Jump to content


Photo
* * * - - 2 votes

Is it possible legally to emulate traction control in F1


  • Please log in to reply
104 replies to this topic

#101 ReeVe

ReeVe
  • Member

  • 178 posts
  • Joined: October 13

Posted 17 October 2013 - 12:29

was it korea or singapore where Rosberg was warned twice not to ride the brakes?



Advertisement

#102 Rubens Hakkamacher

Rubens Hakkamacher
  • Member

  • 1,567 posts
  • Joined: March 04

Posted 17 October 2013 - 17:34

 activation must be logged by the ECU it would be noticed. the ECU also monitors the charging and discharging cycles of the batteries anything unusual would be noticed.

 

 

 What I am describing is between the batteries and the ECU, between the KERS motor and the ECU.   The ECU is going to know the battery is discharging.  The ECU doesn't know the nature of the analog circuit itself.  A simple circuit of a battery pack and a capacitor alone opens the door to Much Shenanigans, and it would be very hard to differentiate what is the natural process of the system and what is "overly helpful" to the driver.  

 

The FIA inspector would have to pour through all the turns and examine the data for minute moments when wheel spin is shown.  He's got a trace of the I/O from the KERS system.  He would see *sometimes* where the impedance of the KERS motor goes down, and where the output of the battery/cap system oscillates. 

 

 Is that TC?  The team would argue that it's hysteresis in the system.  The system has to have a hardware way of protecting the battery from over voltage, too rapid discharge, a battery malfunctioning (consider it's not one battery - it's a pack of batteries), all sorts of combinations of *electrical*, not *digital* situations that requires *hardware* to be present in the circuit.

 

 If it turns out that hardware - batteries, capacitors, coils, even the wiring or the grounding scheme - has a hysteresis that smooths the delivery of power, or "cleans" it up, the FIA inspector then has the onus of proving that it's behavior is intentional.

 

 I might imagine "mystery oscillations" are probably rampant in the KERS data.  Analog electrical components are pretty non-linear in general, and can do some interestingly counter intuitive things suddenly when the temperature changes, or crosses a particular threshold. 

 

 Consider the angle that you can't legistlate the analog side of the ECU, because ultimately the team is going to have a base design philosophy of making a KERS system that is as efficient at collecting and delivering power as possible.  Electrical power is not binary like chemical combustion - it can be modulated or continuous, in varying strengths AND discharged with varying timings and curves.  Unless the FIA wants to open the kettle of worms that regulates the precise timing of these features you can't change the nature of the analog portion of the system.  You'd also find it impossible, because you couldn't regulate a particular continuous curve or temporal range for an energy output without creating disasters where things over run or under run.  

 

The KERS motor would have to be hardwired to a single 100% on/off switch controled by the ECU, and then hard wired straight to the battery, to negate gimmicking that portion of the system.  Which can't happen - you'd have batteries blowing up, motors burning up, uncontrolled wheelspin, engines lugging... that's just not practical.

 

Analog circuitry is not like digital, it's a land of many non-linearities and arcane trickery.  A lot of which may be getting obscured by time, as older electrical engineers are pushed out by guys just educated in control systems.  Between a super genius programmer, and a super genius EE, I would suspect they could find many ways of implementing KERS smoothing that would take a long while for the non-genius to discern it's presence. 



#103 Rubens Hakkamacher

Rubens Hakkamacher
  • Member

  • 1,567 posts
  • Joined: March 04

Posted 17 October 2013 - 17:37

and why go through the trouble of doing aaaaaaaaaaaaaaaaall that just to shave 80bhp off the top, when you are free to switch half the engine off in the traction zones?

 

 Because it would only be doing it in reaction to wheelspin. 


Edited by Rubens Hakkamacher, 17 October 2013 - 17:37.


#104 Kelateboy

Kelateboy
  • Member

  • 7,032 posts
  • Joined: October 07

Posted 19 October 2013 - 03:39

A bit more information on how SECU TAG-320 works in relation to traction control.
 
MES has used Simulink since 2000. "We have been supported by The Mathworks, but have had to do a lot of work to manage the large and complex control strategies," Strafford explained. 
One change from the previous model is to provide the ability for F1 teams to run their own code alongside mandated applications. "We wanted to offer the best of both worlds," Strafford said. 
 
With this approach, each application runs in its own address space and can only access certain signals. "For example, traction control is blocked by not allowing the routine to access wheel speed data," Strafford noted. 
 
Tim Strafford, business development manager for MES, was a control systems engineer with the McLaren F1 team.


#105 Rubens Hakkamacher

Rubens Hakkamacher
  • Member

  • 1,567 posts
  • Joined: March 04

Posted 19 October 2013 - 13:36

Ok, software side:

 

If they let them run their own code, all bets are off.  Again, unless you are a super genius programmer, you can no more fathom the possibilities of exploiting a system than you could coming up with your own version of E=mc^2. 

 

But, off the cuff: even with what is written above, there are still ways to do "things".   

 

TC is blocked simply because you don't get wheelspin data?  You don't need wheelspin data.  They have the "sensor" of the KERS motor telling them what is happening with the torque difference.   If that is what they're relying on for making TC "blocked", that is a complete joke.  Just program it to back off the timing if the KERS sensor's impedance drops.  There is your engine noise, and furthermore you could tweeze it timing wise to further optimize how it's controlling the KERS sensor... I mean, "motor".

 

Outside of that, since you can inline code, there is another exploit....

 

Code execution timing.  You could put your TC routine out in the open, disguised as simply a cleverly inefficient function that takes the processor longer to execute. "When the KERS motor shows a drop in resistance < (x milliseconds) call (Sooper KERS Smoothing Class)" - that maybe shuffles things around inappropriately a few thousand times, or does some inefficient math call, that means while it might look like it's "making the application of energy smoother" it's also retarding the IC timing.

 

Or really anything, if they're counting on that "they can't access wheelspin" point (not that I would bet on that being impossible).

 

The thing to consider is that the teams do have a pretty thorough database of timing data/curves, so while you may not be able to program a traction control class you could implement throttle response in one gear that specifically replicates the engine/KERS timing required to cope with "instabilities in torque application".   While the rules state you can't do that *explicitly*, there is no way of regulating one's specific algorithmic approach to dealing with the ignition timing demands of the IC motor, and the offset of the KERS motor. 

 

I think the notion that TC can't be implemented on both the analog electronic, AND the digital side is patently naive.  McLaren are not using bespoke components for the ECU, they have put together a *system* of off the shelf hardware and software that could have a plethora of ways in which to be exploited.   If the teams are allowed to load their own software, not via an FIA official or McLaren engineer - which I presume they can, since they're obviously constantly hooking their own computers into the cars all weekend long - *you can't know what has gone in and out*.   The most important thing is, has McLaren written their own specific compiler (unlikely), AND do they vet the code going in and out (it appears they don't).  


Edited by Rubens Hakkamacher, 19 October 2013 - 13:47.