Detecting Reverse rotation of the Wheel 1

[santorta]

Hi,

How can I detect the reverse rotation of the Car wheel? I know that vehicles have VRS (Variable reluctance sensor) or Hall sensor for wheel speed sensing. I think, the voltage from the sensor will be the same, for both directions(since the trigger wheel in shaft will cut the flux). Is there any other way say like a algorithm to identify this? for eg, by looking at the gear position and the clutch pedal input, or with brake pedal input.

Looing forward for a discussion...

Bye,
Santo RTA

 

[Thecardoc]

The wheel speed sensors won't be of any help. Because of the design, the voltage rises above 0v when the tooth approaches the sensor, then drops as the tooth leaves the sensor, and it looks the same in either direction.

The newer sensor actually has a hall effect transistor inside it, so all it does is produce a square wave. The best part about those sensors, is the engineers achieved that trick while still only using two wires!

I have some limited ideas of what you may be able to do in a vehicle specific case, but before I type all of them out, what are you trying to create?

 

[GregLocock]

You could vary the pitch of the teeth on the toothed wheel, either ramp the pitch over one rev, or say split it into three sections with different pitches. You could do the same with missing teeth, but I don't like that.



Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[MikeHalloran]

The 'classical' solution would require two toothed wheels, staggered a little, two sensors, and a 'quadrature detector', which outputs a pulse train and a binary direction signal.

Greg's solution is very clever, and could work if you can stand the time/phase delay and inability to detect oscillation as opposed to rotation.





Mike Halloran
Pembroke Pines, FL, USA

 

[MacGyverS2000]

I see no reason to use two wheels, just a properly staggered set of sensors would work just fine.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[reidh]

Two sensors with one trigger point would work quite well.

-Reidh

 

[MikeHalloran]

I'm sure that two sensors on one toothed wheel would work... for optical sensors.

For magnetic sensors and an iron wheel, I'd worry the sensors might get confused.



Mike Halloran
Pembroke Pines, FL, USA

 

[Warpspeed]

This is done all the time with optical shaft encoders, Teeth of a fixed pitch and dual sensors phased at a quarter of a tooth pitch.

The sensor outputs will follow each other, with one leading the other lagging in one direction, and the opposite lag/lead output with reverse rotation.

What you then need is a "D" type flip flop:

http://en.wikipedia.org/wiki/Flip-flop_(electronics)

Connect one sensor output to the "D" (data) input, and the other sensor output to the "C" (clocking) input.

The flip flop will then remain continuously in one state for one direction of rotation, and change to the opposite state for reverse rotation.

The existing ABS tone wheels could be used by fitting a second sensor to each wheel displaced by a quarter tooth pitch.

The flip flop will respond within one tooth pitch of motion to a direction change.

This is a very commonly used technique in robotics and similar applications, and is guaranteed to work.

 

[Warpspeed]

I don't know what happened to my above link, but try this one:

http://en.wikipedia.org/wiki/Flip-flop_(electronics)#D_flip-flop

 

[santorta]

The idea of the toothed wheel is good but the catch is that, in worst cases the reverse rotation of the wheel will be identified only after a minimum of one rev (to identify the missing pulse or varied pitch teeth). This means that the wheel would have moved in reverse by a 7ft (say 30" wheel dia, wud have circumference of 7.85ft).

My requirement is that the reverse motion should be identified within 1ft or 12". since this is a reverse motion(slow speed) occurs after the forward motion, the pulses from the trigger wheel will be very weak and difficult to process the signal for identifying the reverse rotation.

In case of optical shaft encoders, Since it is planned to mount on the wheel axle, i really doubt the durability of optical shaft encoders. will it really identify quickly at any worst environmental condition (say during bumps, vibrations etc..). What about the cost of the optical sensor.

The other idea that i was thinking, can we assume by an algorithm, say that the wheel speed, when the clutch is pressed and when gear is in neutral, and when the previous value of the wheel speed was decreasing to zero, and again increase then we can identify that as reverse rotation. This is just a thought... would like to refine it based on the discussion...

Let me know if i have made myself clear about the algorithm...

 

[Warpspeed]

Read again what I said about using the existing ABS tone wheel.

A typical tone wheel might have 40 teeth, and a typical wheel might have a diameter of perhaps 24 inches. So one tooth is slightly less than two inches of vehicle travel.

This system would be capable of detecting reverse rotation of less than two inches.

 

[GregLocock]

Well, thanks for not bothering to spec your system in your first post. Way to guarantee participation.

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[sreid]

1) Hall Effect Magnetic sensors are not speed sensitive. The sensor switches for "Iron Close, "Iron NOT Close."

2) Two sensors are required spaced as Warpspeed says at a quarter tooth pitch apart. If one calls the two signals A and B, for one direction of rotation the rise of A will lead the rise of B. For the opposite direction, B will lead A.

3) The distance to determine the rotation direction is based on the number of teeth; the more teeth, the shorter the distance.

 

[Warpspeed]

Excellent sreid.

Another minor clarification. The quarter tooth pitch phasing does not have to be on the same tooth.

The sensors could be located diametrically opposite, or physically located anywhere around the wheel. but they must be accurately timed so one has a quarter tooth pitch lead ahead of the other.

 

[Thecardoc]

Hey Mike Halloran.

I'll be at the Herald building in Pembroke Pines on February 5th teaching a basic brakes course. Small world...

Got to admitt some pretty sharp people here. Placing a second sensor, just slightly offset from the factory one would do the trick IMO. No small feat when it comes to doing the modification and prepping the install location, but once measured out it should work quite nicely.

 

[MikeHalloran]

Sorry I won't be there to say hi; I'm working in New Orleans.



Mike Halloran
Pembroke Pines, FL, USA

 

[ot1]

Consider the simple solution of using a G-force sensor, if you just need to know reverse direction for the moment.

 

[carnage1]

I have to say thanks to warpspeed for such a great solution, I had been trying to do a similar thing in a cnc machine and all my ideas were much more complicated and less robust.

 

[Ron07663]

You can obtain the two hall sensors and latching flip-flop packaged in one 2 wire chip. See Allegro Microsystems Direction sensing sensor chip ATS651LSH.

http://www.allegromicro.com/en/Products/Part_Numbers/0651/index.asp