An Old DCT Lag Rant

[Chriskm3]

Have any of the dct drivers been faced with extensive repair bills yet?

[sensi09]

Originally Posted by swamp2

Medium, even speed, perhaps in the 40-60 mph range, gentle to medium (street level) braking, then followed by a medium throttle applied rapidly and immediately after letting off the brake. I've experienced lag which can be about 1 full second. The car simply feels as if it is completely ignoring the throttle application (and it really is).

Going back to the issue of lag, I tried this in D4 and the car doesn't take off immediately, but I wouldn't exactly call it lag and definitely not dangerous. If I skip the braking and accelerate (while in 7th gear in D4), the car does respond much faster though. Only tried this a few times.

One should question how hard you're braking and does DCT downshift during braking.

Someone already pointed it out, but was also wondering if the car is just downshifting:

Originally Posted by SenorFunkyPants

I know its stating the obvious....but if you are in D mode, decelerating, braking and reapplying the throttle then surely a significant number of times the reapplication of throttle will coincide with the DCT making a downshift (and thus a moment of throttle lock out)?

With the car in the middle of downshift and perhaps with DCT adapting to a very leisurely driving style, this could all accumulate into lag.

[sensi09]

Originally Posted by Chriskm3

Have any of the dct drivers been faced with extensive repair bills yet?

Off the top of my head, one of the supercharged guys was due for a new transmission after abusing the car at the drag strip. I'm sure there are others, but replacement prices are not as bad as originally thought.

[e92M3C]

Originally Posted by Chriskm3

Have any of the dct drivers been faced with extensive repair bills yet?

I can only recall a one-off incident where one 2008 M3 DCT completely failed. You can do a search.

Most of the problem with DCT lag has to do with the software. If your car is 2008-2009, you can get a DCT update from dealer after some good amount of complaints. Otherwise, you can get the DCT software reflash from Mike Benvo. I got the Euro DCT software where it starts with D instead of S. Also resolved the cold-start high-rpm shift jerk issue for me.

I think the engine will be blown first before the DCT goes dead from all the abuse.

[Chriskm3]

Originally Posted by e92M3C

I can only recall a one-off incident for 2008 where a member's DCT failed. You can do a search.

I think the engine will be blown first before the DCT goes dead from all the abuse.

Originally Posted by sensi09

Off the top of my head, one of the supercharged guys was due for a new transmission after abusing the car at the drag strip. I'm sure there are others, but replacement prices are not as bad as originally thought.

I heard many dct drivers complaining about end of warranty jitters about the dct tranny. I'm glad no major issues were reported. Supercharged cars fall into another category.

[sensi09]

I think he was trying to "launch" the car without using launch control...done multiple times if I'm not mistaken. With a non SC car, the results would likely be the same.

[bruce.augenstein@comcast.]

Originally Posted by swamp2

Yes, really...It comes down to shift time reduction. If you assume the DCT can shift in about 50-100 ms which is a very reasonable number supported by a large variety of evidence, along with the fact that a good MT shift is probably in the 0.4 s range, just do some simple math. You save about 0.6 seconds over two shifts in the quarter mile. The math really is that simple time saved shifting directly translates to just about the same time spent over a certain fixed distance. For a 3700 lb car use the basic formula for a 1/4 mi time:

ET = 6.1178 (weight/hp)^1/3 (ref:http://www.stealth316.com/2-calc-hp-et-mph.htm , weight in lb, hp obviously as hp)

And you will find almost a 63 hp requirement to have this much gain!

Swamp, you need to go at this a little differently. Assume a 12.4 @ 114 timeslip for the M3 DCT, and I'll stipulate a .05 second shift speed for it. You in turn should allow me to stipulate a .20 second shift speed for the manual car. I say this because we ran tests with an accelerometer way back when on my '93 Vette, and found two tenths to be repeatable, with more than one shift coming in at around .18 seconds. This was with just a very minor clutch "dab" during shifts, and I of course admit it was abusive to the machinery.

So, three shifts over a quarter mile equals a .45 second advantage for the auto, meaning a 3.6% disadvantage in terms of actual acceleration time for the stick car in this venue.

Because the shifts happen early, middle and late during the run, I'll call the .45 second DCT advantage as averaged over the acceleration period. Therefore, because the DCT car is under power for 3.6% longer, think of this as an average added power of 3.6% during the run.

Now, remember that trap speed (and ET) tend to vary as per the cube root of the power to weight delta. With this in mind, figure the DCT will have a 1.2% performance advantage, so a stick car might turn a 12.55@112.65.

Of course, the stick car is lighter, so that further diminishes the difference. Therefore the close racing we looked at a bit earlier in this string.

Originally Posted by swamp2

...Similar advantages exist on a track due, especially when 3rd gear is used a lot, which on many track it is. There is 12% more torque to the wheels comparing both cars in 3rd gear simply because of the gear ratios.

back at ya...

Swamp, you know better. This will be a seesaw battle between the transmissions. When the DCT is in third, sometimes the stick car will be in second, and when the DCT hits fourth, the stick car will still be in third. It's not really possible to predict, and so far, there's been damned little evidence that the auto is quantifiably quicker in an on-track venue.

Bruce

[swamp2]

To all those interested in the OT performance advantage of M-DCT vs. 6MT and in the very closely related issue of shift times:

http://www.m3post.com/forums/showthr...=582295&page=4

A bit tough to track the main thrusts from the players in that debate (myself included of course). In short some of the supercharged M3 drag racer guys felt that M-DCT could be good for an equivalent 50 hp and 10 mph trap speed advantage in the 1/4 mi. I actually argued strongly against that. However, again, it depends on exactly which contest you are trying to compare, 0-60, 60-130, 1/4 mi. time or trap, time to speed, time to distance, etc. In that discussion I actually began using some 0.2 (200 ms shift times). The drag racer guys insist from vbox data that 0.6 seconds is much more realistic. There is no one right answer here but I am fairly convinced most non-pro folks cannot consistently shift a production 6MT E9X M3 car in 0.2 seconds, certainly not without powershifting.

To ascribe an equivalent power gain due to M-DCT is absolutely reasonable however, there is no one single best number for it. Bruce's rough estimate above seems to work out to 15 hp (3.6%). An absolutely critical factor in this is the shift time difference. The best estimates for the M-DCT are in the 50-100 ms range. My prior quick "back of the envelope" calculation showing a 60 hp advantage was not supposed to represent my best guess for the M-DCT advantage, just to show that there is one and how large it might be.

Here is a reasonably large performance database of magazine test results:

http://www.m3post.com/forums/showthr...=70737&page=15

Other than a single outlier in 2 contests the M-DCT cars are overall faster than the 6MT cars in the quarter mile and in many other contests.

The best way too look at this is actually physics based simulations. The formulae Bruce and I have used above are decent approximations for some very quick "back of the envelope estimates". Better than those are a real simulator which is ideal for this type of problem, both because it captures the physics but also, very importantly because it holds ALL variables absolutely constant between the runs (such as temperature, tires, launch, weight, shift times (and other driver effects), parasitic losses, etc.). Below are simulations for a stock E92 M3 M-DCT vs. 4 different 6MT cars, equal power and then +20 hp, +40 hp and +60 hp.

These results are based on a 50 ms M-DCT shift time and 400 ms 6MT shift time. They indicate, that depending on the contest, the M-DCT is good for somewhere between an equivalent 20-50 hp over the 6MT car. If you (or some theoretical you) can consistently shift in 200 ms this equivalent hp advantage figure will probably drop to about VERY ROUGHLY about half of this. I certainly could have run those simulations, but felt a bit lazy.

Anyone should recall this basic fact - it take a lot of hp to make relatively small changes in ETs. This along with shift time savings for the M-DCT is the "equivalent hp" idea and it takes quite a bit of it to account for relatively small time savings.

I think am truly done with this OT debate here. Hopefully, some fellow members can continue my effort in leading the thirsty to the river on this particular point.

[CSBM5]

The basic issue is your assumption of a 0.4 second 6MT shift speed. Seriously, that is a joke. I could teach a 15 year old to shift faster than that within a hour of so of teaching how to drive a stick. Nobody who has any business trying to drive a car at the strip will shift anywhere near that slow. Try cutting that in half.

[swamp2]

Originally Posted by CSBM5

The basic issue is your assumption of a 0.4 second 6MT shift speed. Seriously, that is a joke. I could teach a 15 year old to shift faster than that within a hour of so of teaching how to drive a stick. Nobody who has any business trying to drive a car at the strip will shift anywhere near that slow. Try cutting that in half.

I addressed that concern fully, try not to lose sight of the forest for the trees,

1. A shift is not the time to move the lever. It is the time from the moment acceleration stops and then to the point when the acceleration has fully resumed.
2. Some guys with vboxes and a ton of drag and 60-130 runs under their belts believe the 0.6 seconds is a good shift. Argue with them.
3. I do not believe that any ammetuer drag racers can consistently shift their E9X M3's in 200 ms. Let's get some data if we want to continue this part of the debate.
4. Even if one could consistently shift that fast, the DCT should still be good for as much as a 25 hp equivalent advantage over the MT.

[CSBM5]

Originally Posted by swamp2

I addressed that concern fully, try not to lose sight of the forest for the trees,

1. A shift is not the time to move the lever. It is the time from the moment acceleration stops and then to the point when the acceleration has fully resumed.
2. Some guys with vboxes and a ton of drag and 60-130 runs under their belts believe the 0.6 seconds is a good shift. Argue with them.
3. I do not believe that any ammetuer drag racers can consistently shift their E9X M3's in 200 ms. Let's get some data if we want to continue this part of the debate.
4. Even if one could consistently shift that fast, the DCT should still be good for as much as a 25 hp equivalent advantage over the MT.

Your simulation from Cartest2000 shows the 6MT car trapping at 109mph which is a total farce. The only way it can trap that speed if something serious is wrong with the engine. Trap speed is about the best indicator of power to weight ratio there is, far better than ET, and a normal, healthy M3 will trap in the 112mph range or a bit higher. Heck, in my S62 powered M5, it's trapped at almost 109mph (108.56), and the M3 is a good bit faster than it.

There is no possible way that you are going to see real world differences of 4mph in trap speed between these two cars when the 6MT is actually driven properly. Simulations are fun (I've used Cartest for many years going back to when it was a DOS program), but real world data is much more worthwhile as the model is only as good as it's inputs.

In any event we might have a fun event this Spring sometime if a potential private rental goes through...should be a blast to gather real world data under the same conditions using the same rear wheels/tires.

[swamp2]

Originally Posted by CSBM5

Your simulation from Cartest2000 shows the 6MT car trapping at 109mph which is a total farce.

That is a fair point about the trap speed but ultimately it has no bearing on the primary point about the equivalent power that the M-DCT offers. The power of any simulation tool is much more for differences than absolute values and that is the primary point of the exercise I posted above. I've had some long term issues with getting perfect correlation on trap speeds. The ETs are always much closer. I am using a 12% drivetrain loss which is consistent with the most accurate public hub dyno I know of at rri.se.

One quick CarTest experiment I ran was to change the 6MT M3 shift time to be 50 ms faster. That increased the trap speed by almost exactly 1 full mph. It made a larger difference than I thought it would.

Originally Posted by CSBM5

There is no possible way that you are going to see real world differences of 4mph in trap speed between these two cars when the 6MT is actually driven properly.

If the cars are driven equivalently (as much as that is possible between a MT and M-DCT) and the shift times assumptions are adhered to you will find a 4 mph difference. I do think that 2-3 mph is a better observed difference. Maybe that means that most 6MT test drivers are shifting closer to 300 ms and the M-DCT shift times are closer to 100 ms.

Again, for the hundredth time, the concept here remains completely solid, the M-DCT is faster because of shift time reduction and that can be thought of as equivalent hp. It is not an insignificant amount of equivalent power.

A good side discussion but again, let's please get back to the lag discussion... We've probably lost any of the folks who might participate in that converstation...

[CSBM5]

We can agree to disagree I suppose. On a stock engine car, DCT will not give you 4 mph over a 6MT. All published test results such as C&Ds, for example, so at best a 1mph benefit to DCT although their results show no benefit in trap. All their M3 traps, 2x6MT and 2xDCT were within 1mph of each other at 113-114mph. I'll definitely agree you might eek out 2-3 tenths in ET if you can launch both just right (stock engines), but you're simply not going to gain 4mph trap on stock S65 M3s.

[US///M3]

I'd be interested to see some road course times between the two trannies.

[sensi09]

Originally Posted by swamp2

A good side discussion but again, let's please get back to the lag discussion... We've probably lost any of the folks who might participate in that converstation...

There were a few posts regarding your lag issue in case you missed them.

[luis_m3]

I'll chime in...
I have a lag on mine as well, but it's not bad at all (barely noticeable). I already got used to it and doesn't bother me as I do not do much city driving anyways.

[swamp2]

Originally Posted by sensi09

There were a few posts regarding your lag issue in case you missed them.

Somewhere other than here? Thanks.

[swamp2]

Originally Posted by CSBM5

We can agree to disagree I suppose. On a stock engine car, DCT will not give you 4 mph over a 6MT. All published test results such as C&Ds, for example, so at best a 1mph benefit to DCT although their results show no benefit in trap. All their M3 traps, 2x6MT and 2xDCT were within 1mph of each other at 113-114mph. I'll definitely agree you might eek out 2-3 tenths in ET if you can launch both just right (stock engines), but you're simply not going to gain 4mph trap on stock S65 M3s.

Again, forest, trees...

My point has never been about an exact number. There is no such thing as an exact number, for ET gain, trap gain, hp gain, shift times, etc. There is also no such thing as exact real world test numbers either! Environmental, driver and individual car conditions vary too much. It is a GENERAL PRINCIPLE here. In my very first reply I was only "hitting" back at the thread polluters who were baiting (it worked) with all of the typical "should have got a MT" BS. In that original thread I stated the a M-DCT is good for an equivalent 25-50 hp depending on the contest. You've now "given me" a couple tenths, I think 4 tenths is more reasonable. Hopefully we can finally agree on the big picture and similarly may just be able to agree that there is not one right answer here.

[CanAutM3]

Originally Posted by swamp2

A shift is not the time to move the lever. It is the time from the moment acceleration stops and then to the point when the acceleration has fully resumed.

If we use this as the definition of a shift, I am still questioning why we would not consider the DCT shift time as zero ?

I fully comprehend that there is an acutal time required to engage one clutch while the other disengages and therefore a period of time where none of the clutches is fully engaged. This is essentially the DCT shift time and could well take 50-100ms depending of the shift mode.

However, for the purpose discussed here, wouldn't the two clutches slip simultaneoulsy and therefore never interupt the drive force to the wheels (especially in the more aggressive modes)?

Sorry for bringing this thread off topic again

[CanAutM3]

Originally Posted by Vic311

Very plausible..if you consider the fastest shifting transmission in a production car is now the Aventador @ 50 ms (.05 seconds), an F1 car does it in 40 ms (.04 seconds)

I'm not sure where the DCT stands with shift speeds, but I'm sure its no where near a Lambo or an F1 car

Both the Lambo and F1 cars use single clutch designs. There is the inevitable time required to disengage the clutch, change gears and re-engage the clutch. 40ms to do that is quite amazing.

You cannot compare this with a double clutch setup.

[CanAutM3]

Originally Posted by Vic311

My reasoning is not flawed and I am certainly not the one making a fool out of themselves.

You stated the DCT is good for 25-50 HP. No matter how much subjective data and graphs you try to support this absurd statement and convince yourself.. it doesn't make it so.

I guess a way to look at this is to consider that a drivetrain produces work over the entire quarter mile run. The more work is generated during that run, the greater the trap speed and the shorter the travel time. More horsepower allows more work to be produced in a given amount of time.

However, no work gets generated during a shift, since no force is being applied during that brief instant. The longer the shift, the bigger the work deficit.

Swamp's conclusion is that, to match a DCT M3, a 6MT M3 would need to have 25-50 more hp to compensate for the work it is not producing during the longer shifts.

[swamp2]

Originally Posted by CanAutM3

If we use this as the definition of a shift, I am still questioning why we would not consider the DCT shift time as zero ?

I fully comprehend that there is an acutal time required to engage one clutch while the other disengages and therefore a period of time where none of the clutches is fully engaged. This is essentially the DCT shift time and could well take 50-100ms depending of the shift mode.

However, for the purpose discussed here, wouldn't the two clutches slip simultaneoulsy and therefore never interupt the drive force to the wheels (especially in the more aggressive modes)?

Sorry for bringing this thread off topic again

I did some testing of M-DCT shift times and posted the data here long ago. The data was not very consistent nor high quality and noise was an issue. I never claim in these type of discussions that the M-DCT can shift in zero time. However, some data, in some modes indicated that there may be some zero time shifts where the vehicle acceleration is not interrupted but actually surges during the shift with a very carefully orchestrated clutch overlap period and we feel this as a surge.

I think the best estimate, based on a variety of sources is 50-100 ms for a typical aggressive upshift. Also, shift times will vary depending on the Drivelogic mode and the level of agressiveness in the current driving.