Speed Control For Conventional Trains

This is a demo of the speed control in action. When the actual speed is less than the set speed an increase command is sent to the Track Power controller. This is indicated by a flash on the green LED. When the actual speed is greater than the set speed a decrease command is sent to the Track Power Controller. This is indicated by a flash on the red LED. The voltage changes can be seen on the voltmeter. The train is going up and then down a grade so several increases and then several decreases can be seen. Again there is no change to the train, only the addition of the speed sensor car.

What a wonderful idea!  From watching the video I concur that the system response should be quicker to prevent noticeable slowdown.  Would response be improved if you used a Legacy powermaster instead of a TPC? 

Also: if the root cause of the slowdown is a voltage drop in the layout due to poor wiring or loose track connections, using this type of speed control is a band-aid on a fatal wound.  But if the cause is legitimate, such as a grade, or train load through a combination of sharp curves, how about monitoring the current in the circuit instead of the train speed?  My guess is that current draw will start to increase before the train really slows down, and your servo loop will have a little more time to react.

I really applaud your effort, especially the fact that you're sharing it in the public domain.  I hope many good things come of it, and many benefit.

David - this is pretty cool.  I have a couple of questions, mostly just to clarify my understanding of your system.  You show a buck circuit (of which there are many available on da Bay) to go from track power AC to 5 VDC.  Aren't there some times when the track power voltage is less than 5 VAC?  Or does the TPC have a starting voltage of ≥5 VAC?  As I'm sure you know, there are buck/boost circuits (similarly available) that would cover operations where the input voltage is lower than 5 VAC.  Speaking of TPC, I guess you need to have something like that to provide digital control over track power(?)

My second question is about the speed sensor.  Your diagram shows a Hall effect sensor - but I'm not clear on how it is being used.  Are you using it to sense the magnetic field induced by the track AC or are you using the Hall effect sensor in 'switch mode' along with a small magnet mounted on the axle of the speed sensor car.  the latter would give a direct indication of the actual speed of the car - some of the smoke chuff systems designed to synchronize the chuffing with the rotation of the driving wheels use this method and either a Hall sensor or a reed switch to determine the rotational velocity of the wheel.  I would think - without really having done any of this myself - that this mode would provide the greatest sensitivity to changes in speed.

- Rich

Answers to questions.
TED S:
I am going to look into possibly using a Legacy Powermaster.

I have experimented on other projects to monitor AC current. The current
waveform is very noisy and doesn't change very much. I was monitoring the AC
current when a #60 trolley reverses.

RICHS09:
The buck converters put out 5 volts when the input AC is slightly less than 5
volts. Not many conventional trains run on this low of a voltage. The power on
command to the TPC starts at about 7 volts. The TPC's will go all the way to
zero.

Attached is a picture of the Hall effect sensor. The 10K resistor in the
circuit makes the output digital instead of analog. The software measures the
time per revolution in milliseconds. since one revolution moves the car two
inches. The speed is calculated in in/sec. The data sent by the bluetooth
module could be either one digit or two digits. I was having trouble on the
Master Control end figuring out whether the data was one digit or two digits.
The easy solution was to add ten to the speed data from the sensor car making
the data always two digits. The Master Control then subtracts ten from the data
that it receives.

David - thanks for the additional info.  To go back to your original post - the response lag in your system is due to the TPC response rate?  It would seem to me that on the sensor car side, the digital circuitry would 'know' that the car speed has changed essentially instantaneously.  Presumably there is also the question of how long the integration period is to see if there is a real trend in the speed vs something that is just fluctuating rapidly ...  (I confess I didn't dig into your sketches to see if I could figure out how you set up the speed trendline).  In any event, a very interesting idea.  I've played around with an Arduino a bit - my original intention was to use it for control of things triggered by train movement, like crossing gates or signals or - eventually - control of blocks.  I hadn't really thought about using an Arduino as part of an on-board system.   Without getting too far in front of my skis (exceeding my knowledge), it occurs to me that one might be able to use an onboard Ardunio as a train controller utilizing the BT capability... (?)

- Rich

Nicely done, David!

TRW

David I've been thinking about this... Because of the response lag, compared to fully integrated command and speed control systems that are commercially available, this system as it is may not meet people's expectations for "speed control."

-I do think it would work better with a DC motored loco having a low (numerically high) gear ratio.  For example: Williams Crown Edition brass, Williams old-timer 4-6-0, or Lionel's conventional Atlantic for instance.

-Finally, instead of trying to control the speed by sending commands to the Powermaster, why not just display the speed in real time "speedometer" (and maybe distance traveled.)  Let the operator control the train with a remote, or even the transformer lever!?

My $.02.  -Ted

I'm with Ted. S.  Models with a "grown up" gear ratio need no electronic help maintaining a constant speed.  I have locos with cruise control that will fluctuate more on grades than my WBB 4-6-0 does, with no cruise.  But no, it doesn't go 150 smph.

David,
Since the TPC attached to my Legacy controller responds quickly to a turn of the red wheel (depending on the programmed setting), I think that the lag in your system is not the fault of the TPC, but either the data processing time, its transmission rate, or the amount of data relative to the distance traveled.
My first suggestion is to increase the amount of data, i.e., the number of actual pulses per inch. At least add a second magnet 180 degrees from the present one. A second collar with two more magnets could be added beside the opposite wheel on the same axle, set at 90 degrees from the first two, with a second sensor tied in. 

Dave