[itwasntme]

She has to literally and physically stand on the brakes to stop when in this situation. Very scary and lucky she is not in a lockup for running over pedestrians!

[Sutripta]

Originally Posted by DerAlte

Not sure I understand why it would be required, but no, Prof!

Depends on the philosophy of detecting a locked/ about to lock wheel. (The algorithm then follows the philosophy, backed by very sound mathematics.)

Originally Posted by Viju

Sure. I currently work for Continental Automotive Corporation, Japan as a Vehicle Test Engineer for Motorcycle ABS.

Great. Always wanted to know (in some detail) the differences in implementation of ABS on motorcycles, and on permanent 4x4 vehicles from the ordinary road car.

Is there any difference in the stopping distance with and without ABS (theoretical/ measured) when the vehicle is almost at a standstill, and brakes are applied?

Regards
Sutripta

[Viju]

Originally Posted by Sutripta

Is there any difference in the stopping distance with and without ABS (theoretical/ measured) when the vehicle is almost at a standstill, and brakes are applied?

Regards
Sutripta

Simple answer - No.

But actual result will depend on:
(1) Surface - if it is a high friction surface, stopping distance may be exactly same, where as on a low friction surface, stopping distance will most probably vary due to ABS intervention.

(2) Type of brake application - if it is a 'soft' application, stopping distance may be same, but if it is a 'hard' application, it may vary.

(3) Vehicle / brake configuration - again, depending on tyre characteristics (how easily they go into slip) and sensitivity of the brakes, chances of ABS control starting will vary.

Overall, considering the case of vehicle being almost at standstill, I would say, no, stopping distance will not vary.

[DerAlte]

Originally Posted by Sutripta

Depends on the philosophy of detecting a locked/ about to lock wheel. (The algorithm then follows the philosophy, backed by very sound mathematics.) ...

Sir, in ABS the context is limited to the wheel's rotational speed. Impending wheel lock cannot be deduced from the translational speed of the wheel, only the rotational speed.

'about to lock' is inferred (mathematically) from the rate of deceleration of wheel speed, where a heuristically determined value of deceleration serves as the threshold for that inference.

The speed of the vehicle to which the wheel is attached to is an incidental high level information, of interest only to humans - not to the ABS operational logic.

[mayankk]

I read that the sensors gauge relatively. If any wheel senses a skid under braking while any of the others are rotating, it kicks in.
In such a scenario, it's possible that abs fails when on, say, ice, all the wheels lock inside the window of abs pulses, that all wheels skid.
Is this a possibility?

[DerAlte]

Originally Posted by mayankk

I read that the sensors gauge relatively. If any wheel senses a skid under braking while any of the others are rotating, it kicks in. ...

The sensors don't gauge relatively - they sense only the wheel they are mounted on. Probably you mean the central ABS logic looks at data from the other wheels to influence the on-off ratio of the brake cycling of the wheel tending to lock. No, it doesn't. This logic is used by ESP to stabilize a vehicle, and the slipping wheel is braked.

Originally Posted by mayankk

In such a scenario, it's possible that abs fails when on, say, ice, all the wheels lock inside the window of abs pulses, that all wheels skid.
Is this a possibility?

No. There is nothing called "inside the window of ABS pulses". Since the wheels are monitored continuously, if any wheel seems to decelerate rapidly (well before wheel actually locks), the brake of that wheel is pulsed to relieve the braking action.

[Sutripta]

Originally Posted by DerAlte

Sir, in ABS the context is limited to the wheel's rotational speed. Impending wheel lock cannot be deduced from the translational speed of the wheel, only the rotational speed.

'about to lock' is inferred (mathematically) from the rate of deceleration of wheel speed, where a heuristically determined value of deceleration serves as the threshold for that inference.

The speed of the vehicle to which the wheel is attached to is an incidental high level information, of interest only to humans - not to the ABS operational logic.

Determining that a wheel hes locked/ about to lock in real time is easier said than done. This requires a multipronged approach.

Data from all four wheels is read in (obviously!) processed and filtered (filtered in the broader, and correct sense of the term.)
One of the intermediate values calculated is the vehicle speed. It is an estimate, not a precise hard figure, with its own associated confidence values. (Which is why I said 'estimated vehicle speed'). One of its uses is that if one knows the vehicle speed with confidence (and confidence increases with data), as a starting point, and as a cross check, one can check individual wheel speeds against it.
Of course it is not the only strategy. In fact it is a very minor part of deciding whether a wheel is locking up or not. But vehicle speed is computed.

What you are saying, if I have understood you right, is that the ABS algorithm never ever computes vehicle speed. Is that correct?

Regards
Sutripta

[im_srini]

Originally Posted by DerAlte

The lower limit is just an arbitrarily declared figure to put the monkey on someone else's shoulder.

Originally Posted by Thad E Ginathom

As I understand it, ABS operates just before wheel lockup, to prevent the wheels becoming locked.
It must work then, when the wheel rotation speed is almost zero.
I don't see any relevance in the original speed of the car at the moment that the breaks are applied.

Originally Posted by DerAlte

The key criteria is the rate at which the (rotating) wheel is slowing down.
The ABS ECU just calculates angular speed difference every few milliseconds.
If the deceleration rate is higher than a trigger point (which is a pointer to impending wheel lockup), ABS acts.
It has nothing to do with the speed of the vehicle.

Originally Posted by Sutripta

Determining that a wheel has locked / about to lock in real time is easier said than done.
In fact it is a very minor part of deciding whether a wheel is locking up or not.
But vehicle speed is computed.

If I understand this right, the ABS kicks in just when the wheel is about to lock, correct ?
That is, when it's angular velocity ( as determined from pulses from the reluctor rings on the individual wheels ) approaches 0.
In a very naive implementation, won't the ABS relieve brake pressure just as the wheel ( & the vehicle ) comes to a stop ?
So how do you decide when to stop ABS-ing, I mean, at some point you have to decide to stop ABS-ing otherwise you're never going to stop, correct ?
Is 'all wheels not rotating' a good enough indication that the vehicle has come to a stop ( seems an invalid assumption to me ).

Also, does the ABS motor actually modulate brake pressure ( as in threshold braking ) or is it that it just pulses ( relieves pressure & then pressurizes ) at some pre-defined frequency ?

[DerAlte]

Originally Posted by Sutripta

Determining that a wheel hes locked/ about to lock in real time is easier said than done. This requires a multipronged approach. ... if one knows the vehicle speed with confidence (and confidence increases with data), as a starting point, and as a cross check, one can check individual wheel speeds against it.
Of course it is not the only strategy. In fact it is a very minor part of deciding whether a wheel is locking up or not. But vehicle speed is computed.

What you are saying, if I have understood you right, is that the ABS algorithm never ever computes vehicle speed. Is that correct? ...

Sure, a lot of different approaches can be taken, especially ones that feel academically sound. What is actually implemented practically are simple plausibility checks - quite simply in the interest of keeping costs down. One would definitely use vehicle speed for comparison IF there are multiple point associated with vehicle speed for specific ABS behavior. That is, IF ABS behaves differently at different speeds. It doesn't.

With 4 signals coming in, AND plausibility check on each sensor being done continuously, one doesn't need to. Even ASR / ESP etc. are able to work without vehicle speed. To put it simply, these systems act iteratively, and cross checks action versus effect. Simpler methods prevent cascading of faults (KISS principle) - in this field liability cost is humongous!

Originally Posted by im_srini

If I understand this right, the ABS kicks in just when the wheel is about to lock, correct ?
That is, when it's angular velocity ( as determined from pulses from the reluctor rings on the individual wheels ) approaches 0. ...

No. "Approaching 0" is distinctly different from "has a deceleration profile that can result in wheel lock".

Originally Posted by im_srini

... In a very naive implementation, won't the ABS relieve brake pressure just as the wheel ( & the vehicle ) comes to a stop ?
So how do you decide when to stop ABS-ing, I mean, at some point you have to decide to stop ABS-ing otherwise you're never going to stop, correct ?
Is 'all wheels not rotating' a good enough indication that the vehicle has come to a stop ( seems an invalid assumption to me )....

No. It is continuously releasing pressure (release-hold-release-hold...) till that point. Well, it doesn't know that the wheel and the vehicle has come to a stop. Yes, "all 4 wheels coming to a stop" is a good enough indication (apart from the driver taking foot off the pedal) - with a couple of riders:
* All 4 wheels stopping could mean all 4 wheels locking, so it is not a reliable deduction method
* The braking action does result in changes in the deceleration rate (it is continuously computed), which indicates to the ABS that it is no longer required to act

Originally Posted by im_srini

... Also, does the ABS motor actually modulate brake pressure ( as in threshold braking ) or is it that it just pulses ( relieves pressure & then pressurizes ) at some pre-defined frequency ?

No 'motor' - it is just a set of electronically controlled hydraulic valves.

The latter. More precisely, relieves pressure then does nothing (the line pressure is still high as the brake pedal is still pressed). There is a predetermined frequency - but duty cycle changes based on wheel speed. As the wheel slows down, the hold period increases to improve braking. Difficult to figure out from the pedal pulsation, since circumstances are not conducive for accurate observation when ABS acts!

Originally Posted by Viju

... in relation to a reference velocity. ...

Missed this earlier. Is there a "reference velocity v/s wheel slip" map somewhere? The usual algorithms don't need any. For cars (I haven't checked for bikes), the knee points (at what d(WheelDeceleration)/dt to do what) is heuristically calculated and stored at track testing time - different for different cars - but nothing to do with slip and velocity.

[SS-Traveller]

Originally Posted by Sutripta

On another matter, below a (pretty low) threshold, I believe it's common to switch off ABS.

That's right. For your reference...

Originally Posted by kozhissery

ABS does not get activated in low speed. Anyway that what my Swift's User manual says

[DerAlte]

Originally Posted by SS-Traveller

That's right. For your reference...

What an understatement: "You may feel the brake pedal moves a little while the ABS is operating"! Should be more like "Scares the daylights out of people experiencing it the first time"

[im_srini]

Originally Posted by DerAlte

No 'motor' - it is just a set of electronically controlled hydraulic valves.

I'm not really sure about this, remember seeing an YouTube video where an ABS module was dismantled.
The cylindrical structure on the ABS module was a pump which through something akin to a scotch yoke acted on pistons in the 'block' ( rectangular end of the ABS module ).
The cylinders in which the pistons moved had valves at their ends, which of course, as you say were electrically actuated.
Besides, if the only pressurizing mechanism is the foot, then each time the ABS module releases pressure, won't the pedal too depress a bit ?
Let me see if I can find that video...

[DerAlte]

Originally Posted by im_srini

... The cylindrical structure on the ABS module was a pump which through something akin to a scotch yoke acted on pistons in the 'block' ...

The ABS oil pump. Keeps the pressure up when the valves act, yes. You are right, without it the pedal will gradually go down. The pump works only when there is positive pressure on the pedal, not all the time.

It is also associated with EBD (it is assumed that ladies can't stomp on the brake pedal hard enough) to improve line pressure when it is required - in this case brake pedal is pressed 'fast'.

[im_srini]

Originally Posted by DerAlte

The ABS oil pump.

By 'oil pump' I'm assuming you're referring to the 'brake fluid pump'.

Not exactly the video I was referring to, but close enough...

https://www.youtube.com/watch?v=IKi5O3ZlSu4

Another one, a bit longer...

https://www.youtube.com/watch?v=UUUbPo9pY1E

[DerAlte]

Originally Posted by im_srini

By 'oil pump' I'm assuming you're referring to the 'brake fluid pump'.

...

Yes, the brake fluid pump.