Machining Tolerences for Side Rods

I've just been studying some dimensioned plans by Bagnall for coupling rods for a 0-6-0 diesel shunter, dated 1958, and was rather astonished at some of the small machining tolerances.

Between crank pins:
4' 5" plus 0.002", minus 0.002"
4' 4-3/4" plus 0.002", minus 0.002"
3' 3" plus 0.027", plus 0.029"

Between crank pin and tongue pin joint:
7" plus 0.001", minus 0.001"
7-1/4" plus 0.001", minus 0.001"

This must have been jig-boring territory, or by using a machine fitted with 'optics' in the 1950s, along with some control of ambient temperature. Readily achievable with the CNC machines of today - but after the steel has stabilised from machining of the general profile.

Were they kidding themselves, or did they really achieve this? Bagnall would need to have upgraded their equipment surely from the days of steam manufacture?

-Brian M.

 

In the days before CNC and digital read outs it was all done by graduated feed dials. That level of tolerance may not a problem but remembering the number of turns the wheel has made could be a bit of a challenge. A steam loco chassis has to have some play between motion components to be able to run over less than perfect track. In my opinion tolerances were a bit of a running sore between the drawing office and the shop floor. They would be on the drawings but the assembly shop would make it work. I also doubt that such tolerances would be generally achieved in those days as the machine shops were all open plan with no temperature control, but then again the measuring equipment was also at shop temperature. I think that the resultant fit was more important than the absolute measurement.
Years ago I had the privilege to watch Mike Satow and Steve Hart machine various motion parts for locos, most of the 'marking out' was done with chalk marks on the bed and the final position was achieved by the graduations on the machine dial (or in one case a counter like a milometer). The big gantry mill that Steve had in his workshop was a bit of a beast and I believe came out of the machine shop at Head Wrightsons where Steve had worked his way up from apprentice, a great man.

 

Thank you for your interest and for responding. However, the accuracy of large 1950s machine tool feed screws would not be adequate for machining components of that size, via the hand wheel to the dimensions specified, without some supplementary assistance (only plus or minus 1 thou on the 3' 3" rod).
Thinking more about this, they may have used "end measuring rods" (Lufkin brand can be viewed via Google). These rods used out in the workshop would assume the same temperature as the component, thus irrespective of the ambient temperature could give a good level of control.

Even today, it's unlikely acme type feed screws attain the accuracy specified for this work, without the use of encoders. Ballscrews might.

Would appreciate hearing from anyone experienced in this work. It's quite a step up from the demands of steam work.

 

IMHO steam locos aren't built to thous If they were they would soon wear to a more practical clearance., as given in the Lama handbook.

 

Was it Crewe or Horwich where clearances, even on new locomotives, were once described as being "a rattling good fit"? I think it was Crewe

 

What other delights can be found in that handbook?


Best grazing in the Andes?
Difference between a Llama and an Alpaca.

 

Horwich did indeed use thous, much to the amazement of Derby in 1923!

 

Some years ago, and just out of interest, I measured the temperature of the frames in the firebox area on a former GWR 57XX and found that they were some 50 degrees F higher than the coupling rods. This meant that the distance between the crank axle and that of the trailing coupled wheels was some 1/32" more when hot compared with them being cold. With a longer loco, a 4-6-0 for instance, with a longer narrow firebox between the frames, and with a higher boiler pressure (meaning a higher temperature), the expansion of the frame compared to the coupling rods would be more. Something has got to give, but it does tend to make a mockery of crankpin and other clearances being in 'thous'! I'd be interested to know how Mr. Wardale proposed to overcome this little problem with roller bearings on both axles and crankpins on his proposed 5AT............

 

Just what does give, I wonder. I understand why for a while designers preferred to stick to singles if they could.

 

Returning to the original post, the tolerances are not out of line for those given by the LMA (Locomotive Manufacturers Association) handbook for steam locomotives. My 1951 edition gives the following for coupling rods:

Centre to centre for crank pins +.003" -.003"
Centre knuckle to centre bush +.001" -.001"

Even though expansion may need to be accommodated by various clearances, it still make sense to keep the center distances to a tight tolerance so both sides match. The very tight tolerance on the knuckle to center bush dimension (pin to tongue pin joint) also makes sense as this tolerance is additive to the basic center distance tolerance for the overall rod.

With respect to machining practice, it is only with the advent of DRO and CNC that the machine itself can be relied on for setting dimensions, other than using the feed dials for final adjustment after an initial measurement has been taken with a suitably calibrated measuring device. As Brian points out special templates or measuring rods may have been used. Alternatively a method such as "toolmakers buttons" could have been used to establish the centers.

 

I have a fairly comprehensive set of drawings for the SAR Class 25 (roller bearings on all axles and crank pins). There's a devious little detail in the coupling rods, one bore per rod isn't round. They're dimensioned horizontally and vertically, and the horizontal dimension is 20 thou bigger than the vertical. It should be noted that in this application, the roller bearing is a press fit on the crank pin, but the rod is a clearance fit over the bearing outer ring.

This is for the original Timken rods rather than the SKF design retro-fitted to many of the class (British Timken made a mistake with the bearing spec originally and a lot of the rods were replaced before they came up with a suitable alternative bearing), but I think you can assume the SKF design would have included a similar dodge. You can also get roller bearings with specified radial clearance around 10 thou for the sort of bearing diameter used on the 25. The SAR application is of course for a wide firebox 4-8-4, so wouldn't have the same direct frame heating issue, but more clearance could certainly be arranged.

I'm sure, with his SAR experience, Mr. Wardale knows all about this, unlike the operators of a certain large American 4-8-4 in recent years, who apparently decided to tighten up the (plain bearing) coupling rod bearing clearances because somebody didn't like the clanking. It didn't go well...

 

I recommend reading Swindon Steam 1921-1951 by KJ Cook published by Ian Allen. An excellent book which covers (amongst many other things) optical frame alignment and coupling rod machining tolerances.