Refusing to accept that poor performance was normal for Z scale model
trains, an engineer sets out to devise a better electrical contact to
eliminate stalling at slow speeds
This story does not relate a saving of multi-millions of company
profits or an earth-shaking engineering fix. Rather this is about a
simple electro-mechanical design improvement to a relatively
unimportant hobby item that still made many hobbyists happy This falls
into the category of ‘spare time at home’ engineering problem solving.
Many electrical, mechanical, civil etc engineers are, will be, or have
been into the art of diorama and model railroad construction (or
railroad modeling if you prefer). So are many non-engineers both male
and female. The hobby appeals to all with an artistic and creative
bent, and offers technical challenges and satisfaction to those who are
inclined to tackle artistic and technical challenges. In their spare
time they create accurately detailed and realistic three dimensional
sculpture and animated operational miniaturized scenes past and
I’ve been an avid railroad modeler for thirty years, starting with
hand-spiked HO scale (1:87), then progressing to N scale (1:160) with
good results. Then eight years ago I decided to try Z scale (1:220)
with very disappointing results at first.
The problem? Accustomed to reliable performance from HO and N scale
locomotives, I discovered that the performance of the Z scale
locomotives left much to be desired. Smooth slow speed running without
stalling was impossible to obtain no matter how clean I polished the
rails and wheels. When I called to inquire, the locomotive manufacturer
claimed that no one had ever complained about this before. The online Z
scale community accepted this as one of the drawbacks to this scale –
the locomotives just did not make very good electrical contact to the
rails. Vendors claimed “That’s just how these things work Poor
performance is normal.”
I almost gave up at that point, then I got mad and swore that
mechanical beasties the size of my pinky were not going to get the
better of me. I took them apart and saw immediately what the problem
Larger scale (N, HO, S, G) model locomotives have springy (beryllium
copper or phosphor bronze) wheel wiping contacts for electrical contact
to the wheels which do the electrical pickup from the rails. This
particular brand of Z scale locomotive did not have any electrical
contact wheelwipers. The electrical path was the haphazard contact
between the wheel axles and internal axle journals. The slightest bit
of dirt in these inherently poor contact areas caused these locomotives
to consistently stall at slow (realistic) operational speeds of under
30 scale mph.
Obviously the electrical contact to the wheels had to be improved. But
I tried several things. First was to verify that my failure mode
assumption was correct. I uncoiled little railroad coupler springs to
obtain springy metallic wire, then fitted this wire into the
locomotives to press on the wheels from above and electrically bypass
the axle journal contact. Saw an immediate 1000% improvement in
performance. But this approach had a problem – if the locomotive was
accidentally driven across an isolated block where the opposing block
polarity was reversed (operator error), the thin wires burned up before
the PTC devices in the power packs would shut down from the overcurrent
OK, I needed a more robust wheel wiping contact that can handle a few
amps during operational goofs. So I used flat phosphor bronze contacts
from another part of the same locomotive (supplied by the locomotive
manufacturer whom by then was beginning to take an interest in my
efforts) and found that they solved the problem in both wheel contact
and overcurrent tolerance. But they were a pig to install.
So finally I did some careful measurements and did a phosphor bronze
design that could be easily fitted into the existing mechanical
structure and would bear down on the tops of the wheels to press each
wheel independently onto the rail and at the same time provide
electrical continuity. Had a few thousand of them etched from a single
sheet of phosphor bronze by a local metal shop. Cost me a bit of money
while unemployed but decided the gamble was worthwhile – I was still in
the mode of custom-built model railroads and was hoping to sell a
reliable locomotive as part of the kit and kaboodle. This design worked
amazingly well, could not believe the first tests that they actually
fit into the wheel mechanism and kept running at low speed without any
After verifying that the slow speed stalling problem was no longer an
issue, I wanted to find out how robust the wheel wiper design was. I
built a small tabletop railroad that I could run an upgraded locomotive
24/7. More problems…
Many times after running overnight, I would get up in the morning to
find the test locomotive off the rails, in many cases on the table, in
a couple cases on the floor. For many days this was a mystery. Then I
got lucky and finally witnessed during daylight hours the cat attacking
the moving locomotive and batting it off the rails. To a cat, a Z scale
locomotive is about the size of a mouse. All subsequent tests got moved
to the garage so feline interference was no longer an issue. After 400
hours with almost undetectable wear I decided the pressure on wheeltop
approach to wiping contacts was viable.
Wheelwipers: Simple concept, superb conductivity.
Even then I did not know what I had. At a following train show I had
some of my modified locomotives running, there was another gent
displaying an incredible Z scale empire. We admired each others’
displays, then he asked me “How do you get your locomotives to run so
slow without stalling?” I showed him my wheelwipers and he said “Can I
buy some of those?”
Since then I have been selling the wheelwipers online and installing
them for those who would rather not do the work themselves. A nice
Another upside to this is that manufacturers of Z scale locomotives are
now providing product with reliable electrical wheel contact. A big
improvement in the hobby quality.
This article takes me back to year1992. I converted LEO hand pulled toy train engine compartments into electric train by fixing them with casette player motors,capstan shafts and pulleys.The metal wheels were from steei made in a lathe. The tracks were made from aluminium U channels(inverted) with wooden repers. Connect them with a battery adopter it use to run nice for some time and stops later due to the carbon deposited on the tracks and wheels. I got the idea of applying switch cleaning oil on the tracks and tested and found that it runs continiously for hours together.all the children around played the train and it was amusing!
I remember from many years ago, a quick Google search shows December 1965, a Popular Electronics article on a pulse power supply for model trains.
I believe it used a constant pulse (width and amplitude) plus a variable DC to allow smooth control at slower speeds. Even larger gauge trains run poorly at the slowest speeds, but
the article claimed that pulse power allowed slow operation.
Aluminum develops a non-conductive surface oxide which will insulate current flow to the wheels. Model rail tracks today are almost always made of nickel silver (actually copper nickel zinc alloy) whose oxide is electrically conductive. Your switch cleaning oil probably retarded the buildup of oxide.
It sounds like you had a lot of fun with the conversions. Nice when folks can use their engineering skills for "play". Not many careers provide that option.
Yes, pulse width modulation in various forms is almost universal today thanks to the transistor. It makes a small motor develop a higher low speed torque than pure DC, very useful when performing operations such as realistic yard shunting or spotting freight cars on a siding. It can also overheat a motor if the voltage is too high with the duty cycle low (I^2R with very little back EMF), so the pulse peak must be kept within the limit of the motor full speed DC voltage rating. Many a Z scale motor has smoked at low speed when used with a PWM power pack intended for HO scale.
In the article pulse width modulation was used but did not help the main problem which was the opening of the current flow to the motor. This was simply dirt and poor design.
One of the fun things about model rail is combining the hobby with some of the cool things that can be done with electronics and mechanics. Sometimes after a hard day's debugging it is relaxing to come home, pick up a soldering iron, and build a gadget that flashes LEDs at a grade crossing. For examples see
Model Railroad & Misc. Electronics
Way to go, Glen! That's the spirit that makes American ingenuity world famous. As a lifetime follower of model railroading I contend that my study of Dad's cabinets full of Model Railroader magazines is one of the primary reasons I'm really good with complex embedded computer systems today...
I was disappointed how unsophisticated the electronics was in model rail, but like so many real-world situations there is so much legacy kit investment out there you can't innnovate overnight. Every newcomer confidently attempting to sweep away the troubles of past systems completely underestimates the challenges of power distribution, control technique and packaging of the circuits. Model rail as a whole is its own universe, with only passing similarities to other fields.
So at the same time I have to be impressed with how much modernisation is getting done, much by hobbyists with no engineering background.
Nice work Glen. Lets hope you're on your way to becoming a wheelwiper tycoon... Have you patented it?
I had a 000 gauge train set (slightly bigger than your Z gauge I think, made by a British company Lone Star, and now defunct) when I was a kid, and remember similar problems. I had quite a lot of stuff but heaven knows what happened to it. It's a nice hobby in that it's wide open to engineering applications. And if you can make it pay a bit on the side, good on ya!
@DWilde1 and Sharps_eng - there are control systems available that interface with personal computers for manual or programmed operation of multiple trains. One system uses the traditional block method, another more sophisticated is standardized Digital Command Control (DCC) in which every locomotive is fitted with a programmable miniature digital decoder for motor and lighting and onboard sound control. Track power is bipolar encoded data which addresses and commands locomotives individually. Speed lookup tables can be programmed to compensate for mechanical variations between multiple locomotives coupled together so that acceleration and deceleration effects are matched. Motor back EMF is measured as feedback for a constant speed control loop. Other equipment such as signaling and turnouts (track switches) are also addressable from the same power/data bus. Many manufacturers provide inter-operable off-the-shelf equipment, the decoders are usually surface mount and can be made small enough to fit into a Z scale locomotive.
Much of the home-brew electronics is unsophisticated, that is the ideal way for an electronics beginner to get started by learning how to build something on perf board and from that how transistors etc function. It was the other way around for me, I needed something for my Heathkit H8 to do and that led to learning model railroading and assembly language. It was fun programming the driver wheels to spin a few half revolutions on startup just as the big steamers used to do, then gradually incrementing the pulse-width register to simulate acceleration.
That's the whole point - combine your mechanical, civil, electronic, photographic, and artistic skills and use them for fun.
Thanks David. The patent application was eventually approved but would have cost a small fortune if continued to completion.
Have to wonder how many of today's engineers got started from having a model train as a kid? Or coming from related scientific hobbies such as ham radio or astronomy?
Blog Doing Math in FPGAs Tom Burke 2 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...