Differences with the DCT Upgrade

[footie]

It's great to see that Swamp and myself are in total agreement over all of the improvements with the new upgrade. Every opinion on how each different discipline is experience also seems to be the same.

One other thing I noticed today while seeing if the short shift was cured as the best way to get your M3 off the line without using Launch Control. Without a doubt the stumping on the throttle while stationary is not the best way, all that happens is a build up in revs and lots of wheel spin. No I found that a very brief push of the throttle, just enough to get the car rolling and then pressing the throttle steadily to full throttle, say about 1/3 of a second produced near zero wheel spin and an acceleration which was blistering. I am not completely sure but without sampling LC I can't believe it will be that much quicker.

On the next good dry day I promise to set both this method, dumping the throttle and LC against the iPhone program and will post my findings.

[MDCTFTW]

Man I need to get my car in there.

[swamp2]

Originally Posted by ajj

the surge is great, but the reason i believe they got rid of it is that it can be dangerous in my experience. when accelerating at WOT at high speeds (100+ mph), the car "skips" a little when shifting gears due to this surge. this is since the increased force is not all forward...but in fact, it appears to give the car a small diagonal force vector...i.e. small lateral component . this causes the car(at least mine) to slightly "skip" laterally and briefly lose traction at high speeds when traversing in a straight line. i think car and driver described it as "losing its composure". this can be a bit unnerving in my experience.

I can see this, although my experience is that the surge really only surges in the forward direction. Can you point out the article you mention? Thanks.

[ihyln]

Does the car still start out in manual mode?

[swamp2]

Originally Posted by footie

It's great to see that Swamp and myself are in total agreement over all of the improvements with the new upgrade. Every opinion on how each different discipline is experience also seems to be the same.

One other thing I noticed today while seeing if the short shift was cured as the best way to get your M3 off the line without using Launch Control. Without a doubt the stumping on the throttle while stationary is not the best way, all that happens is a build up in revs and lots of wheel spin. No I found that a very brief push of the throttle, just enough to get the car rolling and then pressing the throttle steadily to full throttle, say about 1/3 of a second produced near zero wheel spin and an acceleration which was blistering. I am not completely sure but without sampling LC I can't believe it will be that much quicker.

On the next good dry day I promise to set both this method, dumping the throttle and LC against the iPhone program and will post my findings.

I noticed this too. Just turn DSC off and manage the wheelspin yourself. However, since I have not tried LC and do beleive it has some bugs before it is premature to say human>computer. If the new DCT better manages the clutch drop and uses DSC to manage the wheelspin there is absolutely no reason it won't be much more consistent and perform better than a human can.

[swamp2]

Originally Posted by ihyln

Does the car still start out in manual mode?

Yes, unfortunately...

[man02195]

Originally Posted by MDCTFTW

Man I need to get my car in there.

WOW, just got my car back, what an amazing improvement. There's now more anticipation necessary for shifting, it happens when you tell it to. I tried a few threshold braking shifts, and they were all executed, but I'll have to wait to see it on the track next month.

[ajj]

Originally Posted by swamp2

I can see this, although my experience is that the surge really only surges in the forward direction. Can you point out the article you mention? Thanks.

can't verify there is a small lateral force vector, or if the additional force is just not evenly distributed to the wheels, but my car appears to slightly "skip" sideways when shifting under wot high speed conditions. scared the crap out of me the first time this happened as i was well over 100mph. the article i was referring actually states that "when M DCT finally does shift, it often slams into gear so violently it upsets the car."

http://www.caranddriver.com/reviews/...r_+page-2.html

[footie]

I remember mentioning this very problem with have surge prior to anyone outside of BMW actually driving one. Do you remember Swamp?

I said at the time that surge would cause the car to become unsettled if it happened in a corner. Well what do you know, that is exactly what happens, though when it happened to me it was only at about 40mph going into 2rd gear and on a slightly damp surface. The lateral skip I think is purely the traction being broke causing the tail to move but the effect could be unnerving at big speeds, plus the reason I think it is happening is because the corner isn't sharp enough to cause the kind of lateral G for the computer to stop the surge (which the system is suppose to do).

ajj,

Am I correct in thinking the corner it happened on was slight, not enough to have a decent amount of steering lock?

[grayshrk]

Swamp - My car is also at Irvine BMW getting the update. I'm very happy to hear that everything went well with your car and they seem to know what they're doing. Unfortunately I gave the SA a little grief yesterday because I had just had the car in for the break-in service on Tuesday of this week and they didn't mention the update being released. I guess it's my fault for not checking the forum prior to taking the car in for service. Anyway, after reading these posts, I'm really looking forward to picking the car up later.

[man02195]

Originally Posted by grayshrk

Swamp - My car is also at Irvine BMW getting the update. I'm very happy to hear that everything went well with your car and they seem to know what they're doing. Unfortunately I gave the SA a little grief yesterday because I had just had the car in for the break-in service on Tuesday of this week and they didn't mention the update being released. I guess it's my fault for not checking the forum prior to taking the car in for service. Anyway, after reading these posts, I'm really looking forward to picking the car up later.

Our local dealer didn't get the software until wednesday according to the techs, so it might not have shown up unless you had a puma case.

[ajj]

Originally Posted by footie

Am I correct in thinking the corner it happened on was slight, not enough to have a decent amount of steering lock?

actually happened to me a couple times on a straight road...when it happens at low speed, doesn't really bother me, but at high speeds, it really is very disconcerting.

[grayshrk]

Well, it turns out the reprogramming crashed while in process with my car at Irvine BMW. They've got it going again but it still has 4+ hours to run according to my SA. Looks like I won't be picking up my car tonight after all.

[swamp2]

Originally Posted by footie

I remember mentioning this very problem with have surge prior to anyone outside of BMW actually driving one. Do you remember Swamp?

I said at the time that surge would cause the car to become unsettled if it happened in a corner. Well what do you know, that is exactly what happens, though when it happened to me it was only at about 40mph going into 2rd gear and on a slightly damp surface. The lateral skip I think is purely the traction being broke causing the tail to move but the effect could be unnerving at big speeds, plus the reason I think it is happening is because the corner isn't sharp enough to cause the kind of lateral G for the computer to stop the surge (which the system is suppose to do).

ajj,

Am I correct in thinking the corner it happened on was slight, not enough to have a decent amount of steering lock?

I do recall this discussion. However, there is much more to this.

The amount of surge is directly related to both the mode and the amount of throttle so in some ways it is user controllable. Secondly the DSC system on the car should be smart enough to lessen the surge when cornering even if all of the other inputs call for a large surge. BMW has directly promoted a similar feature that automatically engages the disengaged clutch when cornering if engine drag creates a brake induced oversteer condition. I don't quite recall if the reduction of surge from shifting during cornering is an official capability of the car, can you quote anything on this? I know this has been discussed.

If the diff is operating properly and a straight line is being taken the surge should not cause any sideways forces on the car. If it does that is a problem.

Either way, the surge is still the fastest shift style when fractions of a second count.

[swamp2]

Originally Posted by grayshrk

Well, it turns out the reprogramming crashed while in process with my car at Irvine BMW. They've got it going again but it still has 4+ hours to run according to my SA. Looks like I won't be picking up my car tonight after all.

Mine had to stay overnight as well. I really can't believe the time it takes to send software. 5+ hours. Unreal. The possibilities to me seem straightforward:

1. Gigabytes of code.
2. A data transfer protocol/bit rate from the 1980s.

Neither of which seem likely. Oh well not really something to spend much time thining or worrying about despite how odd.

[KKSigs]

Got the update installed today, there is definately an improvement. The shifts are almost seamless in the lower S Modes with half throttle. In S5, it shift but not as dramatic when flooring it but still very fun. Also the initial lag or hesitation is gone when you first step on the pedal. Good work BMW.

[swamp2]

Originally Posted by KKSigua

Also the initial lag or hesitation is gone when you first step on the pedal.

I did not notice a change here in the responsiveness of the pedal during low to medium speed take offs. Again for clarity purposes I would not call this identified occurence "lag". I will have another "feel".

[grayshrk]

Originally Posted by swamp2

Mine had to stay overnight as well. I really can't believe the time it takes to send software. 5+ hours. Unreal. The possibilities to me seem straightforward:

1. Gigabytes of code.
2. A data transfer protocol/bit rate from the 1980s.

Neither of which seem likely. Oh well not really something to spend much time thining or worrying about despite how odd.

No doubt. When they told me it was an all day job I thought they were kidding or padding the estimate. There must be some serious validation testing taking place for it to take so long.

[Sticky]

Originally Posted by swamp2

Mine had to stay overnight as well. I really can't believe the time it takes to send software. 5+ hours. Unreal. The possibilities to me seem straightforward:

1. Gigabytes of code.
2. A data transfer protocol/bit rate from the 1980s.

Neither of which seem likely. Oh well not really something to spend much time thining or worrying about despite how odd.

Maybe they analyze line by line or bit by bit? Uploading the data at a high rate of speed probably isn't an issue, but calculating each line of code and analyzing as you are doing it probably takes some serious processing power.

[newtom3]

Originally Posted by Sticky

Maybe they analyze line by line or bit by bit? Uploading the data at a high rate of speed probably isn't an issue, but calculating each line of code and analyzing as you are doing it probably takes some serious processing power.

uh, no. the cpu power to calculate the correctness of data is not alot even for a ten year old controller. It probably is quite a bit of a code going over a slow serial link for reliability... you'd think they would have snuck in a 10/100 ethernet chip in there... its like $2.

[Sticky]

Originally Posted by newtom3

uh, no. the cpu power to calculate the correctness of data is not alot even for a ten year old controller. It probably is quite a bit of a code going over a slow serial link for reliability... you'd think they would have snuck in a 10/100 ethernet chip in there... its like $2.

I don't know, do we really know how many lines of code this ECU is? Maybe they download and analyze the program already there before uploading the new one? I don't think going over that much code is really that quick. I don't mean calculating if software installed is correct, but analyzing the logs and if anything has been changed probably takes a while.

Look at how long a decent virus scanner takes, even on a SSD.

[Dascamel]

With modern tech, 4 hours is a very long time given the amount of info involved. A virus scanner would scan that amount of data in minutes...