Archives

Anybody know exactly what is going on when the car judders?

I know that it involves the start clutch, but it strikes me as an odd clutch problem. Usually when clutches fail they slip, and that isn't happening here since there is no problem once the car is moving. Also there is no "slow down" judder, so the problem is when the clutch is locking up, but no issue when it is releasing. It definitely feels like the clutch is grabbing and releasing rapidly, but why? Is it that the control circuit lacks feedback and just applies a fixed ramp of pressure, which is inappropriate for some levels of torque, or is it the control circuit itself that oscillates, applying too much pressure, then too little? Or to put it another way, if one had a pedal to control that clutch, could the car be started smoothly by applying the appropriate amount of pressure for the corresponding amount of torque (gas pedal position), or is the juddering a symptom of an inherently unstable clutch?

In

http://alexandria.tue.nl/repository/books/613056.pdf

starting on page 28 there is a copy of the paper "The influence of the interface coefficient of friction upon the propensity to judder in automotive clutches". Very technical. It is not exactly the same system, but if the judder studied and that in Honda's transmissions are similar it suggests that the failure of the startup clutch involves changes to the curve of friction coefficient vs. slip speed that result in mechanical oscillation:


The choice of the friction material can be quite significant in order to damp out the effect of clutch judder in as short a window of oscillations as possible. It shows that friction materials with a positive gradient of coefficient of friction with slip speed provide a better damping effect and little or no self-excited vibrations occur and that friction material characteristics with negative slopes increase the propensity to judder. The test procedure reported here can therefore be employed as a 'sign-off" quality test for the choice of friction materials.

Still, one would assume that whatever the shape of that curve a properly designed control system could keep the clutch out of oscillation, at the cost of a slower startup. ( Whether the slope of that curve is positive or negative, the coefficient of friction itself is always positive, one can always start a vehicle smoothly, albeit slowly, by using the minimum amount of clutch pressure.) Or in the more complicated sense, in the way that aircraft are designed that are intrinsically unstable but may be flown anyway because the computer control systems can dynamically adjust for the instabilities. If the control system isn't taking this change in slope into account it is easy to see where it could get into trouble once the clutch ages enough for the slope to change. If it sees the slip decreasing and expects positive slope it would assume that the friction was decreasing and increase pressure to increase friction. However, with negative slope as slip decreases the friction is actually increasing, and it might actually have to decrease pressure to maintain constant friction. If it doesn't do the right thing that could easily lead to control oscillations (but a lot faster than a beginning driver on a manual letting the clutch out too fast, then jamming it back in, etc.) And that's just assuming the slope is everywhere the same sign, and not something much more complex.

Is there a start clutch calibration procedure? Seems like if the properties change it would have to be done periodically for optimal starts.

That is just guesswork, I don't actually know what is going on in Honda's CVT. Still, I think that the control system probably is mostly at fault in not being able to smoothly operate a start clutch which has a different friction vs. slip speed curve than that of a brand new transmission. All the expensive fluid changes and burnishings are Honda's way of working around this issue. That is, instead of fixing their control system to handle the clutch/fluid aging gracefully. Or to put it another way, it isn't that the start clutch stops working as a clutch, it may be just that it stops working the way the rest of the transmission expects it to.

Warped clutch disc. caused by heat from friction!H

Well, all I can tell you is that when the judder is observed the list of suspect causes is always the same and it is up to the technician to identify which is the most likely by a process of elimination. The more experienced (and well trained) the technician the more likely he'll be acknowledging patterns that will shorten the diagnosis process.

In your case all I can do is provide the list of these suspect causes:

-Start clutch defective (or just about there)
-CVT fluid quality deteriorated
-Defect in CVT valve body assembly
-Start clutch pressure control valve is defective
-Defect in the speed sensors at the Input OR output shaft(drive pulley)
-CVT speed sensors or secondary sensors defective
-PCM with hardware or software abnormality

If a breach of operational thresholds occurs a DTC is always generated, this usually points to a loose solenoid or CVT control harness connection issue. Also check and clean the ATF cooling lines while you're at it - bundle a CVT filter change in the servicing as well. ;)


Cheers;

MSantos

re: Is there a start clutch calibration procedure?

I'm sending you a PM that contains some info regarding this.

Well, all I can tell you is that when the judder is observed the list of suspect causes is always the same and it is up to the technician to identify which is the most likely by a process of elimination.

As far as I can tell the roll of the Honda service department with regard to CVT problems is to do everything humanly possible to avoid actually repairing the car. This includes saying that they will repair it under warranty and have ordered a part, then saying they can't find the part, then saying they found it but won't do the repair because they shouldn't have said they would in the first place, then saying they would contact Honda N. America to see if they will cover part of the repairs, then admitting that nobody ever called them, then refusing to call me back. Need I go on? And this is the better of the two Honda dealer's I have dealt with so far. The first service department tried to sell me all sorts of repairs the car didn't need, and did sell me one by outright lying ("you need to replace all the plugs because the service interval is 90K miles", actually it is 105K miles), messed up the steering a bit on one of the repairs they did do (which at least they fixed later), and basically should have had a big "empty your wallet here" sign over the counter.

I really need to find a good independent who works on these cars, because the dealer's are 0 for 2 for people I want to deal with.

As far as that list of defective components, my point was that I'm beginning to think that the start clutches that judder no matter what the service technicians do are not defective as clutches per se, they just don't work with the control software, which expect a clutch to behave in a certain way, and that is not how they behave all the time. This is kind of like how in a manual as the clutch wears the point at which it engages in the pedal travel moves. So the driver compensates. But the CVT control software doesn't seem to be able to compensate for some normal states of the clutch. I mean, if the clutch was truly defective, a burnishing wouldn't do anything, yet it often (temporarily) eliminates the judder.

re: Is there a start clutch calibration procedure?

I'm sending you a PM that contains some info regarding this.

Got them, thanks. Will try the start clutch calibration procedure. Turning on the headlights and decelerating from 40mph to 20mph without touching the pedals seems like something that a lot of people would do on a daily basis. However, I almost never drive at night, so I'm not one of them.

Turning on the headlights and decelerating from 40mph to 20mph without touching the pedals seems like something that a lot of people would do on a daily basis. However, I almost never drive at night, so I'm not one of them.

Yeah, the procedure doesn't sound like much, does it? I'll have to pay more close attention to the actual speeds, but I think I do that every night when I drive home from work. If so, maybe I'm constantly doing a calibration for the start clutch? I dunno...it still judders sometimes, even though the start clutch has been replaced once already.

Yeah, the procedure doesn't sound like much, does it?

Especially since during the procedure that clutch is presumably fully locked up. One would have thought that to actually see what the clutch was doing the car would have to have been driven in some special way from 0 to 20 mph, that is, clutch fully slipping to clutch fully locked.

Speaking of judder, my car usually judders when starting uphill. A couple of weekends ago I had to start uphill on a truly terrifying grade (at a stop sign, having to turn left, opening out onto Sunset avenue in Hollywood.) Lots of traffic. I was pretty sure we were going to die. Anyway, set the parking brake and when a hole appeared in the traffic stepped on the gas, released the brake, and - no judder. I was expecting to have my teeth shaken out but it was a completely smooth start. If I can find some place safe to repeat that I will. In the meantime I'm wondering why a little slope would cause judder, but a huge slope would eliminate it. Again, sounds like a clutch control problem, not a clutch problem per se.

There are certain places in my regular drives where I know the car will judder. These are not large inclines, but for some reason the car judders at these locations. Lately I've been shifting to S when I need to start from a stop at these places, and it seems to help. Once I've gotten up to speed, I shift back into D.

Got them, thanks. Will try the start clutch calibration procedure.

Did it yesterday.

1. warmed car up (drove it about half an hour).
2. pulled over and turned car off
3. turned car on
4. turned on headlights
5. accelerated to 40mph (two people in car, so slight judder even though the street was flat)
6. decelerated to 20 mph without my foot on either pedal.
7. pulled over and turned car off.
8. turned off headlights.

I can't detect any difference in the way it drives - still judders.

Hi Pasadena_commut:

Excellent link!

Have you changed the tranny fluid yet? I know it helps and since your car did not judder when new, I would hope you are attempting all avenues starting with the least expensive.

Wayne

Have you changed the tranny fluid yet?
Repeatedly. Actually, looking back through the records, the CVT fluid changes are approach the frequency of oil changes. Also it was burnished which "fixed it" for a couple of thousand miles, then the judder came back.