About
Pull Testing and By-The-NumbersAbout
Pull Testing and By-The-Numbers
Browse the N Scale Database
Browse the On30 Scale Database
Browse the Large Scale Database
(A is the loco, B is a string hooked to the coupler and C is the string. D is the pulley and E is the weight.
F is the digital scale reading, and G is the test bed.)
About Pull Testing and By-The-Numbers
Pull Testing is a subset of By-The-Numbers (BTN) which also includes Speed Testing. Combined, this can help the buyer choose a locomotive which is more likely to meet their needs. If you don't care about a locomotive's pulling power or speeds at various voltages, BTN will tell you nothing you need. For those who have an interest in model locomotive performance, here is an opportunity to share some information.
I have long felt that performance information can help match a person to a product. One of the exciting things about being involved with a publication such as Model Railroad News is the opportunity to watch people learn about products and make informed decisions. We never wanted to be like the consumer magazines and enter into a semi-negative relationship with manufacturers. Also, we cannot possibly review all of the items that come on the market. Shortly after commencing my review writing career with MRN, I hit upon a vast ocean of modelers who wanted to know how well a locomotive could pull. Most wanted to know this information in car counts pulled.
I attempted to set up various arrangements which would accomplish tests with strings of cars. Unfortunately, I discovered that the two principles of testing were being violated. The first is reliability which wants to know if a test can be repeated later with comparable results. The second is validity which asks if the test conclusively proves anything. Car pulls are not reliable because no two cars are the same. Factors such as track condition, axle journals, curves, grades, and a thousand other variables kept me from getting the same results with the same cars on my own layout, time after time. Asking someone to then try the same number of their cars on their layout meant that their results would bear little relation to mine.
So, not only are car-pulling tests not reliable, they really aren't valid, either. I felt that it would be inherently misleading to present such numbers and then send fellow modelers out, only to have many of them come home with a product which would not perform as mine had. Nothing is gained by hiding this truth; many people will understand the situation and buy the product anyway. In an interesting wrinkle, it would appear that people will accept a locomotive for what it is, so long as they know that information up front.
We chose a method which would allow modelers to duplicate the process. It involves a length of track mounted on a board. A small pulley available from most hardware stores is mounted so that the groove on top is the same height as the standard coupler height. A digital scale makes the process easy, but other weight systems will work well, too. A weight placed on the scale is attached to the loco coupler via a string over the pulley. When the locomotive is powered, it pulls on the string, lifting some of the weight from the scale.
This is a classic method for measuring horsepower and tractive effort. It should be said up front that the term "tractive effort" is not based on measured pull but upon mathematical calculations involving factors of locomotive design. At MRN, we coined the term "Tractive Efficiency" to get away from the hoopla surrounding Tracive Effort. Our efficiency rating is expressed as a percentage and the actual weight pulled by the locomotive is termed "Weight Pulled."
Early on, I had to confront an inconsistency in method which could destroy the validity of the process. Some locomotives reach their peak pulling power before their wheels slip; others will continue to increase the weight pulled even though their wheels are slipping. The ideal method would be to provide a graph showing weight pulled against voltage applied with a mark where the wheels begin to pull, but that would have demanded too much in space, time, and resources. Instead, I tested several locomotives and concluded that a spike in pulling power just before the wheels slip is not a reliable measurement of a loco's pulling power. Better, we reasoned, to take the locomotive to full slip at full voltage and see where it was then. This would be a value most modelers could expect to achieve. If they can feather the throttle right next to slipping and pull a few more cars, that's a bonus.
Speed testing involves math. A measured length of track and a stop watch are the main tools. Varying lengths are useful. At slow speed, a short length should be sufficient. An inch doesn't sound like much, but I had a Bachmann N-scale 2-8-0 which took 32 seconds to travel a single inch. In practice, I use a formula something like this: ((Track Inches X Scale Factor/12)/5280) X (3600/test time)= scale speed. What this means is that I would take that inch, multiply it by the N-Scale factor of 160 and divide that result by 12, converting from inches into feet. So 1 inch X 160 = 160 / 12 = 13.3333 / 5280 = .002525. 3600 / 32 seconds = 112.5 X .002525 = 0.28 miles per hour. Basically, you are converting the distance into a scale mile equivalent and then multiplying that times a factor of an hour. The database computes the scale speed based on information put into a field for Track Length in inches, Scale Factor, and Test Time in seconds.
In order to get valid results, I like to have a length of track long enough for at least 10 seconds of test time at maximum voltage. I typically use a loop of track, accepting the impact of curves upon performance. This is a low-order factor; engineering rates the impact of a factor upon test results starting at (1) to indicate the greatest impact and so on downward. I carefully measure the distance around the loop, using measurement and checking with calculation. I may mark off short, straight sections where I can perform slower speed tests.
BTN provides a lot of numbers on a given locomotive. When scrolling through a list of tested locomotives, one should be cautioned to be careful of what generalizations they draw. First, each locomotive is just a sample. Others from the same production batch may vary considerably. Second, weights listed in italics represent a unit with added weight, making it pull harder than a stock unit. Third, some used locomotives will actually pull better because their wheels get scuffed and will bite into the rails better. Fourth, the vagaries of locomotive design can mean that a locomotive that outpulls another on the flat may find the situation reversed on grades.
Finally, please don't draw conclusions of a negative sort about MRN as a result of viewing this information. I own this equipment, and I provide the testing as a service to MRN. As such, I am also providing the testing as a service to you. It would be highly inappropriate to beat up on a manufacturer over the perceived deficiency in some number. If you develop a testing rig and get very different results, I will help you to troubleshoot the situation, so long as everything remains positive.
Finally, this website is my own and is not owned or subsidized in any way by Model Railroad News. You may see images here that appeared in the magazine, but that in no way makes this an official function of Lamplight Publishing. Errors and omissions here are entirely my fault. Some of the information in these tables are incomplete. Often, I no longer have contact with that locomotive and so cannot complete the data. In other cases, I simply haven't gotten around to finishing the testing. You will usually find more comprehensive data in the published By-The-Numbers than in these tables. In the interests of space, I only put in three speeds and none of the extra amperages in the speed tests. A (T) indicates a locomotive with traction tires installed during the testing.
HO LOCOMOTIVES as of 5-15-2002
| HO LOCOMOTIVES Locomotive Type | Manufacturer and Series |
Loco Ounces |
Pulled Ounces |
Percent Efficiency |
Stall Amps |
Scale Speed | ||
|---|---|---|---|---|---|---|---|---|
| Min V | 6v | 12v | ||||||
| Allegheny 2-6-6-6 (T) | Walthers/Rivarossi | 26.5 | - | 0 | - | 7.6 | 26.1 | 66.7 |
| Alco S-1 | Life-Like Proto 2000 | 9.5 | 1.9 | 20 | - | - | 3.4 | 41.4 |
| GP40 | Atlas | 15.7 | 3.2 | 20.4 | 0.34 | 14.9 | 36.9 | 84.1 |
| SD45 | Athearn | 19.5 | 4.5 | 23.1 | 0.75 | 6.7 | 34.7 | 100.2 |
| AC44CW | Athearn | 21 | 6 | 28.6 | 0.61 | 4.9 | 31.6 | 84.9 |
| GP9 | Athearn | 16 | - | 0 | - | - | 9.9 | 87.9 |
| U30B | Athearn | 17 | - | 0 | - | 8.5 | 40.7 | 107.2 |
| GP40 | Bachmann | 15 | - | 0 | - | 11.1 | 34.7 | 79.4 |
| GP9M | Walthers | 15.5 | - | 0 | - | 10.9 | 0 | 0 |
| Y6b | Rivarossi | 22 | - | 0 | - | 11.6 | 0 | 0 |
| SD7 | Life-Like Proto 2000 | 16.5 | - | 0 | - | - | 0 | 0 |
| BL2 | Life-Like Proto 2000 | 13.5 | 2.1 | 15.6 | 0.4 | 3.5 | 4.1 | 48.8 |
| J class 4-8-4 | Bachmann | 21.5 | - | 0 | - | - | 0 | 0 |
| GP38 | Atlas | 14.5 | 2.7 | 18.6 | - | 6 | 0 | 0 |
| 4-8-8-4 Big Boy | Marklin/Trix | 34 | 6 | 17.6 | 0.32 | 8.4 | 0 | 0 |
| EMD F3A | Athearn | 17.5 | 3.6 | 20.6 | 0.54 | - | 0 | 0 |
| EMD F3A/B | Athearn | 17.5 | 3.6 | 20.6 | - | - | 0 | 0 |
| RDC-3 | Life-Like Proto 1000 | 21.7 | 2.8 | 12.9 | 0.28 | 5.1 | 5.4 | 40.8 |
| 0-6-0T Steam switcher | Bachmann | 4 | 0.5 | 12.5 | 0.21 | 6.1 | 11.8 | 39.8 |
| F3 lashup, 4-unit | Athearn | - | - | 0 | - | 3.3 | 11 | 55.6 |
| Alco FB-2 | Life-Like Proto 2000 | 13.8 | 2.3 | 16.7 | 0.38 | 3.2 | 12.2 | 56.2 |
| Alco FA-2 | Life Like Proto 2000 | 13.8 | 2.1 | 15.2 | 0.41 | 3.3 | 8 | 54.5 |
| AC-12 Cab Forward (T) | Rivarossi | 20.6 | 3.6 | 17.5 | 0.33 | - | - | - |
| Alco F-2 AB Lashup | Life-Like Proto 2000 | 27.6 | 3.9 | 14.1 | 0.77 | 3.2 | 9.2 | 56.2 |
| Life-Like Proto 2000 | SW8/900/600 | 8.6 | 1.7 | 19.8 | 0.23 | 1.1 | 8.7 | 42.7 |
| GP38-2 | Athearn RTR | 13.4 | 2.3 | 17.2 | 0.44 | 3.5 | 36 | 97 |
| SD40-2 | Athearn blue box | 20.1 | 3.2 | 15.9 | 0.51 | - | - | - |