Saturday, June 29, 2019

Comments on truck maintenance

By “truck maintenance,” I primarily mean correcting any deficiency in how an in-service model freight car truck operates, but it may be almost as important to recognize the need to get new trucks into suitable operating condition. That could then become part of what I’ve called a “rookie test.”
     I have posted several comments about my concept of a “rookie test,” meaning the testing I do with new freight cars entering service. The components are obvious things like freedom of truck swing, coupler centering, coupler height, trip pin clearance, and coupler knuckle spring operation. That test and some ways I implement it were described in the first of these posts (see it at: ).
     I have expanded on the original description of my rookie test in a couple of places. Perhaps the most relevant to the present post is at this link: . I’ll show a photo similar to one from that post to show an important part of my car testing:

What you see in the lower part of this “aerial” view of Shumala on my layout is a test train, backing through the reverse or diverging side of three turnouts, two of them no. 5.
     This test has become an important part of what I now do as part of a rookie test or else a “return-to-service test.” I couple up a string of cars, some that are being tested, along with some already qualified, and use a locomotive to run them back and forth through a sequence of the diverging sides of turnouts, at slow speed and then at faster speeds, to see if they behave. And when that test goes all right, I rearrange the car order and repeat, and then maybe create a third or fourth ordering of the cars and continue testing.
     What has caused me a certain amount of heartburn is that freight cars that easily pass the normal rookie test may not consistently pass the test shown in the photo above. Yet they are consistent in another sense, in that they continue to pass the rookie test. The question becomes, why can’t they pass this “running test?” Repeated examination of cars, and repeated testing (I am giving a short version of this, believe me) has shown that the problem almost always lies in the trucks.
     The truck problems are usually one of two issues. First, the truck molding may be a little warped, so that all four axle ends are not in the same plane. This tends to raise one wheel a little, and over curved track, especially turnouts, this drives derailments. You can detect this if all four wheels do not sit flat on a surface, but do this with the truck both right side up, and upside down, because the axle-end hole in the truck is bigger than the axle end.
     I have tried heating up warped truck moldings in very hot water and trying to bend into correct shape, but this nearly always fails. A truck like this just has to be replaced.
     The second problem that I have seen is when one of the axles in a truck doesn’t roll as freely as the other, and I mean a big difference. This may also arise from a somewhat flawed original molding. My test is to spin each wheelset and see how freely (and how long) it spins. It looks a little like this:

     Usually all four wheelsets in the two trucks are the same. But not always. When one wheelset doesn’t roll freely in curved track, where the two ends of the wheelset have to travel different distances, this too can facilitate a derailment. I have tried “cleaning out” the journal hole in such trucks, which rarely helps. Usually the dimensions are at fault, and a wheelset that is shorter will fix the problem.
     An example recently was a truck with wheelsets with 1.015-inch axle lengths. Three of the wheelsets worked fine, and spun freely in the journals. But one of them would hardly even rotate. Experimentation showed that in that journal pair, only a Reboxx wheelset, of 1.000-inch length, would work. With all four axles now spinning freely, the truck worked fine.
     This of course is the great advantage of Reboxx wheelsets: they come in a range of lengths, and I have been surprised how widely the journal width of commercial trucks varies. I have improved many poorly performing trucks, just by replacing the wheels with proper-length Reboxx wheelsets (you can visit their site at: ).
     Wheelsets over the years have been taken for granted, in the sense that we expect them to fit any truck we happen to pick up. But of course experience will soon teach that it just isn’t so. A wheelset that runs freely and smoothly in Truck A may hardly rotate in Truck B, and might be so short as to almost fall out of Truck C with no urging. I happen to like the Kadee Code 88 replacement wheelsets, and have used a lot of them in various trucks. But though Kadee doesn’t tell you so, their length over axle tips is 1.020 inches. That’s fine in many trucks, but too long for others. Again, Reboxx supplies an answer to this.
     Some readers at this point may be muttering, “Why don’t you just replace the entire truck and be done with it?” and sometimes that is indeed the only solution. But if the truck has a sideframe design you can’t readily duplicate, and it’s correct for the freight car in question, you need to try and save the truck. Or if the truck mounting boss on the freight car is not friendly to any other truck design, you just have to figure how to make that bad truck work.
     Incidentally, I should mention that I wrote a related post recently, entitled “Improving Model Freight Car Trucks” about primarily improving the appearance of model trucks, but the present post is more about the operational quality of trucks.  (If interested in the previous one, here is a link: .)
     So even though my rookie test reveals lots of problems with freight car operation, issues in the trucks can be too subtle to be entirely visible that way. More analysis, of the kind described in this post, may be necessary in addition.
Tony Thompson

No comments:

Post a Comment