Hello

Just throwing this out there to see if anyone thinks this can be done and if so would be willing to provide the brains of a future project (my brain will not do).

Today with digital slot cars each car had a decoder chip in it. The track rails are powered with around 13vdc all the time and carry digital packets provided by the control unit to the cars via the braids. These packets are interpreted by the decoder chip. The decoder chip then outputs a dc voltage to the motor. An option with today's slot car tracks is ghost cars where you program the car to travel at the fastest possible single speed without flying off the track allowing you to race against them. The draw back here is that they can only travel at a single speed making them less competitive.

So my question is does anyone have any ideas to be able to operate ghost cars a variable speeds where the car would accelerate on straights and slow down on corners? Some guys have achieved this with photo sensors and bar graph type stickers placed strategically on the track altering the cars speed. Others have used reed switches with small magnets but this seems less successful as many cars have traction magnets. None of these guys are sharing how they have achieved what they have so I thought I would throw the challenge out there. Anyone interested in helping?

Thank you

 

What about hardcoding the cars with maps of the track? The car is programmed by first racing it around the track manually.

 

I have read discussions on replaying human controlled laps however there are a few negatives to this I feel.
1/ Changing the track presents obvious issues
2/ If the recording was to be purely a matter of varying the motor output based on time from a starting point, when the car drifted or experienced wheel slip this would likely throw the replay out ultimately to a point where the car would accelerate at the wrong point of the track and de-slot.

Thanks for the input MCU88

 

What about some sort of angle position sensor that detects when the car is going around a corner? I have forgotten the name of this sensor. Kind of like with smartphones when you turn them around the screen changes from vertical to horizontal and vice versa...

 

I have read about these sensors being used to map g force on corners and a pc based program being used to "learn" how fast the car can go without de slotting. Sounds in depth though.

What about a photo sensor that picks up bar code type stickers on the track? Anyone have any experience with this?

 

Just blue skying. How about an IR LED mounted between the rails at each end of a curve and an IR sensor in the car. First detection, the car slows down; second detection, it resumes speed.

 

What about some sort of angle position sensor that detects when the car is going around a corner? I have forgotten the name of this sensor. Kind of like with smartphones when you turn them around the screen changes from vertical to horizontal and vice versa...

I think you're talking about an accelerometer. It measures the G-forces. But that isn't enough; the car will still need to map out the track and remember where the turns are (record the g-forces experienced for every given position on the track). Because simply reacting to real time experienced g-force isn't going to cut it. The car needs to slow down BEFORE it gets to the turn, not after it's already into the turn and in the middle of rolling over.

 

I think you're talking about an accelerometer. It measures the G-forces. But that isn't enough; the car will still need to map out the track and remember where the turns are (record the g-forces experienced for every given position on the track). Because simply reacting to real time experienced g-force isn't going to cut it. The car needs to slow down BEFORE it gets to the turn, not after it's already into the turn and in the middle of rolling over.

I agree with the car needing to slow down in preparation for a corner.

 

djsfantasi I think your on the right path. It would be nice to not need to hard wire the track up which would make it easier to accommodate layout changes. Building on your idea what about using some sort of photo sensor in the car which picked up label stickers on the track. Different stickers command a different speed set point (perhaps 3 setpoints?). I have seen a video like this but no plans. Easy to change!

 

Here's a optical/ phototransistor that can be used to detect light and dark transitions. Perhaps it could be used to scan your stickers.
https://www.sparkfun.com/products/246
You're going to need a micro to process the data. Do you have one in mind? What scale are your cars?

 

This strikes me as a pretty tough project. If you go with optical marks, you have to plan on reading them at some fast but unpredictable speed, and the car may not be exactly in line with the track during reading. Plus, there may be some amount of ambient light getting under the car, so contrast may be an issue too.

I like the accelerometer better. You could probably run the car once around at a slow speed to learn the track, and then it would know where the curves were, and after that it could predict where it needed to slow down, calibrating its speed based on what it was telling the motor to do, mixed with when the curves actually showed up. It could be an "artificial intelligence" project, where the device would improve its performance as it gained more experience.

 

Here's an article that does what you are talking about

http://asar.biz/elises/slots.pdf

 

This strikes me as a pretty tough project. If you go with optical marks, you have to plan on reading them at some fast but unpredictable speed, and the car may not be exactly in line with the track during reading. Plus, there may be some amount of ambient light getting under the car, so contrast may be an issue too.

I like the accelerometer better. You could probably run the car once around at a slow speed to learn the track, and then it would know where the curves were, and after that it could predict where it needed to slow down, calibrating its speed based on what it was telling the motor to do, mixed with when the curves actually showed up. It could be an "artificial intelligence" project, where the device would improve its performance as it gained more experience.

Yeah that's what I had in mind too. One minor detail about it though, is that (FYI I know nothing about slot cars) I assume the car has no indication of where it is on the track. I'm imagining something like an encoder keeping track of pulses corresponding to wheel revs, and therefore distance traveled and position on the track. Or if it's a BLDC, then number of electric cycles. But I don't know how much traction the cars keep, and if this method could be accurate. Perhaps some low Hz signal could be put on the rails and the car could read doppler or phase shift to resolve position. Or maybe the data from the accelerometer could be totalized for position.

 

No, I was just thinking of calculating position using motor speed setting and time. So at an 80% speed setting, it should take this long to get from the point where the track straightens out after that turn, before it's time to slow down for the next turn. And we know what that turn is like, so we know how much to slow down and how long to keep the speed at that level. If the speed isn't 80% but some other value, then adjust the time accordingly. And if things don't seem to be working as predicted, then the car should be capable of re-calibrating, to make an update in the parameters.

Edited to say that I just read the paper by the New Zealanders that Djsfantasi referenced. That's some smart work, and it does demonstrate that the optical pickup method is workable. But I still think accelerometers would be more fun.

 

No, I was just thinking of calculating position using motor speed setting and time. So at an 80% speed setting, it should take this long ...

Yeah that would probably work and be much simpler. I've done that speed totalizing type of control before with hydraulic through a cogged motor on a toothed track and it was surprisingly very accurate.

 

Yeah that would probably work and be much simpler. I've done that speed totalizing type of control before with hydraulic through a cogged motor on a toothed track and it was surprisingly very accurate.

The important thing is the toothed track and cogged motor make for an excellent encoder. On a slot track, there is slippage, especially when accelerating, deaccelerating and turning, making calculation of position very difficult if not impossible. Slot cars are relatively light and scale tires do not provide a lot of traction.

 

The important thing is the toothed track and cogged motor make for an excellent encoder. On a slot track, there is slippage, especially when accelerating, deaccelerating and turning, making calculation of position very difficult if not impossible. Slot cars are relatively light and scale tires do not provide a lot of traction.

Yes I am aware as I commented earlier:

I don't know how much traction the cars keep, and if this method could be accurate.

On the surface, a toothed rack and pinion would seem much better suited to such a measurement as you point out. But as I learned, there is slippage on the flip side, internally in the motor. the seals in a hydraulic motor are not (not designed to be) seals like in valve. When brand new, they're only capable of a true seal at pressures below about 35PSI, and hydraulic systems typically run 2000-4000PSI. They allow a lot of fluid to bypass, and the amount of bypass is not constant; it depends how much torque the motor is exerting. Some hydraulic motors even have a "case drain" - a separate dedicated line to bleed off this waste fluid. You could consider this your fluid equivalent of "burning out" or "drifting" or whatever you want to call it.

Despite that, I was able to tune the totalized flow feedback (position) scaling to be accurate within 0-3% error, because I was using scaling factors derived from empirical testing and not calculated from the displacement of the motor. Using real world data, the slippage (leakage) was factored in, and although it wan't constant, the end result was close enough. I suspect the car would be just as well suited to this method. Tire slippage will surely increase as the car makes a turn, but after completing a lap, you should be able to do the math and figure out the amount of error caused by slippage, and subtract that from the position reference.

The amount of error caused by slippage I assume is nonlinear, but directly related to speed and G-force. Using empirical data it should be totally possible to develop an equation (or at least a lookup table) that examines power output and g-force to determine actual speed (if the accelerometer doesn't provide actual speed already). From there, it's simple; distance= speed * time

 

I am up for anything but please remember the following.
-I have no idea about programming
-I have an excelent chance of sourcing the required parts and building a circuit off a diagram supplied
-I have an ok chance of gaining an understanding of what the circuit is doing
-I have little to no chance of being able to design the circuit.

Just to give everyone an idea of where my abilities are.

 

Well, many of the ideas here would require programming. Personally, I could not do it without programming.

 

There have been some really good but complicated solutions to this. What I am looking for is something that can be put together as a project by the average punter that can read a diagram, solder and owns a slot car track. Perhaps this is simply not possible but I wanted to give it a chance. Simplicity is the key here.