• Protip: Profile posts are public! Use Conversations to message other members privately. Everyone can see the content of a profile post.

Thoughts on the 9 speed transmission

Why can't the new NSX just have a CVT and be done with it?

I was at UC Davis years ago for unrelated business and if memory served me right they were working with Nissan and I think Audi on a CVT that was:

150lbs
can handle 900lbs TQ
Drivetrain loss between 5-9% (they were working on this at the time I was there)
Infinite gears

It was designed for an 18-wheeler.

Now I dunno what happened with the project, but we need this.

It was explained to me that the CVT, coupled with other tech can be the best form of launch control and TCS.

Example would be that if there is any slippage, a computer can simply go "up" to whatever gear ratios to slow or eliminate the slippage without getting off the throttle so that momentum isn't lost.


there's something mind numbingly boring about accelerating while the engine stays at it's most efficient RPM.
 
there's something mind numbingly boring about accelerating while the engine stays at it's most efficient RPM.

Very boring! Also the CVT on the V6 Camry achieved better 1/4 mile times when it has virtual 6 speeds (gear ratio) with shift points versus holding the engine RPM at one state.
 
It is possible, is it not, that a dual-clutch transmission has continuous torque transfer through a shift notwithstanding that the shift takes some finite time? Crudely speaking, engage the new gear then disengage the old one. It may be that nobody does this because of clutch-life issues, but it's not clear to me that there has to be a period of no torque transferred through the transmission as with a single-clutch transmission.

Currently... No. Even the best of the best: PDK, Ferrari, GT-R, etc all have some downtime between shifts. I'm not saying its impossible but we haven't quite seen it yet. If absolute 100% timeless shifts is something the NSX could do, it would be GROUNDBREAKING.

This is the whole point of the DTC. You have two different gears engaged at the same time, on two different clutches, and clutches swap status (engaged vs not). Sure, in comfort mode they are going to let off the throttle and ease things a bit, but in performance mode that's unnecessary. Per this article the 650S's 7-speed "even slightly overlaps the clutches above 5000 rpm in Track mode as an “'inertia push.'” Even "just" in sport mode all I perceived was continuous torque - no downtime between shifts when under heavy acceleration.

Most-likely explanation is the most straight-forward: They're simply achieving closer gear ratios by having more gears...which has very little penalty with DCT.

The two aspects of the transmission I'm most-interested in learning about:

Where is the electric motor in relationship to it? They've used terms like "direct drive" for the rear motor, which to me would imply it is on the transmission output shaft to directly drive the wheels (even without clutches or gears engaged, it's putting power down). However, video of Ted speaking off the cuff at Detroit auto show he says "between" engine and transmission. Is it part of the flywheel (like old-school Honda hybrids)? Etc.

Dry or wet clutches? Higher-horsepower cars generally require wet, but dry is more-efficient. Maybe part of Honda's secret sauce is making dry viable in this application (the motor is after the clutches, the clutches are sufficiently beefy / of the right compounds, and the close-ratio of the 9-speed reduces how hard the clutches have to work on under-power shifts). They've often been about getting the most from every horsepower generated, and this would be one way to do it.

Probably "no" on the second question, but fun to try to speculate what (if anything) they're doing outside the box with this that will be a surprise.
 
It is possible, is it not, that a dual-clutch transmission has continuous torque transfer through a shift notwithstanding that the shift takes some finite time? Crudely speaking, engage the new gear then disengage the old one. It may be that nobody does this because of clutch-life issues, but it's not clear to me that there has to be a period of no torque transferred through the transmission as with a single-clutch transmission.

You can't engage two gear ratios at once. It would lock the transmission....




.
.


KABOOM!!!


Although, I suppose if there were a way to instantaneously disengage the current gear upon simultaneously engaging the next (no time loss) then a "shift lag" of 0.0 seconds could be achieved. In other words, the engagement of one gear would have to simultaneously cause the disengagement of another gear.
 
Last edited:
This is the whole point of the DTC. You have two different gears engaged at the same time, on two different clutches, and clutches swap status (engaged vs not). Sure, in comfort mode they are going to let off the throttle and ease things a bit, but in performance mode that's unnecessary. Per this article the 650S's 7-speed "even slightly overlaps the clutches above 5000 rpm in Track mode as an “'inertia push.'” Even "just" in sport mode all I perceived was continuous torque - no downtime between shifts when under heavy acceleration.

Most-likely explanation is the most straight-forward: They're simply achieving closer gear ratios by having more gears...which has very little penalty with DCT.

The two aspects of the transmission I'm most-interested in learning about:

Where is the electric motor in relationship to it? They've used terms like "direct drive" for the rear motor, which to me would imply it is on the transmission output shaft to directly drive the wheels (even without clutches or gears engaged, it's putting power down). However, video of Ted speaking off the cuff at Detroit auto show he says "between" engine and transmission. Is it part of the flywheel (like old-school Honda hybrids)? Etc.

Dry or wet clutches? Higher-horsepower cars generally require wet, but dry is more-efficient. Maybe part of Honda's secret sauce is making dry viable in this application (the motor is after the clutches, the clutches are sufficiently beefy / of the right compounds, and the close-ratio of the 9-speed reduces how hard the clutches have to work on under-power shifts). They've often been about getting the most from every horsepower generated, and this would be one way to do it.

Probably "no" on the second question, but fun to try to speculate what (if anything) they're doing outside the box with this that will be a surprise.

I don't care about all that, continuous acceleration during gear change has not been achieved by ANY manufacturer. Are you suggesting McLaren road cars are technologically superior to F1 cars..?
 
You can't engage two gear ratios at once. It would lock the transmission....

No. it would just cause one of the clutches (or both) to slip. Like happens to a clutch during every shift where the revs are not perfectly matched. It's kind of the whole point of a clutch. And when it happens to much, it's what ends the life of a clutch. So it seems reasonable to think that there is a way of controlling the system to achieve the right balance.

I don't care about all that, continuous acceleration during gear change has not been achieved by ANY manufacturer. Are you suggesting McLaren road cars are technologically superior to F1 cars..?

F1 cars cannot use dual-clutch transmissions. So, it's quite plausible that McLaren road cars can have a lower time of no torque transmission. And it's quite possible for that time to be zero when using a dual-clutch transmission.
 
You can't engage two gear ratios at once. It would lock the transmission....
Unless someone decided to give them descrete inputs controlled by separate clutches. AKA a dual-clutch transmission, which we are talking about.

I don't care about all that, continuous acceleration during gear change has not been achieved by ANY manufacturer. Are you suggesting McLaren road cars are technologically superior to F1 cars..?

Like CVTs, DCTs are banned by F1. DCTs can (and do) actually achieve this. It has been achieved. The sequential transmissions on F1 cars are so advanced these days they effectively (though maybe not literally) also achieve it.
 
With embedded tech u can program the "feeling" of shifting.... I sold this tech in a past life..

which car with a CVT can I find this feel in? Genuinely asking since I've never experienced it.
 
More gears keep the engine in its peak operating range better = better performance. FWIW, many 7 speed dcts have 2 overdriven gear ratios for mpg. So essentially they are 5spd transmissions. If the nsx 9spd has 2 over driven gears, it'll be a true 7spd performance transmission.
 
No. it would just cause one of the clutches (or both) to slip. Like happens to a clutch during every shift where the revs are not perfectly matched. It's kind of the whole point of a clutch. And when it happens to much, it's what ends the life of a clutch. So it seems reasonable to think that there is a way of controlling the system to achieve the right balance.



F1 cars cannot use dual-clutch transmissions. So, it's quite plausible that McLaren road cars can have a lower time of no torque transmission. And it's quite possible for that time to be zero when using a dual-clutch transmission.

Unless someone decided to give them descrete inputs controlled by separate clutches. AKA a dual-clutch transmission, which we are talking about.



Like CVTs, DCTs are banned by F1. DCTs can (and do) actually achieve this. It has been achieved. The sequential transmissions on F1 cars are so advanced these days they effectively (though maybe not literally) also achieve it.


These are both false. The point of gear ratios is to turn the wheels at a speed relative to the engine. The wheels cannot turn two speeds at once. If both clutches are slipping, then either there is no power going to the transmission (it is free-spinning, meaning that the "instant" engagement of dual clutch transmissions is wasted), or it will bind up, since one object (out-put shaft/drive shaft/cv joint, etc) can only spin one speed at a time.
 
These are both false. The point of gear ratios is to turn the wheels at a speed relative to the engine. The wheels cannot turn two speeds at once. If both clutches are slipping, then either there is no power going to the transmission (it is free-spinning, meaning that the "instant" engagement of dual clutch transmissions is wasted), or it will bind up, since one object (out-put shaft/drive shaft/cv joint, etc) can only spin one speed at a time.

Both clutches slipping doesn't mean there is no power going to the transmission - just like with a manual car, where getting started involves a bit of clutch slip in order not to stall the engine and some power is still transmitted during slippage since the car gets moving. So in a DCT, as one clutch engages and the other disengages, they can both be partially engaged for a short period of time, and like when shifting a manual transmission, one or both clutches will slip a bit until the full switch from one clutch to the other is completed.

The momentum of the moving car is so great that I bet there still wouldn't be any binding if both clutches were fully engaged - instead, one or both clutch would slip so much that they'd very quickly burn up and fail, but there wouldn't be any broken transmission gear teeth or locked up drive wheels.
 
Both clutches slipping doesn't mean there is no power going to the transmission - just like with a manual car, where getting started involves a bit of clutch slip in order not to stall the engine and some power is still transmitted during slippage since the car gets moving. So in a DCT, as one clutch engages and the other disengages, they can both be partially engaged for a short period of time, and like when shifting a manual transmission, one or both clutches will slip a bit until the full switch from one clutch to the other is completed.

Exactly. Take a 1-2 upshift. Assume shift occurs at 6000RPM and wheel speed of those two gears at that RPM is 30 and 60mph, then evaluate these options...

A: Instantaneous disengagement of odd-shaft clutch at exact same time as engagement of even-shaft clutch.
B. Fully disengage clutch on odd shaft before engaging clutch on even shaft. There will be a brief interruption of power/acceleration to the wheels.
C. Overlap the disengagement and engagement of the clutches such that the odd clutch never slips due to engine power (RPMs going up) but does start slipping when the even clutch is engaged sufficiently to handle the engine power plus torque of slowing engine down from shift.
D. Fully engage even clutch before starting to disengage odd clutch.

Because each of the clutches takes some time greater than zero to be actuated, thus are not totally binary, option A cannot technically be done.

Option D seems like a bad idea. That is a lot of forces for the drive train and the clutches to handle, especially/specifically the clutches. You have one clutch trying to drive the wheels at 30MPH and the other trying to drive the wheels at 60MPH (or trying to drag the engine down to 3000RMP).

Option B is obviously a possibility (the only possibility if you are naaman or Krea.tion). Generally the computer would cut engine power briefly during that very short period so RPMs don't jump between gears. If power is cut long enough for engine RPMs to drop to match target gear, this is a means to achieve essentially zero clutch wear during shifts.

C is how you get uninterrupted acceleration. It is real. During this shift there will be a period of time where engine speed to wheel speed will not just be a ratio somewhere between these two gears, but will be an analog transfer/slide between the two ratios. This is why DCTs are banned in F1. This is what the 650S does in track mode and maybe what other DCTs already do as well (if not already, others will follow). With the even-shaft clutch trying to make the wheels go faster / engine go slower, than before, this type of shift would not just keep engine output to wheels from being continuous, but boost that wheel acceleration slightly (by the amount of torque necessary to slow the engine to get 2nd gear fully engaged). Obviously keeping the throttle at 100% while doing this as fast as you can is a harsh/performance thing. With different parameters (little slower actuation, some degree of throttle cut) overlap could provide things other than max performance, like smooth/comfort-oriented gear transitions without any or as much loss in acceleration between gears. The added clutch wear may keep this from seeing widespread application outside of max performance...even though doing it with a well-programmed system would likely mean less slipping and wear than the average manual transmission driver driving "normally."
 
Last edited:
Wow
Reading latzke's post and considering all the variables and performance results available I can why the manual shifted gearbox is either not being used or being phased out of high perfomance vehicles.
The slow shift speeds of the manual make it uncompetitive.
 
Last edited:
So, I have been discussing the new NSX with some coworkers and we have decided a couple things.
The drive train will probably have no gaps in power delivery. The electric motors will make torque from 0 rpm and should give a boost to torque through out the rev range. The turbos will likely be lower boost to reduce lag and to give it a nice even delivery of torque. We feel that the engine is not about revs (seems like 7500 is what the redline will be), instead it is probably closer to the new m3/m4 in power delivery. We also thought about how the 9 speed was going to be used. The sheer number of ratios could just be to keep the revs in the sweet spot, but with the broad torque curve, why have such close ratios? It seems like it would almost slow it down to have them be too close together. It seems likely that the first two or three ratios could be very low to get it to 60 quickly with the next three or so a bit wider to take advantage of the torque. That leaves the last couple gears for over drive, but why have two? There is always the chance that Honda spaced them so that this car tops out at 60 in second gear. If they did, I bet the next several gears will be at 80, 100, 120, 140, 160, 180, 200 or something like that. Even with a high final drive, there would be no gaps in power or dips into the powertrain's "dead zones" and it would still have a high top speed and a decent cruising gear.
 
Both Ford and GM are planning 10 speed automatics - suddenly a 9 speed DCT doesn't sound excessive!

That's a lot of moving parts. There has to be a formula to see the ratio of moving parts to possible failure. One of my early cars was a 70 Nova with a slip and slide powerglide. Fun to drive and pretty simple by comparison. I really just don't see the point of the extra gears. I get the closer gear ratios and OD gearing but you could do that with a six speed. Is it all about the "more is better" mentality that we Americans have (and I know Americans don't have a patent on that)? Maybe I'm just getting old and cranky. My wife has been calling me a Curmudgeon lately. :0
 
Exactly. Take a 1-2 upshift. Assume shift occurs at 6000RPM and wheel speed of those two gears at that RPM is 30 and 60mph, then evaluate these options...

A: Instantaneous disengagement of odd-shaft clutch at exact same time as engagement of even-shaft clutch.
B. Fully disengage clutch on odd shaft before engaging clutch on even shaft. There will be a brief interruption of power/acceleration to the wheels.
C. Overlap the disengagement and engagement of the clutches such that the odd clutch never slips due to engine power (RPMs going up) but does start slipping when the even clutch is engaged sufficiently to handle the engine power plus torque of slowing engine down from shift.
D. Fully engage even clutch before starting to disengage odd clutch.

Because each of the clutches takes some time greater than zero to be actuated, thus are not totally binary, option A cannot technically be done.

Option D seems like a bad idea. That is a lot of forces for the drive train and the clutches to handle, especially/specifically the clutches. You have one clutch trying to drive the wheels at 30MPH and the other trying to drive the wheels at 60MPH (or trying to drag the engine down to 3000RMP).

Option B is obviously a possibility (the only possibility if you are naaman or Krea.tion). Generally the computer would cut engine power briefly during that very short period so RPMs don't jump between gears. If power is cut long enough for engine RPMs to drop to match target gear, this is a means to achieve essentially zero clutch wear during shifts.

C is how you get uninterrupted acceleration. It is real. During this shift there will be a period of time where engine speed to wheel speed will not just be a ratio somewhere between these two gears, but will be an analog transfer/slide between the two ratios. This is why DCTs are banned in F1. This is what the 650S does in track mode and maybe what other DCTs already do as well (if not already, others will follow). With the even-shaft clutch trying to make the wheels go faster / engine go slower, than before, this type of shift would not just keep engine output to wheels from being continuous, but boost that wheel acceleration slightly (by the amount of torque necessary to slow the engine to get 2nd gear fully engaged). Obviously keeping the throttle at 100% while doing this as fast as you can is a harsh/performance thing. With different parameters (little slower actuation, some degree of throttle cut) overlap could provide things other than max performance, like smooth/comfort-oriented gear transitions without any or as much loss in acceleration between gears. The added clutch wear may keep this from seeing widespread application outside of max performance...even though doing it with a well-programmed system would likely mean less slipping and wear than the average manual transmission driver driving "normally."

In a traditional transmission, engaging two gears at once prevents the transmission from turning at all.

I won't claim to know the details of DCT, and if I'm totally wrong, so be it. But the explanation above doesn't account for how uninterrupted power can be transferred through the same output shaft from two different gears.

If a clutch is "slipping" it means that power is partially being transmitted, but the ratio is not in effect at this point. There are more clutch rpms than there would be if the clutch were fully engaged. Based on what I understand, the gear itself is being (partially) powered by the engine, but that power is not being transmitted to the output shaft, since the lower gear is still engaged. The DCT cannot put full power to two gears at once. That is all I was trying to say.

Now, if the higher gear's clutch is slipping, it makes sense that the engine rpm would not be forced down by the higher gear, since it is not fully engaged. If the computer can calculate the optimum time to fully disengage the lower gear, then time lost between gears is minimized, but probably not truly 0.

In any case, the engine can only turn one speed at a time. Same for trans (output) and same for wheels/axles/driveshaft. If there were some way to do that thing that quarks do (the positive one flips at the same instant as the negative one... or something... regardless of the distance between them), then we could possibly see a truly instantaneous gear change (unless the engine can't slow down fast enough...?).
 
The brochure for the McLaren sport series very clearly asserts that McLaren uses an overlapping clutch transition to not only eliminate dead time in shifts but also to take advantage of the inertia of the geartrain inertia:
McLaren_upshift.jpg
... as posted here.
 
Back
Top