Drove a new Golf today for a while and used its cruise control for quite a way on the A12.
Whilst driving though I was thinking i've absolutly no idea how it works so could someone explain in great detail cruise controls innermost workings?
|
Happily, basic cruise control is really straightforward.
From the speedo, the system obtains an electrical signal that is in some way proportional to the speed of the car.
When you select a set point, you create a reference - the cruise control then subtracts the vehicle speed signal from the reference signal. This leaves you with an error signal.
If the car is going too slowly, the error signal is positive, and after passing through the controller, this error signal will act to increase the throttle opening.
If the car is going too quickly, the error signal is negative, and the controller will then reduce throttle opening.
Obviously, the really clever bit is inside the controller where various mathematical functions are applied to the error signal - typically some combination of proportional + integral action (I don't think there's any need for any derivative action on this type of controller, but I could be wrong). The design of controllers for closed loop systems is quite a deep, but extremely interesting subject area - one of my favourite areas of engineering. The abstract mathematics makes it particularly fun!
As this closed loop controller can review the vehicle speed very quickly, it can regulate the throttle opening, and control vehicle speed smoothly.
Outside of the closed loop, there is some fairly simple logic that shuts down the system and releases the throttle if the clutch or brake gets pressed, and another simple bit of logic that allows the reference point to be incremented by the drivers control.
Number_Cruncher
|
|
wOW! thanks very much for reply.
|
|
Cor ! I will never press the button again without thinking about that !!
|
In modern cars they are now an integral part of the cars can bus and computer network.
For example my cruise control takes its speed reference signal from the stability control system, which in turn takes its speed reference signal from the pulse sensors on the front wheels. If the stability program cuts in (as it did when I hit standing water with one wheel) the cruise control kicks out.
It is very accurate with its set speed. Less than 1 mph as checked by sat nav.
------------------------------
TourVanMan TM < Ex RF >
|
|
Same with the latest BMWs. Speedo and cruise control settings very close to Sat Nav read out. I read somewhere that CC was an option on the latest BMW bikes which also have CAN BUS wiring, that explains how they do it.. Having read up on it on my bike it is an incredible system.
|
|
|
|
|
auto.howstuffworks.com/cruise-control.htm/printable
|
|
Interesting, if rather jingoisitic, site. Prefer an euro-box which handles like something though.
|
|
|
|
|
|
I've just had a quick play, simulating a very basic cruise control system, and proportional plus integral was OK, so I think I'm on fairly safe ground to say that there isn't any need for differential action.
Proportional control is just where the controller output is simply the error signal scaled by some gain. So, if the error is 10 mph, open the throttle by X, if the error is 20 mph, open the throttle by 2X, etc, etc.
Integral control is a bit more complex. It increases the control action the longer that the error signal exists. It is used to get rid of what are termed steady state errors. Say your car encouters a hill and its speed drops - the proportional controller on its own will never apply quite enough control action to keep your car's speed constant - it will drop briefly. But as the integral of the error signal (or if you think of it graphically, the area underneath the error signal) builds up over time, the integral action adds that little bit extra to keep your set speed.
Proportional + Integral control is simply a controller that takes the error signal, acts on it in both ways, and sums the output.
Number_Cruncher
|
|
You'd get on with my wife.
|
Cruise control was available on Honda Goldwings and Heavy Davidsons for years and they don't have can-bus or I-bus or number 15 bus.
I don't know if the £50 retro fit kits use any differential mathematilistic witchcraft. They usually consist of a speed sensor thats attached to the propshaft and a mechanical actuator arm that attaches to the throttle linkage. The usual clutch and brake safety cutout switches are present also, but of course it doesn't tie into any traction control devices so if you hit puddles, ice, diesel etc....hang on.
|
Cruise control was available on Fiats years ago as well, My dads 124 i think.
It was a knob in the dash with a length of bowden cable to the throttle lever.
------------------------------
TourVanMan TM < Ex RF >
|
|
|
what about the cruise control on a early 1970's LWB L/Rover.
It was a ratcheted lever that pulled from left to right with brute force and left the engine racing as you changed gear or dipped clutch. My how things have changed!
When patroling the airfield at RAF Laarbruch in Germany, you could set it on and drift around the perimeter, double de-clutching from 3rd to 2nd and dipping the clutch to reduce speed when turning at each corner.
|
|
Surely you need the differential part to measure acceleration? Isn't it called PID (proportional-integral-differential) control for that very reason? ;)
|
Surely you need the differential part to measure acceleration? Isn't it called PID (proportional-integral-differential) control for that very reason? ;)
Although you can have all three terms active, you don't always **need** to, it depends upon the dynamics of the system you are controlling.
Some systems can be controlled perfectly well just using proportional control. Typically, these are where you don't have large disturbances, and you aren't too bothered about the system following the set point accurately.
You might begin to use integral control if you want to get rid of any steady state errors, and/or if your system is subject to wildly vaying disturbances.
The problem with integral action is that it can de-stabilise a system. This can be countered with differential control action - BUT, the problem with differential control is that when you look at a real signal, it has noise. A differential controller will produce a large control signal from these small spikes of noise (as opposed to integral action which effectively smooths or filters the error signal); it is this sensitivity to noise in the signal which makes differential action a tool best avoided if possible.
You can think of a general PID controller as an amplifer with three "volume" controls, which you can tune by twiddling the controls ,P, I, and D. In many cases, the knob marked D is best left turned right down!
There are tuning techniques, most commonly based on the so-called Ziegler-Nicholls method which allow you to make some tests of the system dynamics, and then give you suggestions for setting the three gain terms on a PID controller to give "good" system response.
David, if you have access to Simulink in your department, I can send you a simple model to demonstrate.
In short, no, I don't think you **need** to measure acceleration in this application.
Cheers,
Number_Cruncher
|
>>In short, no, I don't think you **need** to measure acceleration in this application.
Having thought a bit more about this, I now think that even the integral action isn't strictly needed, although it might be desirable to reduce steady state error.
The reason for this is that the albeit simple system I have considered is unconditionally stable, so, simply by setting a sufficiently large value for the proprtional gain, reasonably good control can be achieved.
In control terms, the plant is simply a pole at the origin, and therefore, the root locus exists only on the negative real axis, hence the transient response is simply a decaying exponential, hence, unconditionally stable.
(In terms of feedback that we've all experienced, this is a bit like saying the microphone is a long,long way away, and hence shielded from the speakers, so we can turn up the volume without the system squealing!)
Number_Cruncher
|
|
|
|
what about the cruise control on a early 1970's LWB L/Rover.
That was not a cruise control, it was a hand throttle! Poor landy! You naughty rock-ape you!
|
|
The cruise control on both my Hondas (92 and 98) appears to work by working the accelerator via a cable from a small vacuum servo. Very effective on both.
|
|
I had cruise control on a Citroen GS. The hot pedal used to get stuck now and again. Needed some oil but was fun while it lasted.
|
I did'nt realise my car had cruise control,just been out to have
a look-could not find any switches, anyone got an idea?
1962 Austin A60 Cambridge.
|
|
|
|
|
|
|
|
