bilbo
|
| posted on 26/2/08 at 05:37 PM |
|
|
DIY ABS
OK, following on from this thread:
http://www.locostbuilders.co.uk/viewthread.php?tid=83837
How difficult would it be to develop a DIY Anti-Lock braking system?
What I'm thinking of is something like the megasquirt/megajolt ethos, but for ABS - 'MegaStop' if you will.
Now I don't see the electronics side of things to be too difficult. Sensors as well should be fairly easy - some sort of trigger wheel on each
wheel perhaps?
But it's the hydraulics side of things that I guess would be harder. It would be nice to have something inline between the M/C and calipers that
can be used with otherwise non ABS components (race calipers, old M/Cs and the like). Perhaps an existing ABS unit from a production car could be used
if the wiring can be worked out?
Just thinking out loud really. I'd be very interested in peoples views.
---------------------------------------
Build Diary: http://bills-locost.blogspot.com/
Web Site: http://locost.atspace.com
|
|
|
|
|
caber
|
| posted on 26/2/08 at 05:48 PM |
|
|
A few years back people said it would be impossible to do modified electronics for EFI systems. We now have the megasquirt , GEMS, etc. etc. No reason
why the same can't be done for brakes except the liability risks are higher for a car that won't stop compared to a car that won't
go or blows up its engine 
Caber
|
|
|
Guinness
|
| posted on 26/2/08 at 05:55 PM |
|
|
I visited WABCO 's factory in Leeds a while back. They produce loads of ABS systems for major manufacturers.
Linky to a PDF:
http://www.wabco.info/intl/pdf/815/00/01/8150100013.pdf?PHPSESSID=b329bf2f7bef7b3c6a47032d933984b2
Linky to their website.
http://www.wabco-auto.com/products_and_systems/products-systems-anti-lock-braking-system-anti-spin-regulation-absasr
The PDF has some interesting information about how the thing works!
HTH
Mike
|
|
|
smart51
|
| posted on 26/2/08 at 06:06 PM |
|
|
Its not the making of the ECU that is the problem, it is the tuning of the system.
Engines are simple. You have a throttle position and an engine speed. You tune air / fuel ratios and spark timing. A rolling road provides all the
environment you need to test.
For ABS tesing, you need wet ice, dry tarmac, gravel ... developing algorythms that recognise the surface and adapt is tricky. For dry grippy
surfaces you need parge pulses to increase the brake pressure quickly to get the maximum braking effort. on slippery surfaces, you need small
increases so that you don't jump straight through the band of grip.
What about split mu surfaces and patchy surfaces such as ice in the shadows of trees but dry spots in between?
Then you have to test at all speeds vs all cornering speeds, combatting lift off oversteer, power oversteer...
The facilities for tuning DIY ABS just arent available to the amature builder in the same way as engine management.
|
|
|
Volvorsport
|
| posted on 26/2/08 at 06:08 PM |
|
|
you also need to be able to program for different co efficients of friction , and weight transfer .
ive heard people can re program the bosch units .
the volvo ones are all on a seperate loom , i thought about it , but it would be much easier to regulate the brake pressure with your right foot !
www.dbsmotorsport.co.uk
getting dirty under a bus
|
|
|
matt_claydon
|
| posted on 26/2/08 at 06:27 PM |
|
|
It's all academic really, there would never be a requirement for ABS on amateur-built vehicles. The current SVA foregoes loads of things that
are compulsory on mass-produced vehicles and there's no reason that will change with IVA.
|
|
|
skodaman
|
| posted on 26/2/08 at 06:38 PM |
|
|
''It's all academic really, there would never be a requirement for ABS on amateur-built vehicles.''
I just hope you're right Matt or a few of us on here are going to be scuppered.
Skodaman
|
|
|
bilbo
|
| posted on 26/2/08 at 06:54 PM |
|
|
quote:
Originally posted by smart51
Its not the making of the ECU that is the problem, it is the tuning of the system.
Engines are simple. You have a throttle position and an engine speed. You tune air / fuel ratios and spark timing. A rolling road provides all the
environment you need to test.
For ABS tesing, you need wet ice, dry tarmac, gravel ... developing algorythms that recognise the surface and adapt is tricky. For dry grippy
surfaces you need parge pulses to increase the brake pressure quickly to get the maximum braking effort. on slippery surfaces, you need small
increases so that you don't jump straight through the band of grip.
What about split mu surfaces and patchy surfaces such as ice in the shadows of trees but dry spots in between?
Then you have to test at all speeds vs all cornering speeds, combatting lift off oversteer, power oversteer...
The facilities for tuning DIY ABS just arent available to the amature builder in the same way as engine management.
Would it really need to be this sophisticated? Earlier production systems never did any of this did they?
---------------------------------------
Build Diary: http://bills-locost.blogspot.com/
Web Site: http://locost.atspace.com
|
|
|
matt_claydon
|
| posted on 26/2/08 at 07:10 PM |
|
|
quote:
Would it really need to be this sophisticated? Earlier production systems never did any of this did they?
No, there are some quite basic systems around that really just compare wheel speeds and release pressure to any wheel that differs from another by a
certain amount. London cabs have a pretty crude system IIRC.
ABS isn't even a requirement at the moment in European Whole Vehicle approval as far as I remember, although there has been a
"gentlemen's' agreement" between car manufacturers that they will fit it universally for a few years.
|
|
|
Puk
|
| posted on 26/2/08 at 07:21 PM |
|
|
The early audi systems used to have an off switch, for use on snow (AFAICR = probly). So maybe a simplified system could be devised.
And if it doesn't work very well - switch it to snow mode.
Realistically I can't see any gifted laymen doing it alone though - we'd want some input from someone that had done it before.
Does anyone know how ABS is tested in an MOT?
[Edited on 26/2/08 by Puk]
Before you judge a guy, walk a mile in his shoes. Then when you judge him, you're a mile a way and you've nicked his shoes.
|
|
|
turbodisplay
|
| posted on 26/2/08 at 08:01 PM |
|
|
I have thought about this - i have someting in the pipeline that will address this. Can`t say until it is patented, if i works that is.
To do a diy system i would take an abs unit either an all in one, bosch, or a kesley hayes unit, take the controller that bolts to the valve block,
laser cut a fitting to join this to the controller.
The early systems used discrete electronics, or very crude microcontrollers, i believe, so cannot be that complex program wise.
For sports cars just having a device that liits slip to 10% dry, 5% wet would be a good comprimise.
The controller would require 2, 3 or 4 processors to allow for safety, any in consistency would result in the system being turned off by a
"watchdog processor".
I would say best bet would be to use a system from a porsche, or rx8 or smilar sports cars which are optimised for maximium braking, not for control
in the wet at the expense of dry and wet brake force. If you need such "muppet safety" then driving a seven style car is probally not for
you!
Another posbility is electronic brake force distribution, as this would still allow some pressure get to the wheels even in a fault (if designed
well).
Who can program bosch units as that would be a good soloution?
I`m using a vectra abs unit as it is very simple to wire, about 10 wires. The later units are ment to be the better as they allow more slip before
cutting in. I can testify mid corner wet braking is excellent in the vectra. But this unit needs to be mounted in the correct orientation as it has a
g sensor inside.
ABS UNITS with a g sensor must be mounted as they are in the car they come from, to prevent them getting confused/ going into LOS mode.
Darren
|
NOTE:This user is registered as a LocostBuilders trader and may offer commercial services to other users
|
Delinquent
|
| posted on 26/2/08 at 08:08 PM |
|
|
quote:
Originally posted by matt_claydon
quote:
Would it really need to be this sophisticated? Earlier production systems never did any of this did they?
No, there are some quite basic systems around that really just compare wheel speeds and release pressure to any wheel that differs from another by a
certain amount. London cabs have a pretty crude system IIRC.
Even in it's most basic form it's still going to out perform 99% of drivers as well, I think a basic "if wheel is locked
pulse" rule would be more than adequate. I imagine Implementation should be fairly simple as well using the major components from an ABS donor
with some careful selection.
|
|
|
tks
|
| posted on 26/2/08 at 09:28 PM |
|
|
quote:
Originally posted by Delinquent
quote:
Originally posted by matt_claydon
quote:
Would it really need to be this sophisticated? Earlier production systems never did any of this did they?
No, there are some quite basic systems around that really just compare wheel speeds and release pressure to any wheel that differs from another by a
certain amount. London cabs have a pretty crude system IIRC.
Even in it's most basic form it's still going to out perform 99% of drivers as well, I think a basic "if wheel is locked
pulse" rule would be more than adequate. I imagine Implementation should be fairly simple as well using the major components from an ABS donor
with some careful selection.
Yupsz... i have the same thoughts to...
its just a matter of releasing the pressure on the axle/wheel that locks..
and if that wheel locks again after releasing than release again...
that will generate the shaken pedal and that will be what we want.
the only thing you want is a speed sensor (from the back axle sensor) sow that you know the car is loosing speed and that its speed is above 5km/h
etc. etc..
i remember our omega stall its wheels if it was to slippery and if the speed was very low.
1 microcontroller will be enough.
you realy don't need 4!
if you use the normal abs sensors then you will need 4 opamps
4 zener diodes
1 microcontroller and some small things..
also proberly a fast output stage for the abs unit. a Fet and driver will be needed...
the hard part will be the hydraulics...
i think they use a a valve block wich has a preset pressure setted with some springloadings.
if you then open the valve the pedal is pushed back (by the hydraulics) and with the force of the spring load.
if you close the valve the pedal is free again.
in this way the normal brake circuit is untouched and will still work unwilling a failure in your abs unit. (the stroke will only be limited by couple
of mm)
Tks
[Edited on 26/2/08 by tks]
The above comments are always meant to be from the above persons perspective.
|
|
|
bilbo
|
| posted on 26/2/08 at 10:00 PM |
|
|
Some really interesting replies here. It's clear that some sort of simple system should be possible, and that I am not the first person to think
about this.
What's really heartening, though, is that whatever this IVA ends up throwing at us, there are plenty of clever people out there in the amateur
car building scene ready to work out DIY solutions
---------------------------------------
Build Diary: http://bills-locost.blogspot.com/
Web Site: http://locost.atspace.com
|
|
|
turbodisplay
|
| posted on 26/2/08 at 11:13 PM |
|
|
1 micro controller would be enought but for safety a min of 2 is required to "watch each other" incase of software or hardware error. It
even possible for cosmic radiation to change bits inside a processor (unlikely but electrical noise can be an issue even on a well built system).
Vr sensor chips do exist which will do the signal processing with hysterisis although op amps can be used with the appropiate configuration.
Darren
|
NOTE:This user is registered as a LocostBuilders trader and may offer commercial services to other users
|
Delinquent
|
| posted on 27/2/08 at 10:01 AM |
|
|
quote:
Originally posted by tks
Yupsz... i have the same thoughts to...
its just a matter of releasing the pressure on the axle/wheel that locks..
and if that wheel locks again after releasing than release again...
that will generate the shaken pedal and that will be what we want.
the only thing you want is a speed sensor (from the back axle sensor) sow that you know the car is loosing speed and that its speed is above 5km/h
etc. etc..
i remember our omega stall its wheels if it was to slippery and if the speed was very low.
1 microcontroller will be enough.
you realy don't need 4!
if you use the normal abs sensors then you will need 4 opamps
4 zener diodes
1 microcontroller and some small things..
also proberly a fast output stage for the abs unit. a Fet and driver will be needed...
the hard part will be the hydraulics...
i think they use a a valve block wich has a preset pressure setted with some springloadings.
if you then open the valve the pedal is pushed back (by the hydraulics) and with the force of the spring load.
if you close the valve the pedal is free again.
in this way the normal brake circuit is untouched and will still work unwilling a failure in your abs unit. (the stroke will only be limited by couple
of mm)
Tks
[Edited on 26/2/08 by tks]
I've not looked into the hydraulics yet, but would hope an existing system could be adapted relatively easily.
|
|
|
turbodisplay
|
| posted on 27/2/08 at 05:04 PM |
|
|
Hydralics choice would be a seperate valve block, seirra, early bosch, or adapt an all in one valve block and mount to own controller.
One interesting problem with abs units is that they have a problem with the valves breaking, due to being made just big enough for the aplication
(faster on/ off cycling). It would be a good idea to have an override bypass in case of valve failure.
Darren
|
NOTE:This user is registered as a LocostBuilders trader and may offer commercial services to other users
|
smart51
|
| posted on 27/2/08 at 06:42 PM |
|
|
ABS valve blocks for cars have 2 valves per circuit. 1 isolates the wheel from the braking system when activated. The other dumps the brake fluid
from the wheel to an accumulator. A pump pumps the fluid from the accumulator to the master cylinder, which is why you feel the pedal move up. The
ABS pulses the isolating valve, letting small amounts of fluid back into the wheel, which is why you feel the pedal go back down again. It cycles
until you find grip or stop. It would be fairly easy to make an ecu that did this. The clever bit is the algorythm.
You need to monitor the wheel speeds carefully to decide when a wheel is about to lock up based on a maximum deceleration rate and differential speeds
of each wheel (eg fronts are 20 MPH slower than the backs). You then need to decide how much fluid to drop out of the offending wheel(s). Not too
much or you lose braking effort. Not too little or your wheel stays locked.
When the wheel is turning again, you need to decide how soon to start reapplying brake pressure, how much to increase it and how quickly. Pulse the
valve to add pressure little by little until you see the wheel start to lock up again. Decide whether to hold that pressure or try your luck and add
a bit more.
If you wanted a very simplistic system, say just to be better than completley locking up, you could wait till the wheel is almost stationary, but at
least 1 wheel is still turning at a reasonable speed, dump most or all of the fluid, wait till the wheel speed is OK, then pulse the fluid back in,
repeating if the wheel locks up again. You have to remember that if all the wheels are less than 3 mph then the car is probably almost stopped.
Don't turn off the brakes.
Such a system is probably better than nothing. Especially if only 1 wheel is locked. The other 3 wheels are left with full brake force, compared
with cadence braking when you lift your foot off all the wheels at once.
I wouldn't want to drive such a system, though, until it was well proven.
edit to say that you need to watch the duty cycle of the solenoids, in case you burn them out.
Also, 2 microprocessors is a good idea. The minimum system has a tiny second micro that challenges the first and expects the correct answer. If it
doesn't get it, it resets the main micro. Of course, the main micro fails safe, so that all the ABS valves are left in their "foundation
brake" state.
[Edited on 27-2-2008 by smart51]
|
|
|
tks
|
| posted on 27/2/08 at 07:40 PM |
|
|
All the common used CPU have watchdog timers in them. They automaticly reset the cpu if it hangs....
Also you don't want a 2nd chip reset the first one because who says the 2nd chip is in good schape after some cosmos mods??? Or that the fault
in the first doesn't give you eratic behaviour?? They both will need to check each other.
1chip if it works the master relais is activated and the solenoid can be pulsed if it doesn't the relais is braked and normal brakes are
gathered.
the 2nd chip could have this simple function. if the 2nd brakes system lights ABS light. if the first brakes the system is disconnected and the light
lights up to .
In that way you will be safe.
I'm OK with everyone that we will bee a ready fitted item for example a bosch component to wich we mount our PCB etc.
in that way we will have the quality and the E marking and safety from normal production cars.
Tks
The above comments are always meant to be from the above persons perspective.
|
|
|
turbodisplay
|
| posted on 28/2/08 at 08:28 PM |
|
|
Watchdog timers are useful - unless the program is looping at the point designed to reset the watchdog.
I`ve looked into microcontroller reliability as i`m putting a traction control system together. That isn`t as important as ABS , but will have 3
processors, one safety, one to monitor the wheelsspeeds, one to calculate engine cut.
I was beng serious about radiation and electrical noise. Background radiation has and can cause errors, a good example is radiation hardened
processors (costing many £££). For use in space with hardware error checking, along with software protection is expected to have an error once in 10
years!
Darren
|
NOTE:This user is registered as a LocostBuilders trader and may offer commercial services to other users
|
tks
|
| posted on 29/2/08 at 06:51 PM |
|
|
quote:
Originally posted by turbodisplay
Watchdog timers are useful - unless the program is looping at the point designed to reset the watchdog.
I`ve looked into microcontroller reliability as i`m putting a traction control system together. That isn`t as important as ABS , but will have 3
processors, one safety, one to monitor the wheelsspeeds, one to calculate engine cut.
I was beng serious about radiation and electrical noise. Background radiation has and can cause errors, a good example is radiation hardened
processors (costing many £££). For use in space with hardware error checking, along with software protection is expected to have an error once in 10
years!
Darren
one to calculate the engine cut??
You should put that in a lookup table my friend... in that way there is now communication between the 2 chips and the failing probility is
halved...
Tks
The above comments are always meant to be from the above persons perspective.
|
|
|
