse. The law of Octuple Time Growth had struck again.


When I tried to pull the valve bobbins out of the valve liners the first discovery was that they were very corroded. It was a mini-brute-force job to remove them. The main cylinder pistons came out more simply but a lot of corrosion was seen in the cast iron cylinder bores. I estimated the main bores had perhaps five-thou or more of corrosion pitting, mainly at the bottom of the bore where water would have lain.







The pistons and valve bobbins appeared to be cast iron and the rings look just like brass in colour, although they might be a type of bronze. They had a little bit of spring in them but their non-sealing abilities were not surprising. The liners, each side in two halves, were removed following a baking at 300 degC to break the sealant. The two little securing grub screws were found and removed. The liners had to be twisted and pulled out as they have shoulders that butt against the block. The Perrier ones do not have shoulders.


Clearly a rebore of the cylinders was needed first. Modelworks had, for some reason, metricated the bore dimensions. In place of the Perrier  1 3/4” (1.750”) diameter cylinder bores these were 44mm (1.732”), and in place of the Perrier 7/8” (0.875”) piston valve bores these were 22mm (0.866”). It was therefore convenient to have a go at opening both these bores out to the Perrier dimensions. The cylinder blocks were mounted on my Harrison lathe using a big 15” face plate and the block clamped to a small 90 deg angle plate. I took some time to align it all. The DTI (Dial Test Indicator) was showing +/- 0.002” at the front of the bore but the back would be +/- 0.005” in error. Since I was measuring a corroded bore this was not precise anyway. I had up to 0.020” on the bore to remove so I tried a 0.002” cut to see where it was touching. By the time I had the bore to a nominal 1.740” all of it was being cut and just the patches of corrosion remained, most deep at the bottom. I used pre-made plug gauges to take the bore to 1.748” by which point all corrosion marks had gone. A three legged hone with medium and then fine stones was used to bring the bore to a good finish.


The main pistons were simple to turn from GR17 cast iron. They were mounted on their piston rods, gripped in a collet, and finally turned to size so that they would be concentric with the rods. They were finished and polished to 320 grit to 1.747” which could be checked as an easy fit in the bores, which after honing were possibly half a thou bigger. Cast iron pistons and bores will thermally expand at the same rate so more clearance is not necessary, but I think that one and a half thou clearance is good to keep an oil film in there. Two piston rings on each piston are 0.077” wide by 0.048” deep and made from carbon filled PTFE (25% carbon). These sit on 2mm viton 75 o-rings. The grooves are 0.080” wide and the depth is set to give the viton rings 0.008” compression. The rings were stagger-cut on a mandrel following the scheme first drawn by R. Etter of South Africa and published in EIM. I have used these successfully on another engine and I know a club colleague has also. To me, the design seems sound by employing the flexibility, sealing and wear properties of PTFE, but addressing the plasticity and thermal expansion weaknesses by placing the o-ring underneath.


The piston valve liners are made from bronze. The ports were measured to provide data for the Walschaerts investigations (described previously). I knew from others that their fit in the blocks was a problem and stories of liner to block leakage and constant exhaust blow were well reported. Following what another writer had tried, I fitted them back into the blocks with plenty of Loctite (648 high temperature permanent cylindrical retainer). Because the bores had been heavily scuffed knocking out the bobbins I chose to mount them on the lathe faceplate again and recut them with a 7/8” machine reamer. They were then honed to a good finish.


Now here I am going to confess to a stupid piece of design on my part. They all say we learn best from mistakes so I hope this short story helps others. New valve bobbins were made from cast iron and tiny grooves cut to take miniscule carbon PTFE rings and 1.5mm o-rings. Well, it works nicely in the main bores so let's try it in the valves! The first time I pushed one of the finished bobbins into its bore it got stuck, so I gently eased it by rotating the bobbin, and a length of PTFE ring squirmed its way through the sleeve exhaust port and was then seen poking out of the steam exhaust port at the back of the block like a tiny black worm! It was hopeless, the soft PTFE was being forced out by the o-ring into the relatively large exhaust port and getting torn up. Oh well, searingly obvious lesson learnt. Lesson to self to re-think three times before plunging into a let's-try-this build decision.


I designed new bobbins with screw caps to hold a pair of cast iron piston rings at each end of the bobbin. The rings would be the 'event timing' edges. The design copied others' previously published. I made the piston rings from SG (spheroidal graphite – stronger and more elastic) cast iron, it was the first time I had done so and all seemed to go well. I used the simple heat treatment technique of holding the rings open on a blade while keeping at red heat. I was not over keen on this way of putting a set in the rings, but others seem to have reported that it worked. All was reassembled in the cylinders awaiting the motion to be fitted before an air test.


The Motion:


I spent a day or so looking at the coupling rods and other motion parts and became increasingly sad and depressed. The finish was dreadful with laser cut edges everywhere and misshapen parts. Horrid big metric nuts were on the crank pins and 5mm clevis pins affixed with split pits held all the motion parts together. The law of Octuple Time Growth raised its head. That's why I felt depressed. But I knew I had to sort it out. I could not run the engine with motion looking like that.


All the motion parts were then reshaped and refaced by hand file and then 180 and 320 grit paper until they looked like they should. Quite frankly, I could have started with new laser cut steel parts and it would have been almost as quick. I spent two weeks on all the reshaping and then more time making oil cups, new castellated nuts and neat motion pins as described by Doug Hewson. The photo shows the old clevis pin and the new item, with the tool used to tighten the front pinned head, while a 6BA spanner holds a flat on the rear nut.








The finished motion was now looking smarter.












New return cranks were made and these replicate the early Britannia two bolt design, although this would have been onto a square crank end. The Modelworks crank end had a centre drilling and four drillings for the LMS later-style cranks. For now I have tidied the crank end with a thin screw on cap.


The Air Test:


I could write a couple of pages on the jigs and procedures for setting the valve timing. Sufficient to say that I collected the writings of several others and distilled the most coherent into a scheme of my own. By the end it all seemed quite simple, but it was not at the start.


The engine was stood on a set of rollers and an air supply connected through a valve and pressure regulator. A light machine oil had been fed through all the ports. When the air was turned on it simply blew out of the exhaust centre collector and the wheels stood still. A lot more pressure got the engine running but there was a constant blow up the exhaust when stationary, wherever it was stopped in its rotation. This meant that either the valve piston rings were not seating or the Loctite liberally spread around the valve liners was not doing its job.


About this time John Baugley was reporting on a web forum exactly the same problem and his use of o-rings to seal the liners into the cylinder blocks. I quickly looked back to Doug Hewson's reports in ME and saw he did the same. Arghh !!!  Why did I think Loctite would be enough! So, all apart, cylinder blocks in the small industrial oven again, this time up to 400 degC before the Loctite let go, and the liners were machined with two grooves on each half, 1.5mm away from the transfer port ring of holes, to hold 1.5mm viton o-rings. The liners were a poor fit in the cylinder blocks so I wanted the o-rings to exert a fair degree of outward pressure to make a good seal. However, I knew from John Baguley's writing that he had problems getting the o-rings to slide in past the long exhaust ports. Sure enough, at 1.25 and 1.30mm groove depth the o-rings caught on the ports. But at 1.30mm groove depth, and a wipe of oil around the ports, they slid in, and out again to let me see that they had not nicked. I finally fitted them ensuring that the outer 10mm of the block and liner were wiped clean and oil free. Into this space I dripped some wicking Loctite 290 in an attempt to hold the liners firmly.


All rebuilt and back on the engine for an air test again. Air was still blowing up the exhaust. Arrgh!! and triple Arrgh!!!


I had previously assumed that the piston rings were sealing and that it needed the liner o-rings. Actually, it would be very difficult to determine where the air was leaking past a liner. You would need a specially designed bobbin that sealed perfectly and slid in and out to test whether the leak occurred as the transfer ports were just opened. I tried making one out of acetyl with o-rings but that would not move past the exhaust ports. I then made a solid PTFE plug, drove that in and proved that when I closed the transfer port the air leak to the exhaust stopped.


So it was the new cast iron rings at fault. But how? I peered at them for 20 minutes but nothing revelatory occurred. I then put them into the ring insertion tool that I used to slide the bobbins into the sleeves and held the tool up to a bright light. I seemed to be seeing something like a solar eclipse when the sun starts to appear from behind the moon. A crescent shape. The sides of the ring nearest to the ring gap were not touching the bore of the cylinder. Perhaps a one or two thou clearance. I looked at all the rings and only two out of the eight made a reasonable fit in the bore. Those gaps would have allowed a big blow past. I had clearly failed in my ring manufacture. I might have made an error with the bore match of a couple of thou, but I suspected the heat treatment part of the process. I think they should be heated in a jig, held flat and over a precision pin, as Chadwick had said in his early papers. In future, I am now a candidate for the cold manufacture technique with the final machining to diameter being with a ring compression jig in the lathe chuck.







The photo shows a valve bobbin with the four cast iron rings and the sleeve that separated them below. I was able to design a simple replacement cast iron sleeve which the end nut tightens to, and thin virgin PTFE tyres. These are shown fitted to the bobbin above.


The design is from Maurice Taylor of Adelaide and has been used by many of his countrymen. A similar design has been used by Howard at NSME but I have lost my reference to that and cannot recall the sizes of the tyres.  The Taylor design uses a thin tyre working out at 0.064” wall thickness in my application. Underneath is a gap of 0.003”. This thin section and under gap hopefully means that thermal expansion is accommodated. Virgin PTFE expands at 0.006 – 0.010” per inch per 100 degC and most of this abnormal growth is in the region between 20 and 25 degC. It means that these tyres will expand their wall by only half a thou, however the whole diameter will also be growing by around 0.007”.  Although the ring will try and push out, the cylinder wall will cause it to expand inwards. I will have to see if the 0.003” gap under the tyre is sufficient. The tyre width is 0.001” more than the fitting gap to minimise steam blow by underneath.


It would have been better to put a very thin o-ring under the tyre with sealing benefits against blow-by under and by keeping the tyre out against the cylinder. It will be interesting to see if my rings still seal cold on an air test after they have been exposed to steam a few times. If not I can pull the bobbins out from the front and add those o-rings.


After reassembly the air test was pleasingly free from exhaust blow by. The engine ran nicely at well under 10 psi, in full forward and reverse, and with the die block taken back to near mid gear and the pressure upped to15 psi. The exhaust beats became a nice even 1-2-3-4 rather than EeeehPhufff-3-4, once I trimmed the valve pushrod lengths by 0.020” and then 0.010” adjustments, a job made easier by the hexagons I had machined onto the ends of the valve rods.


The air test has been posted on YouTube and can be seen from this link or just search on the words “brit 70013 5 inch air test”


11/02/20 – to be continued. Next posting we look at springs, bogies, ashpan and grate.