PSM theory
#1
Racer
Thread Starter
Join Date: Jun 2001
Location: Louisville, KY
Posts: 318
Likes: 0
Received 0 Likes
on
0 Posts
PSM theory
The following was written by a friend of mine who has never read from this board, but I have sent him excerpts from time to time. I sent him the thread started by Mike in Chi (Topic: The 996 guys are on to something... ), which touched off a long discussion about traction/yaw control and people's expectations. My friend Jerry works on these traction systems, and had some expertise and a few thoughts to share. Greg Fishman, you should recognize his letter. Absolutely no flames intended, I think the view from an insider is fresh and very interesting:
There are some misconceptions on what PSM really is trying to do.
First off, I work for a company that designs and manufactures Braking Systems. This includes foundation brakes, ABS, Traction Control, VSC (Vehicle Stability Control), and other driver enhancing products coming in the near future. In particular, I have worked on VSC for the last 3 years. I have not experienced Porsche's PSM, but I have driven several other systems, and have a good fundamental knowledge of various systems. Without going into a lengthy discussion, I just wanted to make some clarifications.
The "mis-fire" that you experienced is not cutting off the fuel flow. It is retarding the spark. (i.e. allowing certain cylinders to not fire a spark) The reason for this is that your Traction Control (TC) is detecting a wheel spin-up (rears in your case) that is about to exceed (or already has) the maximum traction obtainable. If you exceeded that limit on a turn, not only would you lose longitudinal traction, but you would drastically lose lateral traction. Instead, the TC system commands to the PCM (power train control module) a requested maximum engine torque. The PCM then retards the spark as it sees fit (alternating cylinders to minimize vibration) in order to not exceed the torque requested. While this does sometimes hinder track driving limits, when it comes to wet, snow, or ice you'll be glad to have it.
There is no way that PSM can (or would try) to learn someone's driving style. Imagine that you just finished some hot laps racing, then you let someone's grandmother behind the wheel. PSM can't decipher the skill level of the driver. The control would be too "loose" for her ability. Instead, the system is calibrated to the type of car and the typical driver. Porsche engineers have the luxury of getting sports car enthusiasts to buy their cars. They can calibrate their PSM system to allow the maximum leniency to sport driving, and only allow PSM to interact just before things get out of control. Your M3 is slightly more conservative, keeping in mind that it is still a road car with a variety of driver skill levels. Mercedes is probably the most conservative, opting for safety and stability over sport and track.
VSC has the following primary inputs: Lateral accelerometer, Yaw rate, individual wheel speeds, and vehicle speed. These tell the VSC what the "car" is doing. It also has a steering angle sensor that tells VSC what the "driver" is intending. It compares these inputs and determines when there is a certain level of deviation or an impending problem, and acts accordingly. The two Primary corrections are to send brake pressure to certain wheels (to create a moment/torque in the oposite direction of the skid) and/or reduce engine torque. Depending on the type of vehicle, targeted driver, and OEM philosophy, it can be calibrated to be either conervative or "loose".
I hope that it makes it a little clearer. One last comment, PSM and other systems are intended to make cars, trucks, SUV's much safer vehicles on the road where the unexpected can and usually does happen. For the track, there is always the OFF button!
Jerry
Lonn,
What you say makes sense and I would say the same thing if I hadn't driven a car equiped with PSM. I took my E36 M3 to the track one time last year and I kept getting a mis fire as I would come out of the turns. Checked the gas level, etc. After a half a lap I realized it was the traction control and turned it off, it interupts the fuel flow (I think), not what you want on the track (equivalent to lifting of the throttle). What I understand about the PSM is that it learns your driving style and gives you more or less "correction" depending on the situation. If you mostly commute in the car it will correct much quicker than if you are always driving the car hard or on the track.
You need to enroll in a DE or the PDE at Road Atlanta, if you like your car now, you will love it after one of these events!
Greg
What you say makes sense and I would say the same thing if I hadn't driven a car equiped with PSM. I took my E36 M3 to the track one time last year and I kept getting a mis fire as I would come out of the turns. Checked the gas level, etc. After a half a lap I realized it was the traction control and turned it off, it interupts the fuel flow (I think), not what you want on the track (equivalent to lifting of the throttle). What I understand about the PSM is that it learns your driving style and gives you more or less "correction" depending on the situation. If you mostly commute in the car it will correct much quicker than if you are always driving the car hard or on the track.
You need to enroll in a DE or the PDE at Road Atlanta, if you like your car now, you will love it after one of these events!
Greg
First off, I work for a company that designs and manufactures Braking Systems. This includes foundation brakes, ABS, Traction Control, VSC (Vehicle Stability Control), and other driver enhancing products coming in the near future. In particular, I have worked on VSC for the last 3 years. I have not experienced Porsche's PSM, but I have driven several other systems, and have a good fundamental knowledge of various systems. Without going into a lengthy discussion, I just wanted to make some clarifications.
The "mis-fire" that you experienced is not cutting off the fuel flow. It is retarding the spark. (i.e. allowing certain cylinders to not fire a spark) The reason for this is that your Traction Control (TC) is detecting a wheel spin-up (rears in your case) that is about to exceed (or already has) the maximum traction obtainable. If you exceeded that limit on a turn, not only would you lose longitudinal traction, but you would drastically lose lateral traction. Instead, the TC system commands to the PCM (power train control module) a requested maximum engine torque. The PCM then retards the spark as it sees fit (alternating cylinders to minimize vibration) in order to not exceed the torque requested. While this does sometimes hinder track driving limits, when it comes to wet, snow, or ice you'll be glad to have it.
There is no way that PSM can (or would try) to learn someone's driving style. Imagine that you just finished some hot laps racing, then you let someone's grandmother behind the wheel. PSM can't decipher the skill level of the driver. The control would be too "loose" for her ability. Instead, the system is calibrated to the type of car and the typical driver. Porsche engineers have the luxury of getting sports car enthusiasts to buy their cars. They can calibrate their PSM system to allow the maximum leniency to sport driving, and only allow PSM to interact just before things get out of control. Your M3 is slightly more conservative, keeping in mind that it is still a road car with a variety of driver skill levels. Mercedes is probably the most conservative, opting for safety and stability over sport and track.
VSC has the following primary inputs: Lateral accelerometer, Yaw rate, individual wheel speeds, and vehicle speed. These tell the VSC what the "car" is doing. It also has a steering angle sensor that tells VSC what the "driver" is intending. It compares these inputs and determines when there is a certain level of deviation or an impending problem, and acts accordingly. The two Primary corrections are to send brake pressure to certain wheels (to create a moment/torque in the oposite direction of the skid) and/or reduce engine torque. Depending on the type of vehicle, targeted driver, and OEM philosophy, it can be calibrated to be either conervative or "loose".
I hope that it makes it a little clearer. One last comment, PSM and other systems are intended to make cars, trucks, SUV's much safer vehicles on the road where the unexpected can and usually does happen. For the track, there is always the OFF button!
Jerry
#2
Lifetime Rennlist
Member
Very enlightening piece AJ. Thanks. If Jerry has other observations, please post them.
As I've thought about PSM, it raises an interesting question. It can help an experienced driver go faster, and a novice go faster. But for the novice, does it actually slow his learning curve, because he's not learning HOW to correct? If PSM is always there, in one sense it doesn't matter. But then the driver becomes PSM dependent.
Ultimately, I guess he learns more with the button in the off position, as long as he doesn't get in too deep.
As to traction control retarding the spark, I wonder if that's what the F-1 cars do. I thought a few of the cars were srarting to misfire in turns 1 & 2 at Indy this year, til a friend clued me in that it was the TC kicking in.
As I've thought about PSM, it raises an interesting question. It can help an experienced driver go faster, and a novice go faster. But for the novice, does it actually slow his learning curve, because he's not learning HOW to correct? If PSM is always there, in one sense it doesn't matter. But then the driver becomes PSM dependent.
Ultimately, I guess he learns more with the button in the off position, as long as he doesn't get in too deep.
As to traction control retarding the spark, I wonder if that's what the F-1 cars do. I thought a few of the cars were srarting to misfire in turns 1 & 2 at Indy this year, til a friend clued me in that it was the TC kicking in.
#3
The traction control systems in F1 operate by ignition angle changes and selective cylinder ignition cuts where required. Fuel cuts are rarely used (response is too slow), throttle angle control is sometimes used as well (the throttles are under electronic control and generally hydraulically actuated).
#4
Instructor
There is no way that PSM can (or would try) to learn someone's driving style. Imagine that you just finished some hot laps racing, then you let someone's grandmother behind the wheel.
#5
Rennlist Member
I'm not an expert in desgining Stability control, nor do I work for Porsche, but I am an expert in software engineering & design...and this idea that a learning PSM is impossible is just wrong.
Porsche may have implemented PSM using a large-degree of freedom expert system (sometimes known as a neural network). Such a system would be capable of "Learning a new driving style within a few minutes (or laps on the track) depending on the number of degrees of freedom, and the depth of the netowrk.
Essentially what may happen when a new driver takes the wheel is that the system detects a change from it's "zone of comfort" and resets, knowing that it must develop a new set of control parameters for a new driver. This reset would almost certainly take place after a engine shutdown/restart or going from track -> road -> track.
From everything I've read, PSM behaves just this way, operating very conservatively on the track until the algorithm's control parameters are sufficently set (over the course of several laps) such that PSM had
"detected" that the driver has appropriate skill to handle traction closer to the limit.
The statement: "There is no way that PSM can (or would try) to learn someone's driving style" may be correct in fact. However, it's certainly feasible to design a control system capable of "learning" a driver's style - it should not be dismissed out-of-hand.
For more information on the topic, use the web to research "Expert Systems" or "Neural Networks".
cheers,
sean
Porsche may have implemented PSM using a large-degree of freedom expert system (sometimes known as a neural network). Such a system would be capable of "Learning a new driving style within a few minutes (or laps on the track) depending on the number of degrees of freedom, and the depth of the netowrk.
Essentially what may happen when a new driver takes the wheel is that the system detects a change from it's "zone of comfort" and resets, knowing that it must develop a new set of control parameters for a new driver. This reset would almost certainly take place after a engine shutdown/restart or going from track -> road -> track.
From everything I've read, PSM behaves just this way, operating very conservatively on the track until the algorithm's control parameters are sufficently set (over the course of several laps) such that PSM had
"detected" that the driver has appropriate skill to handle traction closer to the limit.
The statement: "There is no way that PSM can (or would try) to learn someone's driving style" may be correct in fact. However, it's certainly feasible to design a control system capable of "learning" a driver's style - it should not be dismissed out-of-hand.
For more information on the topic, use the web to research "Expert Systems" or "Neural Networks".
cheers,
sean
#6
Hmmm, when I instructed a gentleman in is 01 TT he asked me to drive at the end of the day to show him the capabilities of his car. We began and as I built speed the PSM would kick in initially as the car began to drift a bit.It would typically touch one corner of the cars brake to halt the rotation. As we progressed the PSM would come in less and less until at the end we were utilizing full trail braking to rotate he car and going through corners drifting all 4 wheels with no interruption. A friend of mine rented the services of David Murray for a day and was driving his 996 C4. My friend is an instructor also and will be Club Racing a 964 this season. David explained to him that the PSM would actually "learn" his capabilities as a driver and encouraged him to leave it on. As they spent the day together they experimented and toward the 2nd half of the day they ran with it on the whole time. He was able to drive the car with almost no PSM involvement whatever. They used trail braking and 4 wheel drift etc... I am not sure that "learning" is the proper term but it sure seems to have some learning/adaptive qualities that take into account the skill of the driver. One thing I would add, is that at restart it was always back to the PSM being invasive early on and then it would move back up to letting you"drive". Just my observations.
#7
Addict
Rennlist Member
Rennlist Member
A.J.:
That is an interesting piece of information. It is very interesting to read the responses from the different experts that share this board. I had a ’99 E36 M3 that reduced power through throttle angle as JohnM points out. It did not use ignition angle as a method for reducing power as I recall.
Its secondary method of reducing slip was by applying brake pressure to the slipping rear wheel. This was fully active to 25 MPH and exhibited reduced but active functionality up to 60 MPH. I can say that the BMW ASC+T system is very crude when compared to PSM. I know that BMW has introduced Dynamic Stability Control, DSC, which I am sure is much more sophisticated than the mid nineties technology on the previous M.
I drove a friend’s Z3-M around Roebling Road once. Unbeknownst to me, the traction control was active. This increased my lap times by several seconds, as it was no help at all! My friend claims that he was unaware of the traction control being active too. Well, that’s one way to exert a little control while your friend is thrashing your car!
I hope we get a definitive answer to the adaptive question regarding PSM. Inquiring minds want to know!
Thanks for providing such interesting information.
Hank Cohn
That is an interesting piece of information. It is very interesting to read the responses from the different experts that share this board. I had a ’99 E36 M3 that reduced power through throttle angle as JohnM points out. It did not use ignition angle as a method for reducing power as I recall.
Its secondary method of reducing slip was by applying brake pressure to the slipping rear wheel. This was fully active to 25 MPH and exhibited reduced but active functionality up to 60 MPH. I can say that the BMW ASC+T system is very crude when compared to PSM. I know that BMW has introduced Dynamic Stability Control, DSC, which I am sure is much more sophisticated than the mid nineties technology on the previous M.
I drove a friend’s Z3-M around Roebling Road once. Unbeknownst to me, the traction control was active. This increased my lap times by several seconds, as it was no help at all! My friend claims that he was unaware of the traction control being active too. Well, that’s one way to exert a little control while your friend is thrashing your car!
I hope we get a definitive answer to the adaptive question regarding PSM. Inquiring minds want to know!
Thanks for providing such interesting information.
Hank Cohn
Trending Topics
#8
Lifetime Rennlist Member
I find it very hard to beleive that power is conrolled with spark retard or missed spark - essentially forcing misfires in a street car. This is nearly instant death for a catalytic converter - it pumps unburned fuel into the cat, which is hot enough to light it and creat a small explosion in the cat, breaking up the substrate. I can understand this being used as a very fast response in a race car.
#9
Originally posted by Mark D - 993TT CA:
<STRONG>I find it very hard to beleive that power is conrolled with spark retard or missed spark...</STRONG>
<STRONG>I find it very hard to beleive that power is conrolled with spark retard or missed spark...</STRONG>
#10
Interesting thread. I have to agree with Sean that electronics are capable of adapting their responses based on recognising inputs. Besides neural nets, my 89 Supra (model introduced 85?) had fuzzy logic control of the fuel injection. One of it's features was compensation for very slow changes eg debris build up, hose degradation. IE it was learning about the condition of the system and not rigidly adhering to the factory settings.
And I don't know why the PSM system would have to be limited to just one driver "calibration". Many smart automatic 'boxes already have several shifting patterns, and select one based on the "input" eg the way the car is driven. Now something like that should be not too difficult to include in PSM, IMHO of course.
Cheers, Nol
And I don't know why the PSM system would have to be limited to just one driver "calibration". Many smart automatic 'boxes already have several shifting patterns, and select one based on the "input" eg the way the car is driven. Now something like that should be not too difficult to include in PSM, IMHO of course.
Cheers, Nol
#12
Addict
Lifetime Rennlist
Member
Lifetime Rennlist
Member
AJ,
Thanks for the info.
I had my M3 for a couple years and while I liked driving it I wasn't really interested in knowing every nuance of its construction like I am my 993, so I probably mispoke about its method of traction control.
While I am far from a computer or software specialist, I find it hard to believe that Weissach couldn't come up with a system that learns your driving style and adjusts accordingly. I don't have any facts on this just going on conversations I have had with other 996 drivers.
Greg
Thanks for the info.
I had my M3 for a couple years and while I liked driving it I wasn't really interested in knowing every nuance of its construction like I am my 993, so I probably mispoke about its method of traction control.
While I am far from a computer or software specialist, I find it hard to believe that Weissach couldn't come up with a system that learns your driving style and adjusts accordingly. I don't have any facts on this just going on conversations I have had with other 996 drivers.
Greg