Recreating the RA brakes

[Doug]

One of several missing components from my RA project were the distinctive brakes that were a signature feature of the model. While Douglas might have won a Senior and Sidecar TT with machines fitted with these brakes, the Works team only used them for two years and then moved on to a drum brake for the 1925 racing season. Numerous pictures of RA machines repurposed for cinder track racing show brake shoes and linkage removed or the entire lack of any brake components. Consequently, surviving brake hardware for the RA is rarer than the machines themselves. I was able to borrow a rear brake shoe and mechanism that looked like it had been removed unused from a machine. Yet another LDMCC club member had portions of the front brake mechanism and a rear brake ring. Work began on reverse engineering the hardware; as well as examining what was available in period photos or remnant with the few other surviving RAs. It became apparent that like the frames, there were early and late variants of brake hardware. As my RA is DF186 and fairly late in the scheme of things – '3rd variant' frame – it would not have the earliest version of the brake hardware like what would be seen on, for example, the Tom Sheard bike (1923 Senior TT winner). Most likely, being an Australian export circa 1925-26, it was sold directly for use on the cinders and never had brakes fitted. Much was missing and the entire bike will be pieced together from leftovers and numerous donors so we will never really know. For road use brakes, even mediocre ones, would need to be fitted.

Inner operating arm, rear brake. The initial design was a shaft and an arm permanently assembled. I think through silver brazing, though I was never able to determine the exact joint interface. Then superseded by an arm forged as a single piece.





Had my RA been fitted with brakes it would have had the later. For small quantities of reproductions, milling from billet is more practical than making forging dies, albeit wasteful of material and more time consuming. Even so, with some careful nesting the use of material can be maximized; though it meant a lot of milling. It is not as if I had to stand there at wait as the CNC machine was perfectly able to run itself unattended. The alternative to cutting out the blanks with a narrow milling kerf would have been sawing, but the method chosen had an advantage of work holding and uniformity. The billet was tack welded to a sub-plate, then milled through all but the last millimeter of thickness. Sawing would have yielded a narrower kerf, but it is not as if it would have gained five rather than the four blanks per slab size that I had to work with. Another alternative would have been waterjet cutting, but that would have entailed sending out, and I have had trouble in the past with residual abrasives embedded in the surface wreaking havoc with the cutters. Additional milling operations took it from a rough to finished shape.



Slab of raw material tack welded to a sacrificial sub-sheet to provide a means to hold the work.



Milled down to the level of the sacrificial sheet. Four operating arms and four islands of waste.



An original operating arm sitting on several blanks.



Twelve operating arm blanks; in case of attrition. The blanks are now surface ground to aid in uniformity of subsequent workholding.



Set vertical in the milling machine vice, with a fixture to support the operating arm (already drilled and reamed.) Preparatory to roughing the arm profile.



Arm profile rough milled.



Laid down on the milling table, preparatory to surface milling the contours of the arm.



One half of the surface contour of the arm finished.



Sequence of operations from rough profile to half surfaced, with original are the front.



Now flipped over and the second side finished, taking care to align both sides. The fixture/stud on the left remained clamped to the table and provided the registration from part to part, and both sides.



Again, sequence of operations.



Side by side comparison. Note faux forging flash under the flange of the eye to simulate the residual forging flash not entirely filed away on the original. As the original was hand fettled, the new reproduction are much more uniform in shape.



Preparing to turn the shaft.



Turning the shaft. Yet to do is the step where the square taper will be milled, and the threading the tip.



Done, except for milling the tapered square for the outer half of the operating arm.



I did not have an example of the outboard operating arm, so this had to be estimated from photographs. A plastic model was made using 3D printing technology, which was then posted over to the I.o.M. for a final check against the one on the late Bob Thomas RA outfit. Once the size and shape were validated, manufacture was similar to the inner arm; milling from bar stock in several setups to attack the workpiece from all sides. One deviation was that the small post for the return spring to hook onto was originally gas welded to the main forging. I opted to mill it all from one piece because I did not fancy welding thick and thin sections together and at the time I was in a 'lights-out' machining craze of getting the job started and then going to bed, so did not care if it added several extra hours to the job.



Initial work piece blank.



Faced, and tapered square hole added.



Turned on its side and the cup for the brake rod machined.



Turned again, preparatory for roughing the arm.



Roughing in process.



Roughing complete.



Finish pass complete.



A fixture to do the second side.



Work mounted to fixture preparatory to milling second side of arm. A square, tapered plug is fitted to the square tapered hole in the operating arm  under the bolt head to the right.



Roughing in-process.



Finish pass complete.


Probably the most technically interesting aspect was cutting the internal, taper square that connected it to the inner arm. I am not certain how they originally made this, but suspect it was hot punched, much like the later conical spline used on EW kick starter levers and subsequent models. Whatever the method, the challenge with machining would be getting a sharp internal corner. Electrical discharge machining would have been the ideal method, but I no longer had access to that type of equipment and did not want to pay the going rate to farm it out. So, I bought a Ø1mm endmill that had a 14mm reach.



Finish pass complete.


I bought two actually as insurance, in case the first one broke. As often typical with insurance, it was never needed. This was used to clean out 'the corners' yielding a 0.5mm radius which was sufficient to clear the corner break on the male portion of the taper. This delicate cutter did not do all of the work alone. The bulk of the material was removed with an 8mm pilot hole and larger diameter endmill cutters. The secret to success was a shallow depth of cut (0.15mm) and an aggressive (for the diameter) feed rate of 25mm a minute. This made sure the cutting flutes cut rather than rubbed, and combined with the monotonous, uniform feed rate of a CNC machine allowed one cutter to last the entire run of outer operating arms. I should add the new brake arms were made from heat treatable stainless steel, and stainless steel is prone to developing a hard skin through work hardening if the cutter rubs; either because it is dull or through too slow a feed rate. This hard skin causes the cutter to deflect, which causes it to rub and work harden more rather than cut, and pretty soon it deflects enough that the cutter breaks. While I had learned this approach years ago with micro endmills (Ø0.5mm and under) I had never tried it with one this long and slender.

I subsequently found that such a fine surface milling on faux forgings was not entirely necessary. A courser finish could halve the milling time at the expense of a little time spent at the bench with a mold and die polishing stone to smooth out the milling marks. If you have ever spent tedious hours rubbing parts with little bits of sand paper try polishing stones. They work so much faster that they are worth the extra cost.

To be continued...

-Doug

[Doug]

RA Bit by Bit

The frame for my RA project was altered extensively to an approximation of the later DT model.  Missing from the rear lower chain stay was the anchorage lug for the distinctive RA brake system. The lug was subsequently used on the OB model when it came out in 1924. In fact, so cleanly has the lug been removed it appears as if was never fitted. Judging from period photos, into 1926 and '27 RAs were being supplied without some or all of the brake components as the former road racing machines were being sold stripped down for the rapidly growing sport of dirt track racing. Yet where it can be ascertained, they all seem to have that particular lug present. Indeed, it continued redundantly to be seen on early DT frames.

And I needed one. Originally these would have been a steel casting or perhaps even a forging, but for the sake of replicating one plus spares, milling from billet was more practical.


It starts with a block of steel.



Pivot holes and their faces (as counterbores) from the outer face; that side facing away from the tyre.



Stood on end and the hole for the chain stay tube bored.



Flipped over and the faces of the pivots counterbored from the inner face; that side facing the tyre. The smaller hole for the reaction pivot has no counterbore as it is flush with the billet surface



Rough pass, inner face. Note cut has broken through to the counterbore from the opposite side.



Second rough pass with ball tip cutter.



Finish pass.



Thread milling.



Checking with homemade thread go/no-go gauge.



Production!


The first side was relatively easy as there is a square block to grab in the milling machine vise. The second side required some fixturing to hold the part.



Sawing away some of the excess material.



The fixture.



With workpiece installed. A crossbar is passed through the bore for the chain stay tube.



Rough pass.



Second rough pass..



Third rough pass. As this setup would not be as rigid as the first side grabbed directly in the vise, the roughing was done in three stages rather than two.



Finish pass.



All done.


To install over the frame tube, I will either need to split the lug or de-tube the chain stay from the frame. That will be a big job for another day.

To be continued...
-Doug

[Doug]

There are a number of smaller components that make up the RA braking system. All of which were missing from my project. One of these is the link that gives the rear brake shoe a parallel motion as it engages the brake ring. One end is anchored on an eccentric to allow adjustment of the angle of the shoe. The first batch of eccentric bolts for the rear was based on the example seen on the Bob Thomas RA outfit. This turned out to be wrong as that was the 'early style' and my RA, if it ever had brakes, would have been the 'late style'. Oops!


Early style rear link eccentric pins. Eccentric split collet on right for turning and threading the stud offset from the pin journal.


Early style rear link eccentric pin and link on the left. Later style on the right.

The rear link is a made from folded bit of thick sheet steel. The one hole is quite close to the bend, so to avoid distortion – and also to get the holes aligned – drilling and reaming is done afterward. In the next picture, an original link can be seen on the right adjacent to sequential machining operations.


Sequential operations.


Fixture for profile milling the links; one side at a time.



To be continued...

-Doug

[karri]

Some lovely machining there Doug. Id kill to have a cnc at my disposal, but ill have to keep hacking away at it all manually. Thankfully havent had to take on something this intricate yet.

[Doug]

Id kill to have a cnc at my disposal, but ill have to keep hacking away at it all manually.

Well you can do a lot with a hacksaw and file. Most of the sculpted shapes here were forgings filed and sanded by hand and eye; likely by an apprentice. They did not need to be a precise surface milled to within 0.001 of an inch. The CNC mill is a convenient tool to do this sort of thing, but admittedly overkill and it could be done without. I still had to sand these by hand with mold & die polishing stones in lieu of additional time machining at a higher fidelity. At some point of diminishing returns it is quicker to stop milling and fettle it by hand.

[Doug]

The front brake for the RA is smaller and of less surface area than at the rear to befit the prevailing practice to avoid having the front tire skid on the loose road surfaces of the time. All that was available in original components to directly measure was a brake arm, the parallel link and its eccentric pin. Curiously, the link had been intentionally bent on an pronounced arc, though a study of period photos showed it was supposed to be straight. The brake arm formed the second link of the parallel motion and provided the correct hole center distance. As for machining new ones, it was fairly conventional work.



Long front brake arm blank, step milled to thickness. The concave ends to the steps is a result of using a face mill, and will be obliterated by subsequent milling.


The blank mounted on its corresponding step fixture. The hold down screws are standard button head hex screws with the head turned back to a cone for more clearance while milling.


First though, a profile cut. I do not seem to have taken a picture of the rough and surfacing steps of the elliptical sections between the bosses.


The fully machined arm, still in the straight condition. As the arm is symmetrical, it simply needed to be flipped over on the fixture to surface the second side.


The bending fixture. Heat is applied and the arm coaxed around the first bend with a hammer and soft aluminum drift. Then the block for the tip of the arm is bolted down to the fixture bed plate and the reverse bend executed. I found that I could not heat and hammer the arm hard up against fixturing to get the correct form. The fixture had to be cut back more so as to go past the point needed to allow for workpiece spring back. This fixture got it close, with some fine tuning at he bench vise to make the offset parallel.


A pair of finished arms with some of the hollow pins the RA used.


After making a batch of six brake arms and doing a trial assembly, a major blunder was discovered. The brake arm copied could not possibly be correct for an RA and a little forethought on my part would have revealed it was too long to line up with the brake cable running down the rear of the fork girder. The correct arm turned out to be (surprise) the same as used on the OB and CW models; though those used a dummy belt rim brake of different diameter. Eccentric pins and studs also appear to be shared between models. The sample I mistook for a RA front brake arm turned out to be a CW rear brake arm. I had machine the longer arm in the straight condition, then bent to the requisite joggle. Not only was that a bit of a pain in the arse, but I had doubt I would be able to pull it off on the shorter arm with its more aggressive bends. So, second time around I machined the arms with the joggle in-situ.



Roughing the steps on the first side from a block. The little triangles along the vise jaws are a form of inserted teeth for gripping the work securely by a very small shoulder. I bit of fancy tooling I was trying out at the time.


Roughing the second side and drilling and reaming the holes.


One blank at each stage for comparison.


Which in turn is bolted to its own custom fixture.


Profile cut.


Rough and finish surfacing cut.


First side done.


Comparison of the long (incorrect) and short brake levers.


Second side needs its own fixture.


Rough and finish surfacing cut.


Before and after, second side.


After heat treatment. Material is 17-4PH stainless steel.


Next were the pivot stud to make. Thread gauges had to be made and shipped globally to the interested parties to test in their girder fork lugs. Half those surveyed used a pitch diameter 0.006" over standard, and one had a pitch diameter 0.003" over standard, the remainder being standard pitch diameter. Why the variation I do not know. The 0.006" certainly seemed excessive for general manufacturing tolerances, and this is the only case I have encountered on a Douglas where there were various 'oversized' threads.


RA front brake anchorage lug with homemade thread gauge inserted.


Some heat treated posts and their nuts & washers. Note back end of post is drilled hollow to save weight.


Group photo.

The front brake arm forms one part of the parallel link mechanism, and a small link forms the second link. This too was milled from solid, in this case using a convenient size of round bar for material. One side was faced just to have a flat surface to work from, and then counterbored in depth to the face of the bosses. The opposing side could have the bosses faced with a face mill cut. Between the bosses needed to be relived as well, and pocketed in. Once profiled, it would no longer be a pocket.


The blank mounted on the fixture. The surfaces for the bosses top and bottom are already cut.


Profiled, and the step between the bosses cut. The step in the fixture to support the work level can now be seen.


Incremental steps. Top and bottom surfaces of the 'blank' and a finished milled link below. The second side of the link requires no further machining as the step down between the bosses was already cut with the pocket seen on the right.


The last step was to turn down the waist of the larger hub, to save a fraction of an ounce. The setup is not shown, but was simply mounting the lever on an arbor in the lathe and reaching in with a hook shaped tool, avoiding collision with the whirling arm of course!

To be continued...

-Doug