Atlas O Gauge switch problems
We have been working since May of this year
(2006) building a new O Gauge 8' X 20' train layout. We started with all new
Atlas track and switches and ran our first trains about 2 to 3 weeks before
Christmas. Trains consist of a combination of old (1938) Lionel engines and
rolling stock up to current Lionel and MTH engines with DCS control. We have run
into a multitude of compatibility issues but this document deals specifically
with problems with Atlas switches.
Of the 16 new Atlas switches I have
installed on our layout in the last several months, 10 have experienced almost constant problems and several others are
intermittent. Some have exhibited more than one of the problems documented
below. Here is what I have found so far. Switches are all Atlas 60XX series,
mostly Atlas 6070, #6071 and 6085.
Update Oct, 2009: After several years of
troubleshooting and repair, we have finally reached the point where the Atlas
switches work pretty well! I can usually run an evening without a switch
malfunction now. We are now up to 25 switches on the layout including 6 on a
platform extension with a yard. See the update after "Problem 8" for
some of the patches we have installed.
Problem 1.
Many of the intermittent problems are loss of power when an engine crosses the
switch. This switch (shown) failed constantly and that made it easier to find
the problem. The section of the center rail between the points should be powered
by the jumpers shown but the some connections are bad electrical connections. What
makes it hard to find is the fact that the voltage measures OK if you just use a
voltmeter to check it without a current load on the track. However, as soon as you start
to draw current through the section (as when an engine crosses it), the voltage
drops to 0 and the engine stops. I have fixed several intermittent switches by
replacing the jumpers but it is time consuming and problematic. I have not found
any bad connections where screws were used (shown to the right of the arrow).
There are still other switches that have this problem once in awhile but I haven't had
time to find and fix them yet. The one's I have put new jumpers on are
working OK.
Problem 2.
This was a particularly difficult problem to find. The problem was that cars would sometimes derail when crossing this switch from right to left and in
the orientation shown. If you look closely, you can see that the top point is not
tight against the outside rail. When the switch is active, the point travels all
the way to the rail but then bounces back, just a little, when power is removed
from the coil. I had taken this switch apart several times to adjust spring
tension, etc. but was unable to resolve the problem. What I finally realized was
that the lower point was interfering with the fixed rail and the interference was
acting as a spring causing the points to bounce back when coil power was removed.
Other switches have enough of a gap here (lower arrow) to see light but this switch, and one other,
have the rails positioned tight, together. The interference at the base of
the fixed rail and point joint is what is causing the problem.
Problem 3.
Switches not throwing all the way is a consistent problem with 5 of the 16
switches we purchased. Taking the switch apart and testing each part
individually indicates there is no problem as each part works fine by itself.
However, when it is assembled, you can feel a definite resistance in travel as
the slider moves from one position to the other. The switch shown will not throw
completely to the turn-out position unless you manually help it. The slider
moves freely in the slot from the top (shown) to where it is shown then gets
progressively harder to move for the rest of the travel. I cannot find the
reason for this. Adjusting the spring tension, cleaning the parts, etc. have not
made any difference so I assume it is a problem with manufacturing tolerance
buildup but I don't know how to fix it.
I assume you cannot use a lubricant since nothing is mentioned in the
documentation about that but it seems that might help if it were possible. For
now, we operate these switches manually.
Update: I did try using a lubricant for plastic
called Krytox (by DuPont) on one switch that is really hard to get to. It helped
a little but didn't fix the problem so that isn't a valid solution.)
Update: Possible solution? A
reply to my posting on the Atlas forum suggested removing the switch motor and
checking that and the switch mechanism separately to help identify where the
problem was. In the process of trying that, I found that by just loosening the 2
screws (silver heads in the photo above) that hold the switch motor to the
switch body fixed it! I only had to loosen them a half turn each and it freed up
the travel in both directions. I'll continue to monitor this "fix" and
post a new note if any of them fail again.
Update again:
Unfortunately, the above mentioned "fix" doesn't work consistently on
all the switches that have this problem. I think I am going to have to replace 2
switches as I cannot find a permanent solution for them.
Problem 4.
All of our 6070 and 6071 switches have this problem. The rail shown is not undercut
as the other outside rails are so the shoe i.e. "rail jointer", (not installed on this rail) doesn't
fit properly. If you force a shoe on the rail to maintain a continuous ground,
then it bulges out at the top and interferes with the wheel flanges on several of the engines. When an engine goes over the switch, it makes
a sound like (clunk) and you can see it jump.
For some cars and engines, using a track jointer at this point will cause an intermittent
derailment. The Lionel 2-6-6-2 Mallet (steam loco) sometimes derails the front pilot
wheels and there are scars on the flanges of the drivers because of this. The
other switches that are undercut properly work
OK in this respect. I don't know if this is a manufacturing defect or a design
defect but in either case, it is a real problem.
At this point, I have removed the track jointers on the improperly
manufactured rails and installed jumpers across the outside rails on the track
sections that mate to these "fat rails". That seems to work as a
"work
around" for this problem.
Problem 5.
MTK engine derailments when crossing an Atlas 6070 turnout.
This is a new MTH RailKing 30-2422-1 engine crossing an Atlas 6070 (054)
turnout. Roughly 30% of the time, the rear truck on the engine will derail at
this location. I found the problem to be the center pickup roller drops below
the rail from the frog and doesn't ride back up as the engine turns. See the
next photo.
Sometimes the roller makes it back up but if it doesn't, it lifts the truck off
the track. Either the Atlas switch needs the center rail increased in length or
the MTH engine needs a slightly wider roller. (Note that the front wheel on the
truck is off the rail. That's because this is a dummy wheel without a flange and
doesn't ride on track on curves. The 2 wheels behind this one are flanged and
are on the track.
The photo below shows the location on the turnout where the center roller first
contacts the frog. Note the plastic insert in the switch that prevents the shoe
on operating cars from snagging on the center rail. This works for cars moving
from right to left (in the photo) but some cars still derail when coming from
the other direction. The only way I have found to prevent this is to use a very
thin wire and to wire up the shoes on the car trucks. That prevents the
operation of the car as well as the electro-couplers so it is an undesirable
solution but the only one I have found if I want to use those cars.
Update: I posted this article on the OGR forum and
received a number of suggestions for work a rounds to solve this problem. One
idea was to make a replacement insert for the one Atlas provides with the
switch. I tried that and it solved this problem and another as well. See the
photo below.
Excuse the yellow Funtack holding the insert in place (under it) - it will be
replaced with brown caulk after I have some run time on this fix. The insert is
extended toward the frog enough that it holds the center roller up so it doesn't
snag on the track as described above. Also, by extending the insert into the
point where the two center rails come together, it prevents the shoe on cars
with electro-couplers from dropping down on the center rail and uncoupling the
car when it crosses the switch. That was an added bonus but an important
one!
To make the new insert, I used a piece of .118" thick Polystyrene which is
just the right thickness. It sets just below the center rail so it allows the
center roller to maintain contact with the powered rail but is high enough to
prevent it from dropping down and causing the car / engine to derail. This looks
like an easy fix for Atlas as it could be a simple molded part. I made a fixture
and cut them out on a bandsaw then shaped the edges with a Dremel tool. An easy
fix. Thanks to all the folks on the forum that offered help!
I'm going to make a few small ramps that can be stuck to the sides of the center
rail to see if I can fix the problem noted below (#6). I think it may work also.
Yes, I know it looks like crap but so does a derailed engine!
Problem 6.
This problem has been documented in some of the forums. The problem is actually
2 problems.
1. The shoe on the bottom of the truck will sometimes snag on the center rail
(red arrow) when the car is traveling from left to right in the photo above.
Atlas provides a plastic part that you can snap in to prevent the shoe from
snagging when the car is going the other direction. It's shown installed to the
right of the arrow. This seems to work OK for that direction.
Not all operating cars with shoes have the problem when going in the other
direction. The Lionel passenger cars that have the center roller as well as the
shoe all derail in our case. I think it's a combination of the center roller
pressing down as well as the shoe pushing up when it snags on the rail. That
seems to cause enough "lift" that the truck will derail. Some cars
without the center roller will bump over it without derailing.
2. When the car is going the other direction, the plastic insert will prevent
the shoe from snagging on the center rail but then the shoe drops down on the
center rail and momentarily powers the coupler as the car passes over the
switch. In most cases, it is long enough to cause the coupler to open. This is a
problem with all our cars and engines that have electro-couplers that are
activated by the shoe. Some will open every time and some open sporadically.
One of the forums had an article about this and the writer said
the only solution he found to work was to wire up the shoe so it didn't snag on
or contact the rail. I have done that to all our cars (as shown at the lower
arrow). The problem with this "solution" is that now neither the
coupler or any operating function (log or coal dump for example) will work since
the shoe is wired up. This is not an acceptable solution for operating cars.
I am experimenting with a different shape for the rail, at the location shown,
that is wide enough to prevent this problem. I'll post the results if I find a
suitable solution.
UPDATE: I have been wondering why some operating cars seem to work and
others don't. I think I found the reason.
In this photo, an old Lionel operating coal car is in the process of being
derailed. Note that the shoe has started to ride up on the frog and that lifts
the front of the truck up and toward the turnout. In most cases, that will
result in the car derailing when it passes over the Atlas switch. However, not
all of the operating coal cars have this problem.
The car on the left in this photo is the one shown above and is an early postwar
car, before they came out with the magnetic couplers. Note that the shoe in is
the front of the truck. The photo on the right is a newer truck and the shoe is
now located toward the back of the truck. When I replaced the old design with
this newer one, the car doesn't derail since the shoe can't "lift and
twist" the truck now. Although it still "bumps" when the shoe
goes over the switch, the car will still operate! In addition, the problem of it
uncoupling has been eliminated by replacing the electro-coupler with the
magnetic coupler.
Problem 7
I thought I had it made but a new and unexplained problem has occurred. I have
had 2 switch motors freeze for no apparent reason. Both were working normally
initially. As I have had very little "run time" on the platform yet, I
assumed when the first one failed that it was a manufacturing defect and just
purchased a new motor and replaced it. The new one seemed to work OK. About a
week later, I had a second one (different switch) fail the same way. In both
cases, I had not activated the switch prior to the failure so I expected it to
work as it had the last time the platform had been used. This time, I took the
motor off and took it apart.
The plunger is locked in the "in" position and you can't even move it
with pliers. Note the dark color of the coil on the right. When I inspected it
closely, I could see that the plastic in the core had melted and that was what
was locking the plunger in place. Obviously, the coil had really
overheated.
These switches are operated by the switch control supplied by Atlas, the supply
voltage is 17VAC and I observe the Atlas instructions to only power the switch
for about 1 second. So I decided to try and determine what caused the coil to
overheat.
The switch control below the one labeled "6" is where the switch
control is located. I used Atlas #312, 5 conductor cable for all the switches
and the stripped ends are soldered then bent and inserted under the screws so
there are no shorts or other problems there. The supply voltage at the feed
terminals, as measured with my VOM, is 17VAC and there is no activation voltage
to the coils without the switch being depressed so I eliminated wiring as a
problem.
To determine how long it would take for the coil to overheat enough to melt the
plastic, I taped a small thermocouple to the outside of the coil (black tape in
the first photo) and activated the coil using 18VAC applied directly to the
switch terminals. Both coils registered around 22 Ohms before I started the test
so I assumed the coil was OK. The coil temperature increased at a rate of about
1 degree centigrade per second (~ 2 ° F/second). I ended the test at 20 seconds
and the plunger was still locked so the coil would have to have been activated
for at least over a minute to cause the plastic to melt. Clearly, that
eliminates operator error in activating the switch.
The only thing left that I could think of is a failure of the switch control. I
took the control apart and everything looked normal except for one small flake
of black plastic inside. I don't know where it came from but it looked like it
may have been a piece of flash from the molding of the case. I don't know if
that could have held the switch in the "on" position then broke off or
not. I reassembled the switch and was not able to reproduce a condition where
the switch stayed in the "on" position so I am stumped. I have no idea
what could have caused 2 motors to fail this way unless it was some failure mode
in the switch.
Has anyone else experienced this failure or know what causes it?
Since I don't know what has caused the motors to fry, I designed
a simple circuit to monitor if any switch motor is drawing current. Here is a
simple way to tell if any switch is drawing current because of a stuck button on
the switch controller or any other reason:
On my platform, I use a tap on an old Lionel ZW transformer to supply
17VAC to the switches. There is no other current load on this tap - just the
switches. There is a 10amp, fast blow fuse in series with the "hot"
terminal. Since each switch draws about .5 amps (AC) when it is activated, you
can put a 1 ohm, 1 watt resistor in series with the "hot" lead (B/C)
on the supply feed to all the switches. Then I wired an old VOM to measure the
voltage drop across the 1 ohm resistor. If none of the switches are activated,
then there is no current flow and the voltage drop is 0 VAC. If one switch is
drawing activation current, then the drop across the resistor is .5VAC. Each
additional switch will draw an additional .5 amps and add an additional .5VAC
drop across the resistor.
0 sw's activated = 0 VAC
1 switch activated = .5 volts
2 sw's activated (at the same time) = 1 VAC
3 sw's activated (at the same time) = 1.5 VAC
etc.
So now. every time I activate a switch by momentarily depressing
the switch controller button, I glance at the VOM. It should pop up a half volt
while the switch is depressed and then return to 0 VAC. If it doesn't, something
is wrong so I'll know to look for the problem.
I'll post an update if I find the problem.
Problem 8: Operating Atlas switch machines with the MTH
AIU switch control outputs.
The SW outputs on the MTH AIU are a center off, momentary contact (through or
turnout) that simulate using a momentary pushbutton. I thought I could simply
parallel the Atlas controller with the AIU contacts and thereby operate the
switches manually or remotely. Although I had limited success with this, there
was a sporadic problem with the switch assignments on the AIU changing when I
operated an Atlas switch motor. I finally tracked down the cause and it was the
"inductive kick" from the solenoid in the Atlas switch motor. Simply
put, the "inductive kick" is like a voltage spike when you abruptly
open a circuit that has an inductor (the solenoid) that has current running
through it. The voltage spike was traveling back through the wires connecting
the AIU to the switch and that was interfering with the operation of the
AIU. I tried several ways to suppress the spike without success so I ultimately
replaced the Atlas switch motor with one from Z-stuff and that solved the
problem.
Update: 10, 2009
About a year ago, I got so disgusted with the problems with the switches that I
covered up the platform and walked away from it. That wasn't the solution
however since I had nearly $6,000 invested and more than a couple man years of
effort with nothing to show for it. I couldn't afford to replace the switches
with Ross so after about 6 months, I went back to work on solving the problems.
It turned out there were 2 major contributors to the lack of reliability. One
was related to the design of the switch and the other related to the design of
the switch motor.
1. The bottom of the plastic slider that moves the points from
"through" to the "turnout" position is flush with the bottom
of the switch body at each end. Any grit or dirt that gets under that area
causes the slider to drag and that often results in the points not completing
their travel from end to end. Since I used painted Homasite over plywood for the
platform, the surface is somewhat rough and prone to collecting grit. Grit,
especially ballast, gets under the slider and that causes it to drag.
The photo
at the left is an old one but the area outlined is the area where the problem
described is located. To eliminate this problem, I carved out about 1/8" of
the Homasite under the area where the slider travels so grit couldn't cause it
to drag. The carved area isn't shown in this photo since it is an old photo.
I list it as a design problem since there is no protective cover
under this part of the switch and almost anything, even as thin as a sheet of
paper, that gets between the platform and the slider will cause it to
bind.
In any event, carving out under the slider eliminated this
issue.
2. The other issue is the switch motors. As mentioned earlier on this page, I
have had problems with them burning out and no clue why. In Problem 7 I
mentioned a way to monitor the current in the supply lead to the switches so I
would know if one was drawing current. That has helped me prevent future
burn-outs and has also helped me identify the causes. In nearly every case, it
has been a problem with the button controller that Atlas supplies with the
switch. Since it is a 2 step operation to slide it to the end of travel then
momentarily press it down, anything that gets between the slider and the control
housing may hold it down after you release it. Prior to understanding the
problem, my standard mode of operation was to press down and slide at the same
time. If there was something (in one case a mouse dropping) laying on the slider
and I didn't notice it, then pressing down and sliding forward slid it under the
housing and held the slider down after I released it. The other source of debris
has been plastic flash from the manufacturing process that was left in the
control housing but that was only true in 2 cases. Now I inspect the control
panel before operating the trains and clean off any debris that shouldn't be
there.
Ultimately, I will replace all the Atlas motors with Z-stuff motors so I can
operate them from the AIU but for now, I seem to have resolved most of the
operational issues.