Which Acuity parts needed to just remove lateral slop on shifter?

[Victorofhavoc]

Nah


Nah I don't believe in changing my driving style as a patch for that. I'll accept the rev hang (really more like overboost) in the meantime.

The turbo lag is acceptable to me, and that can be mitigated with a flat foot shifting tune option anyway.

Induced rev hang just annoys me.

I've driven some TC racecars and they had rev hang as well. It's not as noticeable because the dual clutch and dog trans setups just slap you into gear faster than the throttle can hang, but you feel it on the lack of engine braking for 200ms or so. My 370z racecar also had rev hang; with a lightweight flywheel, carbon driveshaft, 6 puck clutch, and no slave cylinder it was only noticeable in neutral. If you tried to totally tune out the hang it would throw codes on weird things like crank, cam, and other things. "Rev hang" is also technically how Porsche does their antilag setup in their pdk cars. They keep the throttle plate open, revs going, and turbo spinning so that back on throttle is a touch more immediate.

You won't mitigate really any turbo lag with ffs. You're just keeping the boost up between upshift. On downshift and any throttle on/off the lag will always be very present. That's why on track a turbo car you must do left foot braking to build boost at apex, while an NA car will just coast for a few feet and lfb doesn't add anything.

Sponsored

 

[ashmostro]

That's a cool breadth of cars you've had experience with. I have similar experiences and it's so interesting driving cars with totally different drivetrain configurations, isn't it? So much fun!

I agree with your comments on ffs scenarios. There's nothing you can do about downshifts and lift off-on. But it does help with maintaining constant acceleration on straights which can mean a meaningful lap time increase if the straight is on the shorter side. Anyway, IMHO this is all very detailed and not that important compared to whether the car feels satisfying to drive, which is subjective as well.

To that end, lag doesn't bother me that much. Rev hang pisses me off. And I like a consistent friction point. hehe.

 

[Spart]

So, with those needs in mind, I want a predictable engagement point but with as smooth a vibrational feel as possible.

You may be giving the CMC damper more credit than it deserves.

I can't say that I notice any NVH difference before/after. If I really focus on it when stationary, I can maybe feel a little bit of something that wasn't there before. But it's imperceptible when the car is moving.

Hydraulic clutches, by their very nature, already quell a lot of high frequency vibration and NVH vs a more direct linkage like a cable. And cars with hydraulic clutches did without any kind of "damper" for ages (and most modern cars do not have these.)

I really think it's a solution in search of a problem, and it just causes more problems than it's worth.

 

[ashmostro]

You may be giving the CMC damper more credit than it deserves.

I can't say that I notice any NVH difference before/after. If I really focus on it when stationary, I can maybe feel a little bit of something that wasn't there before. But it's imperceptible when the car is moving.

Hydraulic clutches, by their very nature, already quell a lot of high frequency vibration and NVH vs a more direct linkage like a cable. And cars with hydraulic clutches did without any kind of "damper" for ages (and most modern cars do not have these.)

I really think it's a solution in search of a problem, and it just causes more problems than it's worth.

I don't disagree at all. it probably isn't necessary but given I'd rather have a more damped than less damped response so long as is doesn't appreciably slow re engagement, it doesn't feel worth the cost to remove it (to me at least).

 

[Spart]

I don't disagree at all. it probably isn't necessary but given I'd rather have a more damped than less damped response so long as is doesn't appreciably slow re engagement

Based on my data, it seems to delay re-engagement by 30ms.

 

[ashmostro]

That's pretty small. What's the data say on the CDV delay time?

 

[Victorofhavoc]

Based on my data, it seems to delay re-engagement by 30ms.

How did you measure 30ms?

That's an insanely tight margin, and if you could feel that at your foot then it's time to go racing tomorrow, Mr. Andretti. I pride myself on being fast and reflexive, and under 60ms I have a near impossible time telling variation. 60ms is a dsg fart on a bad day.

 

[Spart]

How did you measure 30ms?

That's an insanely tight margin, and if you could feel that at your foot then it's time to go racing tomorrow, Mr. Andretti. I pride myself on being fast and reflexive, and under 60ms I have a near impossible time telling variation. 60ms is a dsg fart on a bad day.

WIth 1000fps high-speed video.

30ms does not sound like a lot. But when you're moving rapidly, it will affect you more than you'd think.

Give a FPS gamer 30ms of lag and they will have appreciably worse performance. Same thing goes for trying to nail a perfect shift, all of these little things add up.

 

[Victorofhavoc]

WIth 1000fps high-speed video.

30ms does not sound like a lot. But when you're moving rapidly, it will affect you more than you'd think.

Give a FPS gamer 30ms of lag and they will have appreciably worse performance. Same thing goes for trying to nail a perfect shift, all of these little things add up.

30ms of lag on a network is very different from 30ms of lag on a physical feeling. Frame drops are something else entirely, and there's a much more cascading effect there. Your human brain has a reaction time of 100-120ms at best. Even the best drivers out there struggle to get under 100ms for a queue-to-reaction rate.

I agree you can technically feel a 30ms delay, but given all the other inputs happening in a performance driving scenario (and I know because I've raced quite a few types of cars) you won't feel it. Realistically, the average driver will take over half a second between shifts. Even the fastest shifters out there can't get under 200ms on dog boxes. Your brain and input adjustment can easily take 30ms into the equation, and it won't appreciably affect anything.

 

[Spart]

That's pretty small. What's the data say on the CDV delay time?

I want to stress that I don't have individual numbers on only CDV deleted and only CMC damper deleted. Just did not have the time or interest in bleeding my clutch system three times instead of once. I'm just interpreting the data based on differences that make sense based on how each device works.

Ignoring the initial delay (which I attribute to the damper and not the CDV) the amount of time it takes from the time the clutch fork starts properly moving at full speed to reach full engagement with the CDV was 99ms. After the delete, that component of the engagement time dropped to 62ms.

So while the CDV+damper delete dropped the delay to full-speed clutch fork movement nearly in half, it dropped the rate of engagement once motion begins by 37%.

 

[Spart]

30ms of lag on a network is very different from 30ms of lag on a physical feeling. Frame drops are something else entirely, and there's a much more cascading effect there. Your human brain has a reaction time of 100-120ms at best.

People conflate reaction time with the ability to perceive short-duration events. The two are unrelated.

For proof, watch a video with one frame every 100ms.

Humans easily have the ability to perceive events lasting approximately 10ms, despite not being able to react to them in 10ms. It's just two different things.

 

[ashmostro]

People conflate reaction time with the ability to perceive short-duration events. The two are unrelated.

For proof, watch a video with one frame every 100ms.

Humans easily have the ability to perceive events lasting approximately 10ms, despite not being able to react to them in 10ms. It's just two different things.

I agree with this completely. I mean, it absolutely must be true or humans wouldn't be able to play a single sport effectively.

That said, 30ms is pretty small from a power delivery interruption standpoint - small enough that it would need to be a cheap mod to be worth the squeeze. When you start to get to the 100ms regime, things change dramatically in my experience.

 

[Victorofhavoc]

People conflate reaction time with the ability to perceive short-duration events. The two are unrelated.

For proof, watch a video with one frame every 100ms.

Humans easily have the ability to perceive events lasting approximately 10ms, despite not being able to react to them in 10ms. It's just two different things.

The average movie is at 24 fps. That's a lot more than 10ms. We don't perceive it as a slideshow. Yes you can reduce motion blur by jumping to 60fps, but beyond that is nearly impercitble to most. (gaming is a different story because frame drop is very real and that behaves entirely differently when you're running 100fps on a 120hz screen). As far as I'm aware the fastest perception rates have been around the 13-16ms range, but even then the brain is interpolating data and if something happens "outside of the bounds of anticipated physics" the perception range drops dramatically to 40ms+. Meaning if you watch a 60 fps video of a ball bouncing, and for 2 frames the ball changes direction, your brain won't interpret the change. They've done this study where the ball entirely disappears, and no one notices because the brain fills in the details.

At the end of the day, even if the reaction rate were to drop by 30ms, while. You MIGHT be able to perceive it, you certainly would not be able to react to it. To that end, adjust your anticipation is a more effective approach than looking for an unattainable reaction rate.

 

[Spart]

Even the fastest shifters out there can't get under 200ms on dog boxes. Your brain and input adjustment can easily take 30ms into the equation, and it won't appreciably affect anything.

Okay, I want to dive into this and break this down in detail, lest somebody read that and believe it.

For reference, I timed a lot of my shifts on track with the GT350 in the 260-300 ms range. I'll use 260 ms since it's seemingly as fast as my mechanical sympathy lets me shift that gearbox. Haven't tracked the ITS yet, not sure if I will. It feels like it could be similar in terms of shift speed, but I don't have as much seat time on it as I did the GT350 and I don't have the data, so I won't make up a number.

In a shift, you're clutching in and out. So that 30 ms that I'm talking about is happening twice.

But it's actually worse: 30 ms is the difference in delay that I measured. More exactly, with the CDV and CMC deleted there is about 32 ms of delay, and with them in place there is 63 ms of delay.

In other words, there's 32 ms of delay inherent to the system and an additional 31 ms of delay with the CDV and CMC damper in place. But that total delay is still happening - with the CDV and CMC damper, you have 63 ms of delay.

And again, that's happening twice each shift.

You also aren't continuously moving the clutch pedal, there's going to be a brief moment where you bottom out the pedal even if it feels like you never stop. Let's just call it another 30 ms. I don't know what it is and it will vary every time, so just humor me with the 30 ms figure.

So if you're trying to hit a 260 ms shift like I'm capable of - you have 62 ms of delay one-way, 30 ms of dead time at the bottom, and 62 ms of delay the other way. 154 ms out of 260 ms that nothing is happening even though you're "actively" working the clutch.

Well, 260 ms - 154 ms is 106 ms left over to actually get the engagement and disengagement done. 106 ms sounds like a lot, but no. My high-speed data shows that before the CDV and CMC damper delete, it takes a total of 99 ms for the clutch to fully engage if you omit the 62 ms delay (161 ms total.) But that is the one way travel time for the clutch fork. Let's assume that you can clutch in as fast as my data shows the clutch moving after the delete, which is 62 ms instead of 99 ms.

So from the time you start pressing the clutch pedal, you have:

• 62 ms of delay
• 62 ms of disengagement
• 30 ms of dead time at the bottom of travel
• 62 ms of delay once your foot starts going the other direction
• 99 ms of engagement (slower than disengagement because of the CDV)

Total = 315ms.

So if you *try* to shift any faster than 315 ms with the factory setup, you will get some amount of unintended clutch slip. There's basically no way around it.

And unintentional slip is a big complaint on these cars with fast shifting (any car with a CDV, really) for this exact reason. With the slow engagement rate, you just can't effectively work the clutch as fast as you can work everything else. The clutch becomes the weak link and you can't help but unintentionally slip the clutch.

Now let's run those numbers again after the CDV and CMC damper delete.

• 32 ms of delay
• 62 ms of disengagement
• 30 ms of dead time at the bottom of travel
• 32 ms of delay once your foot starts going the other direction
• 62 ms of engagement.

Total = 218 ms.

Needless to say, 315 ms down to 218 ms is a big difference, and it puts the speed of clutch engagement within my envelope as a driver, whereas before I struggled a bit to get good, consistent rapid shifts.

This is to say nothing of the element of timing. Timing is really important for a good shift no matter how fast you're shifting. You want the clutch to engage at the exact right instant, and the more delay there is in the system the less intuitive this becomes. Humans are actually very good at timing things within tens of milliseconds - look no further than musicians for this sort of precision. Again, reaction time has nothing to do with your ability to sync to small time frames or perceive them.

And, having actually experienced this car before and after, that difference in timing is stark. It's more intuitive and the clutch engages harder and more predictably. It made driving the car in everyday situations a lot more forgiving and enjoyable, and when I get more aggressive with my shifts it made a huge difference.

I hope for people reading this, that this is a lot more enlightening than something so dismissive as:

I agree you can technically feel a 30ms delay, but given all the other inputs happening in a performance driving scenario (and I know because I've raced quite a few types of cars) you won't feel it. Realistically, the average driver will take over half a second between shifts.

People have been deleting Honda CDVs for a long time now. The design hasn't changed much in two decades. The CMC damper is just another thing nerfing the system for the lowest common denominator.

 

[Victorofhavoc]

Okay, I want to dive into this and break this down in detail, lest somebody read that and believe it.

For reference, I timed a lot of my shifts on track with the GT350 in the 260-300 ms range. I'll use 260 ms since it's seemingly as fast as my mechanical sympathy lets me shift that gearbox. Haven't tracked the ITS yet, not sure if I will. It feels like it could be similar in terms of shift speed, but I don't have as much seat time on it as I did the GT350 and I don't have the data, so I won't make up a number.

In a shift, you're clutching in and out. So that 30 ms that I'm talking about is happening twice.

But it's actually worse: 30 ms is the difference in delay that I measured. More exactly, with the CDV and CMC deleted there is about 32 ms of delay, and with them in place there is 63 ms of delay.

In other words, there's 32 ms of delay inherent to the system and an additional 31 ms of delay with the CDV and CMC damper in place. But that total delay is still happening - with the CDV and CMC damper, you have 63 ms of delay.

And again, that's happening twice each shift.

You also aren't continuously moving the clutch pedal, there's going to be a brief moment where you bottom out the pedal even if it feels like you never stop. Let's just call it another 30 ms. I don't know what it is and it will vary every time, so just humor me with the 30 ms figure.

So if you're trying to hit a 260 ms shift like I'm capable of - you have 62 ms of delay one-way, 30 ms of dead time at the bottom, and 62 ms of delay the other way. 154 ms out of 260 ms that nothing is happening even though you're "actively" working the clutch.

Well, 260 ms - 154 ms is 106 ms left over to actually get the engagement and disengagement done. 106 ms sounds like a lot, but no. My high-speed data shows that before the CDV and CMC damper delete, it takes a total of 99 ms for the clutch to fully engage if you omit the 62 ms delay (161 ms total.) But that is the one way travel time for the clutch fork. Let's assume that you can clutch in as fast as my data shows the clutch moving after the delete, which is 62 ms instead of 99 ms.

So from the time you start pressing the clutch pedal, you have:

• 62 ms of delay
• 62 ms of disengagement
• 30 ms of dead time at the bottom of travel
• 62 ms of delay once your foot starts going the other direction
• 99 ms of engagement (slower than disengagement because of the CDV)

Total = 315ms.

So if you *try* to shift any faster than 315 ms with the factory setup, you will get some amount of unintended clutch slip. There's basically no way around it.

And unintentional slip is a big complaint on these cars with fast shifting (any car with a CDV, really) for this exact reason. With the slow engagement rate, you just can't effectively work the clutch as fast as you can work everything else. The clutch becomes the weak link and you can't help but unintentionally slip the clutch.

Now let's run those numbers again after the CDV and CMC damper delete.

• 32 ms of delay
• 62 ms of disengagement
• 30 ms of dead time at the bottom of travel
• 32 ms of delay once your foot starts going the other direction
• 62 ms of engagement.

Total = 218 ms.

Needless to say, 315 ms down to 218 ms is a big difference, and it puts the speed of clutch engagement within my envelope as a driver, whereas before I struggled a bit to get good, consistent rapid shifts.

This is to say nothing of the element of timing. Timing is really important for a good shift no matter how fast you're shifting. You want the clutch to engage at the exact right instant, and the more delay there is in the system the less intuitive this becomes. Humans are actually very good at timing things within tens of milliseconds - look no further than musicians for this sort of precision. Again, reaction time has nothing to do with your ability to sync to small time frames or perceive them.

And, having actually experienced this car before and after, that difference in timing is stark. It's more intuitive and the clutch engages harder and more predictably. It made driving the car in everyday situations a lot more forgiving and enjoyable, and when I get more aggressive with my shifts it made a huge difference.

I hope for people reading this, that this is a lot more enlightening than something so dismissive as:



People have been deleting Honda CDVs for a long time now. The design hasn't changed much in two decades. The CMC damper is just another thing nerfing the system for the lowest common denominator.

A sub 300ms shift on a street car with synchros is just ungodly fast.

If you can tap under 300ms in reality AND do it more than once without blowing synchros, then you sir deserve a medal, an fia license, and a place on a very fast team. Especially considering the clunky tremec 3160 is known to typically take 400ms+ for fast and aggressive shifts.

Even dct trans from paddle pull to gear change typically runs 100-300ms. Dog boxes take 100-200ms and half the time they just slam into gear or rip it out before the clutch action.

A 260ms shift on a tremec street trans puts you in the territory of the best drivers on the planet. The vast majority of drivers out there are lucky to get their best shift times into double that.

Sponsored