based on your description it looks like that when the wheels and driveshaft rotate within a certain speed range something in the car is brought into a vibration - a part that has a natural frequency that is more or less equal to the frequency of the wheels turning. normally that is not going to happen, since all the frequency generating parts (driveshaft, differential, engine etc) are mounted through rubber dampening fixtures that should prevent it to happen. the car body itself also has stiffening beams in its structure that should help prevent it. maybe a thorough check of all the mounting components and the bodywork itself (welds that have come loose) may reveal where something is wrong.

i do not think the drivetrain itself is at fault - if that was the case the change of various elements in it would have solved the problem. one thing i would particularly pay attention to is the mount of the differential/angle drive. i would check all the rubber mount components and also check whether the tightening torque is in accordance with what is should be.

Thank you very much for this input. This car has a multiple link independent rear suspension. That means the differential is fixed with bolts to a subframe in the rear. That subframe is mounted to the car with huge rubber beds but also connected with multiple (10) links to the wheel assembly. So in theory, the back end of the system is pretty sturdy if you add that the multi link suspention hold the wheels steady (no loose parts and aligned) and the differential should be kept in check by the axles, subframe, coils and (new) shock absorbers. Also, the shaking is more pronounced in the front of the car.

On the other hand, the transmission is damped only with a fist sized rubber element connected to the frame, much closer to the middle of this system. While that element is old, seems to be in a good shape. Motor mounts... Well, this is an old car...

May be I should add, that the shaking doesn't feel like something is loose. It is more like when something violently shaking the car.

If I understand correctly, you would go for the transmission mount, the motor mounts and the rear subframe rubber... Probably in this order...

The biggest question in my head, before I start throwing money at an (in other aspects great) car, if this is a drivetrain mount problem, or a suspension problem. As the suspention is checked so many times and found sound + the vibration is periodic and it is present in a wider speed range than a tire vibration (between 110-180 km/h) + the opinion of "romke", I would say, this will be a drivetrain mount problem.

Any second opinions? :) thank you!

The coming and going sounds like 2 frequencies modulating each other. If it's a tire or wheel problem, the vibration gets worse when the frequency of each tire/wheel is in phase. Have you swapped out the wheels or just rotated them? I'm thinking a rim or two might be bent.

I have not swapped out wheels (rims). I have mounted multiple sets of tires and the wheels were rotated from rear to front or spare randomly. It also means that the wheels had been checked visually by multiple mechanics. The rims are made of steel. I understand your point, but how would you explain periodicity. I assume wheels turn pretty much at the same speed in a straight line.... Also, there is no side to side shimmy in the steering wheel, but it shakes with the car.

The car is also buttery smooth up to 110 km/h, 70 miles/h (tire problems should come out slower speeds than that I guess)

Oh, one more thing: this car has ASD, which is sort of a limited slip differential system. ASD is supposed to regulate slip in the differential assembly with oil pressure. This system is not functional in this particular vehicle (Fault code). There is no noise coming from the differencial, what so ever. I don't see how could this add to periodical vibrations, but who knows? All wheels are turning at the same speed, as the ABS works properly.

Just throwing this out but we had a State Patrol unit that had a bent wheel and it did exactly the same thing, you went fast eough and it shook, slow down and it went away. Even a small amout of off center weight on an item spinning at that many RPM will start to shake a lot. I think BrianE22 hit the nail on the head.

For sure that is wheel related vibration. 60-70mph is usually when 1st order vibration from wheels or tires starts to excite the base.

Agreed, wheels are suspect.
Visual inspection is not sufficient.
Put a dial indicator on each axle and measure runout of each wheel rim, inside and outside, radial and lateral.

While you are preparing for that, you need to remove the wheels and scrub the accumulated mud/tar off of the rims, something which I think is still part of an 'E' service, and probably not often done.

Mike Halloran
Pembroke Pines, FL, USA

Oh boy. Forst of all, thank you all very much for your inputs. Consensus, it seams, is about a wheel related problem. So here comes my doubts:

Wheel runout problems come out about 40 mph. The car is very smooth up to 70.

Balancing have been done at least 10 times in the past 4 years in 10 different shops while either rotating or replacing tires. The 5th wheel was always involved. The problem never changed.

I never heard a periodic vibration issue caused by wheel balance, runout! How and why it smooths out if it vibrates in the next second?

I never heard of a balancing problem present in such a wide speed range (from 75 to 110 mph) untol the vibration will get so fast that it can't shake the car anymore.

I will check the radial and lateral runout with the wheels mounted. That has never been done before.

A bent rim can be perfectly balanced, statically and dynamically. Force variation comes from the associated runout.

Wheel hop natural frequency is usually between 9-11 hertz which for most cars would be between 60-70 mph. I believe the next natural frequency is the oscillation about the steer axis but I don't remember what frequency that typically is. It's possible your 3rd vibration speed range is caused by the next mode of vibration for the wheel/tire/suspension.

The coming and going of the oscillation is caused by there being 2 frequencies close together. You're getting a modulation producing a "beat" frequency. Suppose, for instance, that one wheel hop natural frequency is at 10.5 hertz and another is at 10.7 hertz. The modulation would produce a frequency of 10.7 - 10.5 hertz or .2 hertz. The period for that would be 5 seconds. You'd feel the vibration coming and going every 5 seconds.

It does seem strange, though, that no one would have thought to check for mounted wheel runout. Maybe your shock absorbers are in real bad shape?

Really good clue! Right now I am measuring vibrations on all the 4 corners of the car. I will post results soon. Shocks are new though and so are shock mounts.

Hi hhy2k,

by "tires were changed" to you mean new tires, or existing tires get swapped around?
How long has this been going on? It did it suddenly develop when you went on a trip and your teenaged kids were left home with the car?

"wavelength of about 3-4 seconds" . I take that to mean 3.5 seconds between the peaks of highest vibration you feel.
3.5 seconds/cycle is about 17 cycles per minute. If my 20 year old Volvo has tires that are about 25 inches tall and spin at about 840 rpm.
17 cpm is about 2% of 840 rpm. A 2% change in tire diameter would be about 0.5 inches. Not inconcievable for worn mismatched tires.


Any "couple", two plane or dynamic unbalance, or even significant single plane unbalance would be inphase on both tires every 3.5 seconds.
One result could be to try to wag the tires back thru the steering linkage. Whether the force is so large it significanlty deflects suspension parts, or the suspension has looseness that allows small amounts of unbalance to achieve large vibration amplituies is hard to say.
yes, less than perfectly round whee;l/tire assemblies will cause 1X rotation (or other frequencies, depending on the shape of the deformationforce no matter how well they are balanced.
Hunter claims theire force balancing compensates for stiffness variation in the tire, which would be leigitimately invisible to a conventional balancer.

When measuring vibration will you be getting frequency info? Frequency information is key to accurate troubleshooting.

• http://files.engineering.com/getfile.aspx?folder=ef8d3b44-f59d-4a9f-a07f-bf

I had a long conversation with a tire shop. They said again and again, periodic vibrations can't be the result of wheel balancing. The vibrations are either present or not (at a constant speed). More than that, they also said that a badly balanced wheels has symptomes at a constant speed only. A slightly higher or lower speed would make the vibration disappear! By the way, that is my experience with tire imbalance as well. In this case though, the vibration is always there over 75 mph, only the frequency will change and the periodicity will be faster.

I get questions about the car and part's age. This is an old car, but it os a good car with amazing engeneering. The problem we discuss is there since I bought the car 4 years and 30 000 miles ago. In order to track down this problem I have changed tires at least 5 times (different tires!), changed shock absorbers, shock mounts, tie rod ends. The car had been checked by professional shops for every suspention element and even minor problems were addressed. I have been changed two flex disks in the drive line and the one suspention bearing in the middle.

I did not change wheels, but as I have 5 of them they are going to various places on the car, or in the trunk. I never had a change in the vibration behavior by doing anything with the tires. The wheels are made of steel, not aluminum.
I haven't change motor mounts.

Now to the measurements. You can see 5 mesurments below, each line with 4 pictures represents one measurement, the lenght of every measurement is 30 seconds, the car travels 140 km/h (!). The app I used runs on an iPhone. The phone lies flat (like on a table) over the four corner of the car, over the wheels. The fifth measurement is in the middle of the car! The order of the measurements is as follows:

1st line: middle of the car.
2nd line: right front
3rd line: right rear
4th line: left rear
5th line: left front

RED represents x axis, side to side motion
GREEN represents y axis, front to back motion
BLUE represents z axis, up and down motion

The last drawing is the motion pattern I read out of the diagrams. Your conclusion might be different of course.

• http://files.engineering.com/getfile.aspx?folder=31ecfa92-2121-4974-aadb-17

First pic... Sorry, i can add only one at the time...

• http://files.engineering.com/getfile.aspx?folder=3c044271-12f4-4480-9eab-e9

One more thing! Someone asked about the real time of the periods. As the pictures I already posted are smaller scale summaries, here I post a full size one where the periods are very clear. This is the side to side motion of the left front wheel for an exact 30 second period at 140 km/h (87 mph).

Thanks for every inch of though put into this problem. I really appreciate it... But it's fun, right? :)

• http://files.engineering.com/getfile.aspx?folder=c731c21f-3e89-4f01-ad06-03

To bad there is no post editing. The last pic is for the RIGHT FRONT wheel.

I'll wait til you publish your vibration data.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376: Eng-Tips.com Forum Policies http://eng-tips.com/market.cfm?

Hello again. I already published the diagrams. They are in the attached pictures. My conclusions after some math, measurements, counting and tire size consideration are as follows.

At 140 km/h (39 m/s) the vibration base band is 6 seconds! This is the periodicity. The vibration frequency is about 90/6s (15/s). Tire rotation is 120/6s (20/s). Drive shaft rotation is about 360/6s (60/s). //engine speed is 3600 rpm (60x60)// >>> note: this is a manual transmission! Symptoms are the same in any gear (3,4,5,N), depressed clutch has no effect either.

The tire rotation speed (120/6s) does not match the vibration frequency (90/6s)

The gear ratio in the differential is about 3:1

no, pictures of waveforms aren't data.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376: Eng-Tips.com Forum Policies http://eng-tips.com/market.cfm?

Using TMOOSE's analogy (thanks for that post):

10cpm (6 sec wavelenght) and by a tire rotation speed of 1200 rpm the wheel diameter mismatch is 0.833 percent. At this tire size this makes about 5mm, 3/16th of an inch difference in diameter. This is very possible, although I do not have mismatched tires on one axle. They are different in the front and the rear, but they are matching pairs (the car is RWD)

Also, how would this explain the 90/6sec periodicity if the tires turn 120/6 seconds?

@GregLocock

I worked from these pictures and found all the data I posted. I am happy to provide more data if you need and if I know (but I really dont know how old is the captain, so don't ask that question please :) but seriously, what is the data you are missing?

numbers, in the form of columns of time and the instantaneous vibration (ie the numbers plotted on that graph). I think you have done a great job at analysing it thus far, I'd just like to look at the raw data.

I've worked on 2 heterodyne or beating problems in my 20 years of automotive NVH, they are tricky.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376: Eng-Tips.com Forum Policies http://eng-tips.com/market.cfm?

Well, columns are good, so here it comes:

Vehicle speed: 140 km/h, 39 m/s, 87 mph
Wheel speed: 20 rps, 1200 rpm (120/6s)
Vibration base wave lenght: T=6 seconds
Vibration frequency: 15 Hz (90/6 seconds)
Drive shaft speed: 3600 rpm (60/s and 360/6s)
Engine speed (no effect)
Transmossion speed (no effect)
Vibration is present over 110 km/h

I can't match the wheel's 20 rpm to the 15 Hz vibration frequency. May be nobody can! :)

No, the actual datapoints.

s0.0012
etc etc

typically a csz or tst file

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376: Eng-Tips.com Forum Policies http://eng-tips.com/market.cfm?

Oh OK, that column! I'm sorry for being such a bonehead. I have this longer run recorded.

• http://files.engineering.com/getfile.aspx?folder=95f170ec-b7dc-457f-9251-f6

The tire shop " said again and again, periodic vibrations can't be the result of wheel balancing. The vibrations are either present or not (at a constant speed).
Would you say, daring to risk an age discrimination lawsuit, based on superficial appearances Is it likely that the tire person could have accumulated significant tire service experience back in the 70s, before the manufacturing tolerances of some mundane components gained a decimal point or two? When even Honda was still using lug centered wheel assemblies? (I'm guessing your Benz's wheels are hub centric.)

The tire shop " said again and again, periodic vibrations can't be the result of wheel balancing. The vibrations are either present or not (at a constant speed).

Did they say that 10 times?

How about this...
An Internet clown says "All 4 tires on a car are not rotating at the same rpm when traveling in a perfectly straight line when traveling at constant speed because even if brand new they (he tires) are not the exact same diameter."
An Internet clown says "All 4 tires on a car are not rotating at the same rpm."
An Internet clown says "All 4 tires on a car are not rotating at the same rpm."
An Internet clown says "All 4 tires on a car are not rotating at the same rpm."
An Internet clown says "All 4 tires on a car are not rotating at the same rpm."
An Internet clown says "All 4 tires on a car are not rotating at the same rpm."
An Internet clown says "All 4 tires on a car are not rotating at the same rpm."
An Internet clown says "All 4 tires on a car are not rotating at the same rpm."
An Internet clown says "All 4 tires on a car are not rotating at the same rpm."
An Internet clown says "All 4 tires on a car are not rotating at the same rpm."
An Internet clown says "All 4 tires on a car are not rotating at the same rpm."

An Internet clown says "All 4 tires on a car are NOT rotating at a constant rpm."

I'd position myself on the highway enjoying the "periodic vibration" at speed in either the left or right lane.
Then, when traffic permits, I'd turn the wheel about 1 inch at the rim to cross all 3 lanes in one gentle swoop carefully noting if during the swoop the period of the vibration changes. When traffic permits I'd return to the originsl lane, again noting the vibration. If a section of the highway has a gentle curve a similar test is conducted automaticallly.
Sometimes I can "catch" the vibration at constant point, because by luck I have set the problem tires at very nearly the same rpm, so the unbalance and/or runout are also maintaining the same phase, whether the relative phase is maximizing (high vibration) minimizing (low vibration) or some intermediate effect.

Then I'd measure (not eyeball, even a tire specialist's eyeball) and record the radial and axial runout of all 4 tires and rims while mounted on the car.
https://groups.google.com/forum/?hl=en#!search/djt...

I think I got the message. Although I don't think he said they all turn at the same speed (if that would be the case, your ABS and all those stability control gadgets were pretty much wasted money). He only said, at a constant speed (a 10-15 mph range) a vibration is either there or not and as this is true for each and every wheel, he doesn't see how a periodic vibration could happen in such a wide range of speed (and I add that the characteristics of the vibration never change). The turning test I have done many times. I never noticed any change in the vibration amplitude or frequency because of a slight turn. The only thing had any effect ever are the very smooth surfaces under the car.

Honestly, I should really do the runout test on the car as you suggest, it is just so hard to grasp the bent wheel concept after all that tire and wheel swapping for years... While I always had the same symptoms. On the other hand if the geometry discrepancy is on the car itself, that would have never came up by switching around the rims, no matter how well the tires were balanced or checked for runout on the balancing machine.

This is clearly the next step I must do. Thank you very much.

Have a friend drive your car while you ride next to it in another car. Look at each of the 4 wheels to see if one is shaking more than the others.

Or get a GoPro or other high def/hi frame rate camera and make video recordings of the suspect areas to try to narrow down the source. Would be easy to use a GoPro underneath the vehicle to observe drive train and suspension areas.

Jay Leno's Garage and other automotive videos makes extensive use of Go Pros. Have noticed quite a bit of detail captured on vibrations or shake that builds up in various areas of a vehicle, for example frame/bumper to body differential movement at road speeds, especially on older vehicles.

All run out had been checked out to be less than a millimeter with the wheels mounted on the car.

I move on to the drive train.

All well and good that the runout was measured.

On a smooth road, have an assistant in another car drive beside you while you drive at the problem speed, and have the assistant tell you which wheel(s) is plopping up and down.

OK, I got that data, I'll have a play around with it. It looks as though you used a 100 Hz sampling frequency. What did you set the anti aliasing at?

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376: Eng-Tips.com Forum Policies http://eng-tips.com/market.cfm?

Here's a waterfall plot of the data, and a synthesised signal of similar characteristics

here's the function to generate that synthesized signal

for i=2:length(t)

if t(i)<20
f=20.5;
else
f=20.5-(t(i)-20)/60;
end
G(i)=sin(t(i)*2*pi*f)+.8*sin(t(i)*2*pi*(f+1/4.5));
end

On the pretty plot frequency is vertical, 22 hz or so at the top, and time runs from right to left (sorry)

Now, the interesting thing is that the waterfall of the synthesised signal does not display as I'd have expected, as the speed ramps down the beating reappears and disappears, quite strangely. The resemblance to your data is reasonable. Sadly i don't think this really gets us much further.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376: Eng-Tips.com Forum Policies http://eng-tips.com/market.cfm?

Very interesting thread gents.

The one thing that jumps out to me is that the wheels are the same ones.

I suggest borrowing a set of good wheels/tyres that don't give these vibrations and go for a run.....