I've always been interested in the design of things as well as the actual construction of them, so in this section of the website I will be looking at some of the valve gears that have been designed for our models by the likes of LBSC, Martin Evans, Don Young ,etc. The idea will be to check out the designs using modern computer simulators, see how good the results are, and, if necessary, try and offer improvements. Some of the designs out there are already very good but others can be improved considerably by just 'tweaking' a few of the dimensions. Let me state right at the beginning that this is not intended in any way to be a criticism of any of the designers for whom I have the greatest admiration and respect. It is purely an exercise in using modern computer software to give a more accurate picture of the valve events given by a specific design. I do not claim to be an 'expert' on loco valve gears, more an 'enthusiastic amateur' but I have spent a lot of time reading up on the subject and I think I'm getting a reasonable grasp of the subject.

It is no easy task to design a model loco etc. from scratch and LBSC etc. did the best they could with what they had available. LBSC admitted he was no expert on valve gears and relied on bits of tin, cardboard, and pins to try and come up with something that worked. Thanks to Don Ashton, Bill Hall, Alan Wallace, and Charlie Dockstader, we now have access to design and simulater programs and equations that enable us to design a valve gear that is as near perfect as possible. Is it worth it you may ask? I think so. Why put up with something that 'just' works when you can have something that works well and does the job it's supposed to!

Of course, there is no point in designing the 'perfect' valve gear and then throwing it together any old how. The dimensions need to be accurate as a 'few thou' here and there can soon lead to errors in the valve events. This means setting out the pins for combination levers etc. using the most accurate method you can e.g. using DRO's or the micrometer dials on the mill or lathe. It is said that 'Curly's' own locos ran better than those built by others because he made the valve gears to more accurate dimensions than he gave in the 'Words and Music', where the dimensions were probably rounded up or down to the nearest 1/64" to make things easier for his 'Tyros'.

There are also far more accurate (and much easier!) ways to set up the finished valve gear than often quoted in the construction notes. For example, the hit and miss methods described to set up the return cranks and determine the eccentric rod length on Walschaert's gear are not necessary. For a certain valve gear design there is only ONE correct position for the return crank relative to the crankpin and this can be set with a simple jig before any of the other bits of the gear are even made. The length of the eccentric rod is also fixed and should not need to be measured by taking the average readings from a pair of dividers.

Similarly, the eccentrics for Stevenson link valve gear can be set once and for all before the axle is even fitted in the loco. Once again, there is only one correct position for them. Don Ashton gives several methods of doing this in his excellent book on the gear. As Don suggests in the book, it is very difficult to set the eccentrics by the usual trial and error method involving turning the eccentrics round the axle until the valve opening 'looks' right. Setting the eccentrics beforehand also means they can be locked securely in position. How many locos do you see limping back around the track after one of the eccentrics has slipped under load! This seems to happen to a poor unfortunate member of our club everytime he brings the loco for a run.

Having looked at a few valve gear designs, there are one or two common errors that seem to occur:

The worst designs for Walschaert's gear seem to be those using a suspension link for supporting the radius rod , rather than the 'slot and die' in the end of the rod method. As the radius rod moves back and forth, the suspension link causes it to move up and down as well whch has quite an effect on the equality of the valve events. I get the impression that the attachment point of the link to the radius rod is just 'guesstimated' so that the link hangs vertically when the reverser is in mid gear! Moving the attachment point a small distance either way can make a big improvement overall.

With Stevenson's valve gear, the common error seems to be to put the suspension pin/s for the expansion link on the centreline of the die slot when it often needs to be slightly forward of the centreline or behind it to equalise the cut-offs for the front and rear halves of the cylinder.

The worst case is when the expansion link is suspended from the top or bottom rather than the middle. This completely screws up the valve events! LBSC's Pansy is a prime example of this where the expansion links are suspended from the top eccentric rod pins. I've had cause to study the Pansy valve gear (See the 94XX page) and the valve events are very unequal. To be fair, LBSC did not have a lot of choice with the Pansy design as there is little or no room to suspend the links from the middle via the usual brackets.

Anyway, hopefully over the next few months I will add the various valve gear designs that I have looked at and show the results obtained from the simulators.

Click on the links below for the various sections:

Basics

Design

Simulators

Examples

Note that the above pages are still under construction!

Return to ME home page