I am not quite sure exactly what year I think people started quantifying in parts-per-million.
But I am sure that I am wrong.
For example, I was wandering around the York National Railway Museum a year or so ago when I came across an unexpected treat….
It was the 1906 North Eastern Railways dynamometer car No 902502.
This carriage is a piece of test gear which goes between a steam engine and a whole load of normal carriages.
It has a spring inside connected to the locomotive in front, and by measuring deflection in the spring the tractive force of the engine can be measured.
Combined with speed, this gives the engine’s power under prevailing conditions.
The treat was the quality of this spring, and the mechanisms that read from it.
Rather than the huge rough-hewn thing I was expecting, the 100 year-old spring consists of many long beautifully-machined leaf-springs set across the carriage.
All the centres are connected together to the rod that goes to the engine, and all the ends are retained in the dynamometer chassis.
There are no simple connections. Everything is joined using techniques that minimise friction – complex mechanism that have not yet got to the bottom of.
I did have a photocopy from a 1906 issue of The Engineer which described the dynamometer car, which I have mislaid.
However, I phoned the National Railway Museum and a guy called Roy Gibson came to the rescue almost immediately with a copy of The Gresley Observer which reports on a from 1948 re-test of the dynamometer car.
It clarifies that there are 30 leaf springs, each 2.3m long and 16mm thick; and that these were the original 30 fitted in 1906, none having had to be replaced over the years.
Paraphrasing slightly, the report says: The figures so far obtained do show clearly that the amount of friction in the draw gear of the North Eastern car less than the limits of accuracy with which either the calibrating device can be set, or the final record read, even with a magnifying glass.”
If I read and interpret the test figures correctly – do correct me if you know better – static testing the figures show an error of 32kg with a load of 2,000kg, dropping well below 10kg if the car was shaken to simulate movement – so better than 0.5% accuracy.
Maximum load is 18,000kg, with 2.3% maximum non-linearity from 0 to 11,000kg, rising to 10% from zero to maximum load.
Hysteresis on removing the load is under 22kg – I suspect this is without shaking – which was measured after 5,000kg load – so 0.44%.
A particularly precise and useful piece of kit in my humble opinion.
Mechanical disc and wheel integrators combined force and speed to indicate ‘work done’, and everything was recorded on rolls of paper chart.
Coal and water consumption were recorded by hand on the charts along with several other parameters.
After 1936, said The Gresley Observer, these chart include a one second tick track provided electromagnetically from an accurate clock.
Not a load cell or a transistor in sight.
or scroll down and respond there.