Drive by Wire - Who Would Use it? [Archive]

MCoach

06-10-2012, 06:22 PM

It's a new concept that I think will become like the aero package in judging. The questions went from: Why do you run aero, can you prove it is effective? Oh, you don't? Ok, that's fine.
to: Why doesn't your team, run aero? Can you prove that it isn't worth your time through simulation of an aero design you've worked on? Once the cool kids start running it, the view on the point will change.Experimenting with it will be cool. There are quite a few things that can be done with it, one article I read recently I will link for you here about Formula 1.

http ://www .racecar-engineering. com/articles/f1/formula-1-2012-round-7-canadian-grand-prix/

This is about the formula 1 race this weekend and most of it focuses on how tight of line they walk when tuning it. Part of it focuses on setting up the car with their electronic throttle. which can be seen in this video at 1:43.

http ://www .youtube. com/watch?feature=player_embedded&v=_Wn1EFLa2C8

Drive by wire would be something to look up some SAE papers on if you are really interested or doing some serious research. The electric SAE teams, as well as the aforementioned snowmobile team, would be good references for gremlins to look out for.

My challenge to you would be to generate ideas on what the integration of that system would take. What would be the benefits of it over a cable actuation system? What could it allow you to do that a cable system wouldn't allow you to do? What problems would it cause over the cable system? Cost of cable vs cost of DBW?

These are things that a judge will be looking for if you pursue (if it is allowed) such a system.

BillCobb

06-10-2012, 06:51 PM

I would like to recommend that you NOT refer to 'throttle by wire' as 'drive-by wire' because its NOT the same control system or theory of integration. Throttle by wire is already in production in passenger cars, with 'steer-by-wire' in existance and awaiting government approval in several countries. Then there is 'brake-by wire'.

When you fly in commercial airlines, all these systems are in operation. Usage in FSAE (IMHO) is a tremendous opportunity to demonstrate modern engineering prowess and mechanization of directional control, redundant safety, packaging, reliability, and haptic sensing subjects (just to name a few).

Welcome to the New World, Captain Rameous...

Cardriverx

06-10-2012, 10:18 PM

They refer to it as drive by wire in the rules, so I did here. But I do understand your point.

Mcoach, you made the great points and provided good info, thanks.

Originally posted by BillCobb:
I would like to recommend that you NOT refer to 'throttle by wire' as 'drive-by wire' because its NOT the same control system or theory of integration. Throttle by wire is already in production in passenger cars, with 'steer-by-wire' in existance and awaiting government approval in several countries. Then there is 'brake-by wire'.

When you fly in commercial airlines, all these systems are in operation. Usage in FSAE (IMHO) is a tremendous opportunity to demonstrate modern engineering prowess and mechanization of directional control, redundant safety, packaging, reliability, and haptic sensing subjects (just to name a few).

Welcome to the New World, Captain Rameous...

MCoach

06-10-2012, 11:01 PM

Originally posted by BillCobb:
I would like to recommend that you NOT refer to 'throttle by wire' as 'drive-by wire' because its NOT the same control system or theory of integration.

Bill, thanks for pointing that out. It is good to note that they are NOT the same, and it may be something key to watch if the same terminology is used if/when rules are published concerning each technology.

Cardriverx,

Did you have any current ideas that you would like to pursue besides getting rid of that pesky throttle cable?

Cardriverx

06-11-2012, 09:09 AM

There are five main reasons I want throttle-by-wire, which I listed roughly in order of importance. I haven't put too much thought into it yet, so I am sure there is more I would want to do with it.

Easy custom throttle profiles for autox/skidpad/accel.

Better secondary rev limit that uses the throttle-by-wire.

Downshift bliping with our air shifter.

Integrating traction control with the throttle-by-wire.

Idle control using the throttle-by-wire.

Originally posted by MCoach:

Bill, thanks for pointing that out. It is good to note that they are NOT the same, and it may be something key to watch if the same terminology is used if/when rules are published concerning each technology.

Cardriverx,

Did you have any current ideas that you would like to pursue besides getting rid of that pesky throttle cable?

Mbirt

06-11-2012, 10:55 AM

Idle control using the throttle-by-wire. That thought had not occurred to us, dude! This is a great reason for TBW, thanks for mentioning it.

jlangholzj

06-11-2012, 05:45 PM

Originally posted by Mbirt:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Idle control using the throttle-by-wire. That thought had not occurred to us, dude! This is a great reason for TBW, thanks for mentioning it. </div></BLOCKQUOTE>

with current rules i see no reason why you couldn't do this already?

Menisk

06-14-2012, 06:33 AM

Another two that haven't been mentioned:

<UL TYPE=SQUARE> <LI> Anti-lagging turbo's
<LI> Exhaust blown aero. (I know jack shit about aero and whether or not you'd get enough out of our engines to actually gain anything) [/list]

MCoach

06-14-2012, 04:21 PM

Originally posted by Menisk:
Another two that haven't been mentioned:

<UL TYPE=SQUARE> <LI> Anti-lagging turbo's
<LI> Exhaust blown aero. (I know jack shit about aero and whether or not you'd get enough out of our engines to actually gain anything) [/list]

YES!!

I was waiting to comment until I saw someone at least mention this area of possibilities. An anti-lag map and a blown aero map aim to achieve the same end result of flowing more air. From what I've seen from data, closing the areo performance gap between entry corner speed and .75seconds into accelerating down the straights is somewhere worth gaining nearly a full second through this. The extra down force gained through blown aero (especially a diffuser) is something not to scoff at, even at our speeds.

exFSAE

06-14-2012, 05:29 PM

FSAE teams have enough trouble getting relatively well-established electrical systems working, much less adding throttle by wire to the equation... for what is IMO a trivially small gain.

Much better things to spend the development time on, like not having suspension compliance rates equivalent to al dente spaghetthi.

Idle control, TCS, launch rev limiters... all quite practical to do within a standard ECU without adding the extra complexity.

Different throttle curves - do you need this? Why? Sure there are different throttle needs for skidpad and autocross but you can just as easily put that in your cam design and have a pretty soft, easy to modulate curve for part throttle through sweepers and skidpad and then a faster rate at the high end for course events.

Then once you're actually out on track for endurance... I don't see any advantage over a push-pull cable.

Menisk

06-14-2012, 10:55 PM

Thing you need to remember is that not all teams are composed entirely of Mechanical Engineers. There's Electrical and Mechatronic guys in this comp and it'd be nice to see something like throttle by wire that opens up a heap of development opportunities for us as listed above. While you do need to think about where to put your development time on a team you need to think about the skills you have on your team. There's no point ignoring electrical and mechatronic opportunities on your car if you've got a bunch of keen mechatronic and electrical guys who are going to be much more productive in their own field.

jrickert

06-14-2012, 11:28 PM

Idle control and blip would be nice.

Menisk

06-14-2012, 11:30 PM

Just realised you could also use it to have adjustable engine braking. Rather than having the throttle fully closed in overrun you could run with it partly to fully open with a fuel cut to adjust the amount of braking to whatever the drivers find easiest.

MCoach

06-15-2012, 12:03 AM

Originally posted by exFSAE:
FSAE teams have enough trouble getting relatively well-established electrical systems working, much less adding throttle by wire to the equation... for what is IMO a trivially small gain.
--
I would disagree. It depends on your aptitude as a team to such systems. As a counter point, some teams have trouble welding a frame together with access to only MIG or stick welders and still have relatively advanced electronics on their car. I would reference several of the up and coming Chinese cars as example.

Much better things to spend the development time on, like not having suspension compliance rates equivalent to al dente spaghetthi.
--
What if you've worked out these problems through a 'legacy design' by revising only key systems over quite a few years with a great deal of knowledge transfer and only fine tuning required? I would point out that many of the German cars don't change a whole lot from year to year. Missouri S&T and RIT also come to mind. Teams in this position would stand to gain the most from having a new area to expand into beyond the now standard approach of investigating an aero package.

Idle control, TCS, launch rev limiters... all quite practical to do within a standard ECU without adding the extra complexity.
--
This allows just one more system to better control these systems. Take this scenario for example. Normally a '2 step' launch control style is used where a set RPM is held and timing becomes retarded to keep it at that same rpm even when more throttle is applied. In the case of DBW, the driver would engage the launch control, put the pedal to the floor and the throttle would open to the angle needed to hold the proper RPM without retarding ignition. For some, it's not about the complexity of just one more electronic object, it's about the precision of execution. This one item allows all of those things that you mention, while currently achievable with the current cable actuated throttle, but to a higher degree of precision and tuning.

Different throttle curves - do you need this? Why? Sure there are different throttle needs for skidpad and autocross but you can just as easily put that in your cam design and have a pretty soft, easy to modulate curve for part throttle through sweepers and skidpad and then a faster rate at the high end for course events.
--
Cam design? Again, some teams are more adept at picking mechanical systems, especially engine parts off the shelf. Different throttle curves would be a sweet thing to have control over. For one, I would like to be able to convince our drivers of several things while they pilot the car. For one, during accel, the sooner they get to full throttle, the quicker our car is (in our case at least), so ~40% throttle input would be mapped to 100% open with a pretty steep curve in there. However, in other conditions I would want them to be able to modulate a bit more and have a wider range of input to a smaller range of opening. So, 40~70% becomes pretty insensitive and allows a small, progressive percentage of opening. Why would I choose to make the throttle insensitive over such a wide range? Same reason I like to see our brake bias, tires, and entire set up changed for a rainy day; it is completely different. Using a rainy day as an example, I would want the driver to have more modulation at the lower values because that is where he will be on exiting a corner and will need a little more help controlling wheel spin on exit. This can all be switched from one to another with a simple rotary dial as well. THAT is something refining your cam design cannot do. Unless you feel like swapping cams between accel and autocross, be my guest.

I for one welcome our new throttle-by-wire overlords.

Cardriverx

06-15-2012, 12:46 AM

Originally posted by Menisk:
Just realised you could also use it to have adjustable engine braking. Rather than having the throttle fully closed in overrun you could run with it partly to fully open with a fuel cut to adjust the amount of braking to whatever the drivers find easiest.

Yes yes, this is the one I forgot to mention. Slipper clutches can be unreliable (look at akron from this year) and using an air ram on the clutch to slip a stock clutch puts a lot of wear on it.

Having adjustable engine braking with throttle by wire would be second on my list.

exFSAE, I think MCoach summed up what my response would have been. While everything can always use improvement, I can't think of one system on our car that is as bad as your "suspension rates" example. Plus we have an EE and me who deal just with electronics and engine tuning, so we have the time and ability to investigate throttle by wire.

Seriously, I think FSAE is already behind the times by not offering TBW, there are very few modern cars with a cable throttle now.

I just really hope the rule will pass.

exFSAE

06-15-2012, 05:12 AM

I don't see any harm in allowing it so long as the mode of failure is well designed by rule to go to shut throttle. I just think for the time investment your ROI will be tiny.

Want a good and challenging electronics / mechatronics project? Do an E-Diff. I bet if someone got that done right they'd absolutely crush competition on track.

But what do I know. I've been out of FSAE and doing the pro motorsport thing for 5 years now, maybe I'm too far removed.

Buckingham

06-15-2012, 06:43 AM

I would agree with exFSAE in that the ROI probably isn't there, unless you can quantify improved fuel economy by controlling how aggressive you let the driver be.

Little history lesson:
2001 U of Wisconsin successfully ran ETC in comp.
2002 5-10 additional teams showed up with ETC. If memory serves correct none of those other teams were allowed past tech with their systems.
2003 it was banned because tech inspection took 3 hours per team and because of the high rate of failure at tech inspection.

ROI was zero for the teams that failed tech and had to put a mechanical throttle back on.

06-15-2012, 08:06 PM

I'm also with exFSAE. No mid-field FSAE team is going to improve their winning chances by going to DBW. Maybe if you've been winning for the last ten years, and are getting bored. Or if there is a cheap, reliable, off-the-shelf DBW package readily available.

But to be blunt, a lot of the above talk about the benefits of DBW is a good indication of how many teams are happy to be losers. Hypothetically, if your lives were on the line (ie. win or die) how many would spend time on DBW? Keep in mind that, hypothetically, there might be talk of "a team experimenting with a simple aero package, and consistently pulling 5Gs lateral...".

Go fast first. Play with yourselves later... http://fsae.com/groupee_common/emoticons/icon_smile.gif

Z

Thrainer

06-17-2012, 05:06 PM

I like most of the ideas mentioned here, but don't think it will be legal to open the throttle when the throttle pedal is at 0%.

With such systems, you are adding a wealth of failure modes. For example, one of our first version of regenerative braking on an electric car made the car go when you stepped on the brakes in a standstill.

Nicky

06-18-2012, 03:07 AM

Originally posted by Thrainer:
I like most of the ideas mentioned here, but don't think it will be legal to open the throttle when the throttle pedal is at 0%.

In Formula One, there is a rule stating that the clutch, brake and throttle position are to be actuated as a one-to-one relation with the driver input. However, there is an exception to this rule. The clutch and throttle can be actuated even without driver input as an "Anti-Stall Function."

Moreover, there are throttle bodies that have an idler solenoid which bypasses the butterfly valve. This is used to control idling rpm. The safety in this is that the engine cannot rev up beyond a certain rpm due to the small size of the bypass channel.

...one of our first version of regenerative braking on an electric car made the car go when you stepped on the brakes in a standstill.

Not a great example to support your point. Regenerative braking is a software feature more than anything else. While it is imperative to have regen so as to reduce the size of the battery and to improve efficiency, teams tend to tune these parameters as like any other system.

In your case, all you had to add were a few checks based on some conditions of drive speed to tackle the problem.

Tell me now, there are numerous failure modes with the conventional system of using the throttle cables as well. I remember that Aristotle Racing Team couldn't complete endurance in 2010 as the idler screw kept turning in and thereby kept increasing the minimum throttle position to such an extent that it wasn't drivable beyond that.

This could've been avoided had someone foreseen the problem. This is relatively easier as it is visible and perceivable. Scrutiny of the software code is more laborious, time consuming and needless to say not easily perceivable. But flowcharts and block diagrams can be checked.

What I am coming to is that just as there is a physical scrutineering for the design and assembly of the components, there must be a basic set of rules that mandate necessary safety features to be incorporated in the code irrespective of which hardware and software a team uses to build their controller.

What can be checked in a FSAE level is whether students have incorporated basic security shut-offs in their codes. For eg:
1. Make mandatory the use of watchdog timers in controllers (resets the controller when it get hung/stuck. In simpler terms forces a reboot of the system if it were to get stuck and reset the system)
2. Make all controllers take an input from the kill switch/engine rpm so as to shut off the respective actuators when the kill switch is off.
3. Define maximum response time of actuators viz. DBW from full open to close time should be say 200ms or DBW position to move to the driver demand within 50ms.
4. Set redundancy in systems:
i. For single Accelerator Pedal Sensor, the DBW should shut down the DBW throttle in event of a failure in sensor.
ii. In case of using 2 or more sensors: the DBW controller can utilize input from subsequent sensors only in the event of the first sensor failing.
5. Minimum sampling rate of the driver feedback to be set. Like say 20ms sampling rates on the Accelerator Pedal.
6. On reset the system must cause the actuator to go to the minimum position and only then accept inputs. For eg: Clutch must go to a engaged state, throttle valve must go to 0%, Power steering must provide a 0% assist, Gear Shift actuator must be de-energized(so as to return to mean position) etc.

I welcome the rule change permitting DBW, but also strongly advise that necessary rules be put in place just as the rules that are in place for the various mechanical components on the car.

Regards,
Nicky
Electrical Head '10
Engine & DaQ '07, '08, '09, '11
Ashwa Racing (http://www.ashwaracing.com)

Cardriverx

06-18-2012, 10:46 AM

I would make the rules so that you have to have at least two position sensors on the throttle and throttle body. Running just one sensor is unsafe.

Also, we would be using a MoTeC M1 with our setup, not a custom controller. All of those features are built in.

And I agree, I beleave that a cable throttle has a similar chance of failing than a proper TBW system.

06-19-2012, 06:44 AM

Originally posted by Nicky:
4. Set redundancy in systems:
i. For single Accelerator Pedal Sensor, the DBW should shut down the DBW throttle in event of a failure in sensor.
Nicky,

How would the DBW controller know that the sensor has failed?
~~~o0o~~~

Originally posted by Cardriverx:
I would make the rules so that you have to have at least two position sensors on the throttle and throttle body.
Cardriverx,

How would the DBW controller know WHICH sensor has failed?
~~~o0o~~~

To clarify, say the sensor(s) puts out 0V for foot-up, 5V for foot-flat-to-the-floor, and inbetween V for inbetween power demand.

The controller reads 5V from the single sensor, so...???

The controller reads 5V from one sensor and 4V from the other, so...???
Here, probably shut-down, and end of Endurance and comp. All because of one crap sensor! http://fsae.com/groupee_common/emoticons/icon_mad.gif And you still don't know which one!!! http://fsae.com/groupee_common/emoticons/icon_confused.gif

So how much disparity between sensors before shut-down and embarrassing failure? Maybe 5V? So now if one sensor wire shorts-to-ground (0V) while you are accelerating, again embarrassing failure...

So how about three sensors...
My head hurts.

Z

SNasello

06-19-2012, 07:51 AM

There are definitely ways to detect inplausibility in position sensor signals. Normally, you would not install a sensor so that it reads 0 or 5V at the extremes, as this makes the plausible signal equal to an inplausible case. If the sensor had more electrical range than the throttle allows mechanically, then 0% throttle would be 0.25V or something and 100% would be 4.75. This would mean anything out of that range is inplausible (either short to ground or short to 5V). Most ETCs have features built in to detect inplausibile signals.

Most systems also have pedal and throttle sensors to make sure that the throttle isnt doing something the driver doesnt want it to do. The pedal sensor should also have a plausibility check for safety reasons. So really, you have 4 sensors.

Funky Luke

06-19-2012, 08:01 AM

So how about three sensors...
My head hurts.

So what? Multiple Sensor TBW systems have been standard equipment on new cars for quite some years now. There should be plenty of Papers etc. available for this problems. Actually a friend of mine, who did the TBW for the TU Graz electric car wrote one. From a MechEng's perspective this is probably a complex issue. From a control systems engineer probably not so much.
Best example are the FSE electric cars which, AFAIK, would regard to TBW as the least of their problems.

Here, probably shut-down, and end of Endurance and comp.

So? This is true for so many things.
Super-lightweight this-and-that: breaks. DNF
Electric clutch: jams open. DNF
No alternator: excessive power draw -> DNF
All shift systems: jam in high gear -> DNF

This is motorsport. We push to the edge and sometimes beyond. If you take too much risk or misjudge the risk you are taking you will most likely be punished. This is true for TBW as well as for any other system.

Also, nobody forces the TBW on you. If you just have a couple of wrench toting greasemonkeys on your team, don't do it. If you have a couple of skilled EE's you can probably gain something.

"Much better things to spend the development time on, like not having suspension compliance rates equivalent to al dente spaghetthi. "

oh come on... as if it was the same people who do one and the other. At least during development there should be no trade-offs between these areas. In testing...maybe.

On the practical side of things I see several advantages:

[LIST]
<LI> Idle control: currently only possible with spark advance which wastes fuel.
<LI>Anti-Lag turbo: with the second coming of the turbo FSAE engine (AMG!) this may be of use, although it wastes quite a bit of fuel. Useul for AutoX though.
<LI>Throttle response curves: Especially useful in rain and skidpad
<LI>together with e-clutch and e-gear this would allow for an almost fully automated acceleration run
<LI>downshift blip: finally. would in many cases remove the requirement for very quick clutch actuation
<LI>driveabilty tricks: together with a good ECM strategy you could do "anti lag" on N/A engines by leaving the air path rather unrestricted and just fiddling with retard. Probably only for AutoX due to fuel Issues
<LI>Launch/TC stuff: maybe too slow but woth a try

-Luke

Cardriverx

06-19-2012, 01:51 PM

Dual sensors is simple, you have a profile defined for each, example sensor 1 - .8v is 0% throttle, 4v is 100% throttle. Sensor 2 - .9v is 0 throttle, 3.9v is 100% throttle. Now the ECU knows what each sensor should report for every pedal position. Now, if one sensor deviates more than X% from the other sensor, it considers it faulty and goes into limp mode.

Buckingham

06-19-2012, 07:38 PM

Anyone looking at designing failure modes for ETC (or mechanical throttle for that matter) should look at the US Federal Safety Standards to which all US production vehicles must adhere.

49 cfr part 571 FMVSS 124
http://ecfr.gpoaccess.gov/cgi/...Title49/49tab_02.tpl (http://ecfr.gpoaccess.gov/cgi/t/text/text-idx?c=ecfr&tpl=/ecfrbrowse/Title49/49tab_02.tpl)

also google NPRM FMVSS 124 for the proposed changes to modernize the standard for ETC/hybrid/EV etc.

Nicky

06-20-2012, 03:16 AM

@Z: I hope the above posts clarified the doubts with regard to sensor redundancies.

Sensors are connected to dedicated pins on a controller. The controller need not average out the values from 2 sensors. It can rely one of the sensors and switch where it takes it input from in case that primary sensor fails. Failure of a sensor, as many pointed out, can be if the sensor reads values outside a given band. This can occur due to a faulty sensor or even cable breakage.

May I point out that the Throttle Sensors currently used on the throttle bodies face the same issue and that too without redundancies. And while we're at this, say your crank sensor wire breaks: DNF. There isn't a redundancy for such a primary sensor.

With an ICE you can still operate in a limp home mode as the butterfly controls the air-flow. The fueling and ignition can be set for a given driver demand(limp home mode). This allows the driver some degree of control on the power output despite a non-functional throttle.

My intention was to introduce standard safety features in the code irrespective of what control system you use.

I would like to add that standard non-contact type throttle sensors are rated for a lifetime of 250,000kms(155,000miles) i.e. about 15years for a passenger vehicle.

@Lukas: Idle control is possible with a purely DBW concept. ECUs used in FSAE have idle control functions to save fuel while idling(Motec M100 series).

I agree that DBW would be have a very slow response rate compared to the standard fueling and ignition control mechanisms.

06-20-2012, 06:56 AM

Originally posted by Nicky:
@Z: I hope the above posts clarified the doubts with regard to sensor redundancies.
Nicky,

The above posts have clarified how little understanding you students seem to have of these issues. I won't bother going into the details, but all the above objections to my previous post are flawed. Think about it...

Bottom line is that Murphy is better at his job than are you guys.
~~~o0o~~~

Luke,

"This is motorsport. We push to the edge and sometimes beyond. If you take too much risk or misjudge the risk you are taking you will most likely be punished. This is true for TBW as well as for any other system."

And therein is the greatest benefit of DBW. It's a cracker of an excuse for yet another DNF! "We were pushing to the edge, but, err..., unfortunately..., went beyond..."

Motorsport is indeed about making excuses, but good engineering is about eliminating risks, failures, etc.
~~~o0o~~~

Anyway, I am completely in favour in FSAE teams pushing the envelope. But I honestly can't see the point of spending time on something that has such a small performance gain, while also having so many potential failure modes (many more than I mentioned above).

There are many more low risk ways of building a much faster (ie. winning!) car.

Z

Funky Luke

06-21-2012, 01:31 AM

Originally posted by Z:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Nicky:
@Z: I hope the above posts clarified the doubts with regard to sensor redundancies.
Nicky,

The above posts have clarified how little understanding you students seem to have of these issues. I won't bother going into the details, but all the above objections to my previous post are flawed. Think about it...

Bottom line is that Murphy is better at his job than are you guys.

</div></BLOCKQUOTE>

So then grace us with your understanding. Right now you are just coming off as condescending

Funky Luke

06-21-2012, 01:33 AM

@Lukas: Idle control is possible with a purely DBW concept. ECUs used in FSAE have idle control functions to save fuel while idling(Motec M100 series).

I think this is what I meant - without DBW/TBW you have to use advance. With TBW you can save fuel.

JWard

06-21-2012, 05:00 AM

I see Z's point though, it's more that the option is now there for teams, like with a paddleshift actuated gearbox. Teams get ambitious and stretch resources and there are much bigger wins to be found than perfecting TBW, apart from teams who have a honed concept that have the resource to do it. Even then - I agree with him, the margins for improvement are small for 98% of the teams out there, perhaps they shouldn't be encouraged.

Case in point - idle control. How long do your cars sit there idling during enduro. How much fuel do you expect to save compared to spending the probably 20+ hours designing / implementing / testing it you could place elsewhere.

I see the advantage in perhaps the skidpad, and perhaps some in saving fuel with traction control during enduro instead of using advance. However, the biggest advantage I see is for teams like wisconsin / cardiff running turbos. Anti lag would seriously improve both those cars if set up right. There's a more acceptable time-spent-implementing : potential-gain ratio there. Again though, they'll have to decide whether it may increase the cost of the car past the additional sensors / actuators required, considering inconel exhaust manifolds required running large amounts of anti lag etc. It's fs masturbation imo, and I think the judges are more impressed with other things. (Although I'm totally ready for someone to prove me wrong next year)

Anyway, there's a lot to be said for the reliability of FS cars already; another complex and potentially likely unreliable system on the car is something perhaps 98% of teams shouldn't touch!?

exFSAE

06-21-2012, 05:34 AM

I still say.. OK even if this is approved and you nail all the electrical and mechanical aspects of it.. how much of a performance gain are you looking at? How many more points are you going to score with this system in perfect running order? A trivial amount?

Incidentally that's a pretty big reality that hits once you start working in industry, or on a professional motorsport team. You can't just arbitrarily chase down whatever minutiae comes to mind. Have an idea for a performance gain? OK, lay out your estimate for man hours to get it done, and ideally using some sort of predictive measure tell me how you're justifying the expense. What's the projected ROI? Then it gets weighed against the zillion other things you can be working on, and maybe it makes the cut.

One of my criticisms of FSAE - and I was just as guilty of this while I was in school - is that you get caught up in this mentality of "It's racing! Every little bit counts!" and then you can almost be more focused on small stuff rather than taking a step back and thinking about the big picture. That and all the engineering development time is free. In reality there is always an order of big gain items to small gain items. That kind of identification of value and prioritization is a HUGE necessity in a high performance / high pressure / tight time frame engineering environment. You will impress employers and managers with that.

I haven't read the FSAE rule book in a few years now, but if I had an electro-mechanical design team and the option of developing an E-Diff or Throttle-by-Wire, I'd leap at the former. Can show quite clearly in a driving or lap time sim how immense differential behavior is on handling and what gains there are... whereas TBW is much more nebulous to justify other than by hand waving.

murpia

06-21-2012, 06:39 AM

I agree with exFSAE, a torque vectoring diff is such an obvious (legal) performance boost I'm amazed I haven't seen one yet...

A lot of the FSUK / FSG electric cars are attempting torque vectoring right now. It will be interesting to see the results.

Regards, Ian

06-21-2012, 06:45 AM

Originally posted by Funky Luke:
So then grace us with your understanding.
Luke,

I have worked with highly "skilled EEs" who did this stuff. They hated it. The problem is that you have to identify EVERY possible failure mode. Then work out (ie. guess) their probabilities of failure, combine them, etc., etc. The key point is that if you miss a single failure mode, then Murphy wins!

Below the students identify ONE failure mode, find a solution, and think the problem is solved. Murphy -> http://fsae.com/groupee_common/emoticons/icon_biggrin.gif.
~~~o0o~~~

Originally posted by SNasello:
There are definitely ways to detect inplausibility in position sensor signals. Normally, you would not install a sensor so that it reads 0 or 5V at the extremes, as this makes the plausible signal equal to an inplausible case. If the sensor had more electrical range than the throttle allows mechanically, then 0% throttle would be 0.25V or something and 100% would be 4.75. This would mean anything out of that range is inplausible (either short to ground or short to 5V). Most ETCs have features built in to detect inplausibile signals.
So the driver has foot off pedal, and the DBW sensor gets a "dirty" short to 5V, or maybe shorts through the brake lamp, or some other resistance, or a million other things, and the DBW controller reads 3.9V. The car accelerates at 80% power into a wall.

There are countless ways that the controller can receive a false signal between 0.5V and 4.5V, from a single sensor.
~~~o0o~~~

Originally posted by Cardriverx:
Dual sensors is simple, you have a profile defined for each, example sensor 1 - .8v is 0% throttle, 4v is 100% throttle. Sensor 2 - .9v is 0 throttle, 3.9v is 100% throttle. Now the ECU knows what each sensor should report for every pedal position. Now, if one sensor deviates more than X% from the other sensor, it considers it faulty and goes into limp mode.
So how big is X%?

You make X=1%. The car is in "limp mode" 90% of the time (sensor drift).

You make X=10%. Now one (of the two) sensors drifts, or dirty shorts, or whatever..., by 9%. The controller has no idea which is the faulty sensor. It might decide to use one sensor as the "primary" (as suggested by Nicky) and that might be the faulty sensor. So now the engine power randomly varies +/-9% from the driver's request.

Either way, the car is undriveable. The team spends the final two weeks before comp trying to track down the problem, with no driver training.
~~~o0o~~~

Luke "So? This is true for so many things.
Super-lightweight this-and-that: breaks. DNF
Electric clutch: jams open. DNF
No alternator: excessive power draw -> DNF
All shift systems: jam in high gear -> DNF"

My point is that teams that do any of the above are obviously too comfortable with losing.
But they need a good excuse. "Oh, we pushed to the edge with super-lightweight this-and-that..., electric clutch..., DBW..., etc..., etc... Ahh, but there's always next year...".

What's wrong with winning? http://fsae.com/groupee_common/emoticons/icon_smile.gif

Z

PS. Ever wonder why modern jumbo jets have quadraplexed control systems, and they still crash?

Cardriverx

06-21-2012, 11:56 AM

Good points Z.

I really like how the thread is going, everyone is being civil and putting up good arguments!

Buckingham

06-21-2012, 05:27 PM

While I agree that the ROI isn't there, don't let anyone over "Z"ealous convince you that it can't be done properly, because it can.

Wisconsin's DBW system in 2001&2002 completed both endurances (the team finished 4th, 2nd overall, respectively) along with from my understanding hundreds of miles of testing/driving before and after those two competitions. Their system was based very heavily on OE automotive methodology, components, and failure modes.

Cornell won overall in both 2001 and 2002 and had a very dominant program in those days (with an in-house ECU?). Wisconsin didn't take 2nd in 2002 in spite of or because of the DBW. Wisconsin didn't make design finals either year.

It CAN be done properly by an FSAE team, it is more likely that it won't be done properly (as mentioned earlier 9 or 10 teams [including Cornell?] that tried it in 2002 could not pass tech with it). But I think almost everyone can agree that the ROI isn't there.

06-21-2012, 07:59 PM

To counter my general negativity above, I should say that I agree with Buckingham that DBW can be done with reasonable (?) reliability in FSAE.

If FSAE consisted of, say, 20 races a year, like most other racing series, then sure, go for it. If the DBW plays up during one race, then the gains might allow you to catch up in the other 19 races.

But FSAE is a once a year event (maybe a few more if your team has the money to travel). Most students have only a few shots at it. Cock it up, and you're pissed with yourself... forever!.

Hence my constant harping, here and elsewhere, to MAKE RELIABILITY THE FOREMOST PRIORITY. Winning is fun! http://fsae.com/groupee_common/emoticons/icon_smile.gif
~~~o0o~~~

Anyone like to speculate on "E-diffs"? ExFSAE? (It is sort of on-topic "Drive-By-Wire".)

I proposed something similar in 2005, but don't know if it has ever been used in FSAE. Big gains possible here (awesome yawing power!). Though also quite tricky (too much yawing power!!!). So only for a biggish team, with smart EEs, and possibly as a background project until well proven.

Z

exFSAE

06-22-2012, 05:22 AM

Without even getting into actively controlled torque vectoring or anything like that... think of what an immensely powerful handling tuning tool you would have if you could control how much differential lock you have between off-throttle (entry/in), neutral throttle (mid), and on-throttle (exit).

Gee, where have I seen that kinda terminology and adjustment before...
http://s1.aecdn.com/images/news/how-to-read-an-f1-steering-wheel-7154_6.jpg

Cardriverx

06-23-2012, 10:19 AM

The issue with the E-Diff is that it would have to be designed from scratch, and would be a huge project. Not to mention it would be hard to keep the weight down (just by looking at OEM E-Diffs, they are quite complex)

exFSAE

06-23-2012, 12:37 PM

Originally posted by Cardriverx:
The issue with the E-Diff is that it would have to be designed from scratch, and would be a huge project. Not to mention it would be hard to keep the weight down (just by looking at OEM E-Diffs, they are quite complex)

What if you took a Salisbury diff's internals and tried to just figure out a way to vary preload electronically?

Would it be a challenge? Sure, there's that on the one hand. On the other hand, it would be an actual performance and tuning gain. A big one. Whereas DBW is... meh. Trivial IMO.

Rex Chan

06-24-2012, 07:24 AM

I don't think I'd trust myself to do it properly. I definitely wouldn't trust a lot of guys I've worked with to do it.

And isn't anti-lag legal already? I know the TB/links must be mechanical, but you don't have to close it when you're "off throttle".

06-25-2012, 07:32 AM

Originally posted by Cardriverx:
... the E-Diff ... would be a huge project. Not to mention it would be hard to keep the weight down ... they are quite complex.
So how about doing it KISS, say with a "Dual Clutch Spool".

My idea is to have a chain or gear drive from engine to the rear sprocket. There might be "change gears" in this transmission, but no clutch. Instead there are two (higher torque capacity) clutches, one each side of the rear sprocket. These are "normally engaged" (as usual) and CV'd directly to the driveshafts out to each rear wheel. So no "diff" of any sort. A hydraulic, pneumatic, or electric system can independently disengage each clutch as required. Also assume a single rear brake acts on the rear sprocket, ie. inboard of the clutches.

If "wrench toting greasemonkeys" are doing the development (or better yet "hammer wielding blacksmiths" http://fsae.com/groupee_common/emoticons/icon_smile.gif ), then the following simple control system can be used. Some simple on/off switches (or valves) are connected to accelerator, brake, and steering. Their operation is:

1. If steering is, say, within +/- 20 degrees from straight-ahead, then both clutches engaged (locked).
(So acts as a spool for straightline driving.)

2. If steering more than 20 degrees left, and accelerator more than 10%, then left clutch DISengaged (open).
(So all drive thrust goes to RR wheel.)

3. If steering more than 20 degrees left, and brakes applied (brake light on), then right clutch DISengaged.
(So all rear braking acts on left wheel.)

4&5. Similar to 2&3, but steering to right.

Bottom line is that I reckon you could remove the steering tie-rods and still drive this car around the AutoX track. Wiggling the steering wheel around the 20 degree switch points would provide enough differential thrust/braking to turn the car as needed (like a ZTR lawn-mower). But it would be a handful at speed!

So you get some smart EEs to smooth out the switching points to give progressive clutch slip. This controlled by inputs from wheel speeds, lat/long Gs, gyros, etc., etc. Put the steering tie-rods back and you have pretty complete control of all four tyre forces. And the clutches can also be used for the mundane jobs of launch control, gear changes, and stall prevention.

Cost is two clutches, some hydraulic, or pneumatic, or electric actuators, and electronics+software.

Benefit is no differential, no engine clutch and its linkage, and awesome yaw control! http://fsae.com/groupee_common/emoticons/icon_smile.gif

Z

MCoach

06-25-2012, 01:12 PM

So how about doing it KISS, say with a "Dual Clutch Spool".

So you get some smart EEs to smooth out the switching points to give progressive clutch slip. This controlled by inputs from wheel speeds, lat/long Gs, gyros, etc., etc. Put the steering tie-rods back and you have pretty complete control of all four tyre forces. And the clutches can also be used for the mundane jobs of launch control, gear changes, and stall prevention.

Cost is two clutches, some hydraulic, or pneumatic, or electric actuators, and electronics+software.

Benefit is no differential, no engine clutch and its linkage, and awesome yaw control! http://fsae.com/groupee_common/emoticons/icon_smile.gif

Z[/QUOTE]

It's as easy as 1,2,3....right?

Z, the development that would have to go into such a set up would be annoying and very time consuming to an FSAE. I would not call this set up SIMPLE. If you compare even the size of what it costs to package these components, I would look at the difference between Audi's implementation to the all mechanical torsen differential compared to their new sport differential that works on a level a little more complicated than ours. links provided to pictures, respectively:

http://media.caranddriver.com/...83603-s-1280x782.jpg (http://media.caranddriver.com/images/11q1/381957/torsen-differential-photo-383603-s-1280x782.jpg)

http://www.caricos.com/cars/a/..._s8/1024x768/61.html (http://www.caricos.com/cars/a/audi/2012_audi_s8/1024x768/61.html)

Slipping clutches to get smooth switching points in an event such as FSAE would probably lead to failure of the clutch plates. I would point to the teams that end up slipping their clutch when the shifting systems fail and end up stuck in 2nd or 3rd gear, unable to get around the hairpins without slipping the clutch. They usually make it another 5 or 6 laps before retiring because their clutch went up in a cloud of smoke.

In regard to the central rear brake, I can't get past the fact that I would not implement it unless I was running a spool for a team, and already sacrificing tractive force or had complete e-diff control in distributing that braking load as needed.

Turning your arguments around. What happens when one of the sensors fail? What would be the failure modes?
Lets go over those a little:

In the best scenario (with regard to operation), one or so sensors fail, and the diff just remains a spool. With no engine clutch, you have a direct drive to the engine and no way to catch the car in a spin or coming to a stop unless it is already in neutral. Car dies. DNF.

In scenario 2, the spool begins to act in the wrong direction, so it causes the wrong side to lock if it is calibrated improperly. Very confusing to the driver at least.

In scenario 3, a sensor or wiring is loose, clutching becomes unpredictable and potential is to send the car flying into some direction other than intended unexpectedly and very rapidly. I would not call it a safe failure mode.

In scenario 4, the sensors or wiring fails and the clutches remain open. Engine revs, nothing happens. DNF. Although this one is one of the most unfortunate, I would call it the safest to the driver.

So, it's a concept so simple that "wrench toting greasemonkeys" can develop and "smart EEs" can perfect. Great.
With the plethora of motorcycle engines that are almost universally sourced for FSAE these days with their own internal clutches, I would not consider this type of differential for the reality of our car. Maybe, if you wanted to focus on running a V-twin Briggs motor with this as your development goal rather than a CVT I could see it as a possibility.

At the end of this post, I would also point at the Ferrari FF while employs clutch systems for driving the wheels and their way of achieving the "perfect awd control". Those who have driven it have complained about the unpredictability of the car and the engagement of the when each wheel is putting down power.

Funky Luke

06-25-2012, 06:41 PM

In regards to DBW in the wider sense - instead of perfecting a complicated and possibly heavy and expensive mechanical solution, I would try to go down the software route.

This will probably be the opening of a can of worms:
As of now, ABS systems are allowed in FSAE and a couple of teams use them.
If you had full control over the ABS valves (i.e. the software) you could at least strategically release the brakes to aid dynamics.
Take an ESP block with its additional valves and you could run an "electronic locking diff" like in many road cars, which brakes the wheel losing traction.
I understand that implementing such a system is not an easy task, even for an control systems engineer. What is an even bigger concern: what if something goes wrong here: there is the distinct possibility of a car locking one or two wheel on one side of the car at maximum speed (acceleration), sending the car right into an obstacle.
Right now, the software on these units is unregulated - but is a potentially equally big safety concern as TBW.

06-26-2012, 04:23 AM

Originally posted by MCoach:
Z, the development that would have to go into such a set up would be annoying and very time consuming ...
MCoach,

Which is why in my earlier post I said,
"Big gains possible here (awesome yawing power!). Though also quite tricky (too much yawing power!!!). So only for a biggish team, with smart EEs, and possibly as a background project until well proven.

My key point is that most FSAE teams seem to think that "progress" = "doing what we've always done..." (eg. using an off-the-shelf motorbike engine + off-the-shelf diff, etc.) but then adding some smallish, extra bit of complication, preferably electronic, and not necessarily having any big benefits. Such as an ... err ... electronic throttle! That is, progress = "let's keep polishing this turd".

My suggestion above, taken to its logical conclusion, would have a single-cylinder engine with chain(s)/gears direct to the rear sprocket, and the two clutches. This is a far simpler overall package than the conventional FSAE car with CBR600, etc., yet has superior dynamic potential. Sure, it also would need some real engineering to get it working properly (eg. sizing the clutches so they don't burn out when slipped) , which is why I doubt it will ever appear in FSAE.

Anyway, I do NOT think the "awesome yawing power" of torque vectoring is necessary to dominate in FSAE. Easier is 60% rear weight + single-cylinder + open-diff + aero undertray, with most development focussed on the undertray.

Efficient aero = much, much faster than any amount of electronics! http://fsae.com/groupee_common/emoticons/icon_smile.gif

Z

MCoach

06-26-2012, 10:53 AM

When sizing projects to be fit to a car, I'm not looking at things from an OEM perspective of trying to develop new technologies for a vehicle 5 - 15 years out for research. I'm looking for problems that occur on our car currently to fix or or articles to add to aid performance somewhere. These projects I organize into categories of things that will work currently, things that need to be developed into next years car, things that take 1 year of engineering and testing, projects that will take 2 years of engineering and testing. I don't try to consider projects that will take 3+ years because of the follow through involved. By the time whoever starts the project graduates and new people take on the development, the chance that they will have the same understanding of the system is very small and therefore the risk is high to not complete the project. Graduate work is not a focus on this team.

This is why I say it will be annoying.

Theoretically, I would invest in the aero before trying to figure out a clutch pack system to work.

My approach to systems and whether or not they will be helpful with respect to an FSAE team is "complex analysis, simple implementation".
I don't need systems on a car which parts are not easily replaceable. A wheel center shatters, we put another on. A suspension arm breaks due to impact, we made spares. Diff blows up, swap in the extra. Motor blows up, find out why. Off the shelf items allow us to minimize costs and the little man power we have to machining new parts.

Reliability is still a concern in this competition, and with our track record, I would trust electronics over the short term development of some experimental mechanical systems.
We don't need to reinvent the wheel. Sure, the Graz teams have built their own engine with AMG, but that is an exception. Creativity is encouraged, but should not make a Rube Goldberg machine out of the car. You can't make vast changes everywhere and expect things to be predictable.

Considering the "progress" that has been made in this competition in the past two decades alone, I would call it real progress. An SAE paper was written in 199(3?) about how these cars will never achieve more than 1.3G in any direction. The lap times pulled at each competition these days are just astounding and I am quite sure they will be surpassed as time continues. Do not insult the competitors and call it "polishing a turd". It's read "evoluzione".

AxelRipper

06-26-2012, 05:56 PM

To digress further into the realm of the e-diff, there is a far simpler version than Z suggested. Get one out of a utility cart. These are simply mechanical locking differentials, usually actuated by a cable. If you were to create a setup with an electronic system to actuate the locking mechanism (possibly just a linear actuator to pull in place of the cable) then you would have an open diff with a programmable lock-up setting that could be used to lock in either acceleration or braking, using whatever inputs you would like.

Then, when your system fails (this is FSAE after all) you would have an open diff or you could rig it into a spool easily.

The far more complex setup with this same differential could use a series of linkages set at different points around the chassis hooked to the steering rack, throttle pedal, and brake pedal that would pull the cable when they got to a certain area and release in others. Or you just give your driver another handle to worry about that he can pull for maximum acceleration.

MCoach

07-24-2012, 10:49 PM

Why not just use cutting brakes to initiate a substantial yaw moment?

An e-diff would be spiffy, but even Audi is having problems with the complexity.

Drift-o's anyone?

Tom Wettenhall

07-30-2012, 05:55 AM

For my 2c, look at E85. Seems stupid at first. You need to carry more of the stuff with you, it gets in your oil and asplodes your engine, and it's a pain to actually get any points benefit out of it. ROI low. Disregard.

On the flip side, once you've grenaded a couple motors and know how it works, it allows more aggressive spark timing, (=more power, if you care about power) the fuel economy offset actually works out pretty well for you, (='less fuel') and sponsors absolutely love the whole renewable fuels thing. (=money) Suddenly it's pretty good.

So, although I don't think TBW is a good idea for my team, I have no doubt that someone would be able to get a big return out of it.

Charles Kaneb

07-30-2012, 01:32 PM

I'd love TBW/DBW to be legal, as if it's properly implemented it will significantly increase the reliablity of our car.

The team I was on last year ran a single. It idled at 6000 RPM - any slower and it wouldn't run or start reliably. What do you think 100 step inputs a second does to the lifespan of lightly-loaded aluminum parts? Remember, you get more vibration from one of these engines under zero load than under full load. If we could get the thing to idle at 2500 with DBW it would be worth it on reliability grounds alone.

Paul Achard

07-30-2012, 06:08 PM

If I might pitch in to the conversation on E-diffs: I had thought a bit about Funky Luke's idea after reading about it in RCVD. You could fit a very lightweight open diff (golf cart?), along with 4-piston, dual circuit calipers on each rear wheel (see AP-racing's 2x2 caliper). Run one circuit to the main braking system and the other to a pair of servovalves or the like. Inside wheel starts spinning, apply braking force to that wheel.
Provided you can find reasonably-priced e-valves, this requires a minimum of moving parts, you get perfect differentiation characteristics (limited by software) and if your system fails you still have your braking and an open diff.

murpia

08-07-2012, 06:47 AM

Originally posted by Paul Achard:
Provided you can find reasonably-priced e-valves
I like this idea, and HydraForce (among others) can supply suitable valves around the 100USD price point.

Regards, Ian

jlangholzj

08-07-2012, 12:32 PM

Originally posted by Charles Kaneb:
I'd love TBW/DBW to be legal, as if it's properly implemented it will significantly increase the reliablity of our car.

The team I was on last year ran a single. It idled at 6000 RPM - any slower and it wouldn't run or start reliably. What do you think 100 step inputs a second does to the lifespan of lightly-loaded aluminum parts? Remember, you get more vibration from one of these engines under zero load than under full load. If we could get the thing to idle at 2500 with DBW it would be worth it on reliability grounds alone.

if you've got a car idling at 6000 you've got problems that wouldn't be solved with DBW.

I say open it up and allow for development. Its going to take 2-3 years to get a successful system anyway. Its not any different than having a dynamic aero package that doesn't break (OU) or the first few years of making a carbon monocoque and not throwing an a-arm through it (RIT) or using a push pull cable and melting it because my dumb ass didn't heat shield it......The problem is making large jumps in development in too short of a time.

probably the safest way to do it is to use a digital sensor with watchdog and CRC's (cyclic redundancy checks) + DED/DEC (dual error detection/correction). You'd be able to tell in a hurry if the sensor was bad.

even with TBW you'd have starting profiles for getting the engine warm (high idle) and the other major gain i could see is a "smoother" throttle response at speeds below a threshold (exiting corners) that are past some steering angle. Plus you'd have some idle control, and a couple other minor things.

I'm sure that there's some more to be had in there somewhere....

another thing everyone keeps forgetting about is packaging. Who says you need a POT for the sensor? Use pressure transducers instead. Very little pedal throw would be needed (<1/8").

brake by wire would also be pretty sweet. Could mount the reservoirs and MC's somewhere remote, tie it into the wheel speeds for front and rear independent ABS.