Testarossa dreaded differential issues and questions | Page 8 | FerrariChat

Testarossa dreaded differential issues and questions

Discussion in 'Boxers/TR/M' started by PineChris, May 28, 2016.

This site may earn a commission from merchant affiliate links, including eBay, Amazon, Skimlinks, and others.

  1. godabitibi

    godabitibi F1 Veteran
    Silver Subscribed

    Jan 11, 2012
    6,252
    Papineauville, Quebec
    Full Name:
    Claude Laforest
    Scott you forgot to also mention all those who HAD a bad break with the oil coming out of the diff with a broken case and cover WITHOUT any symptoms before. On a SLOW drive.

    But we are just a bunch of negative guys. There is no proof of anything about those diff. Please stop talking about it, who cares. ;)
     
  2. qwazipsycho

    qwazipsycho Formula 3

    Oct 30, 2004
    1,176
    Utah
    Full Name:
    Scott
    You mean like me? OH crap! I didn't say that. I didn't say anything. I wasn't here.
     
  3. carguy

    carguy F1 Rookie

    Oct 30, 2002
    3,401
    Alabama (was Mich.)
    Full Name:
    Jeff
    There sure is a lot of brain power on fchat. Sam...thank you for your in depth study. It is enlightening to see the infamous diff issue in statistical terms. Folks need to keep in mind that from Sam's analysis you estimate the "probability" of failure, but there will always be special causes of failure, etc. It would be great to see a fishbone diagram of this issue so we could see all of the contributors involved. I agree with Turbo-Joe that over time the stresses are relieved. If the diff makes it through a certain number of cycles, the data shows that the failure rate drops off...that makes sense to me. There will always be "fliers" and "special causes" for failures that do not fit into this analysis. It's important that the mud-slingers out there understand that. Quote...."you can please some of the people all of the time, and all of the people some of the time, but you can't please all of the people all of the time".
     
  4. ozziindaus

    ozziindaus F1 Veteran
    Silver Subscribed

    Aug 16, 2012
    9,799
    Detroit
    Full Name:
    Sam
    #179 ozziindaus, Jun 15, 2016
    Last edited by a moderator: Sep 7, 2017
    Here you go Jeff. I had to break it up for readability.

    I have limited the fishbone to "Process" but I'll put another one together for "Application" that will include possible causes such as;

    -Driving style
    -Wheel vibration
    -Environmental conditions
    -Oil levels etc

    Also, if anyone knows the material, x-sectional thickness at the weld and diameter, I can calculated the shear stresses for fracture and possibly the fatigue limit based on drivetrain output and no. of oscillations.

    I will say that I think this is more of a fatigue related fracture rather than ultimate shear strength based on the Failure Mode (cracks appear before total collapse).
    Image Unavailable, Please Login
    Image Unavailable, Please Login
    Image Unavailable, Please Login
     
  5. Philwozza

    Philwozza Formula 3

    Apr 23, 2009
    2,045
    Winchester UK
    Full Name:
    Phil Worrall
    There are other important factors in the replacement of a diff than bearing preload or tooth backlash. The thickness of the clutch stack that sits within the casing is vitally important. If the old shim stack is worn and placed within a new casing it is unlikely, but not impossible, that the clutch pre-load is incorrect. We have learned this during the manufacture of HUNDREDS of competition limited slip diffs.

    Phil
     
  6. Philwozza

    Philwozza Formula 3

    Apr 23, 2009
    2,045
    Winchester UK
    Full Name:
    Phil Worrall
    Can I just repeat an old thread.....
    In my experience over many years, the 512 diffs fail at the point when the driver LIFTS OFF POWER. It is the point when due to the design of the tooth helix, the parts are bring forced away from each other. I have no analysed data that shows failure due to bad (donut) type driving. My feeling is a poorly designed part that can fail at any time from new to old age.

    IMO

    Phil
     
  7. ozziindaus

    ozziindaus F1 Veteran
    Silver Subscribed

    Aug 16, 2012
    9,799
    Detroit
    Full Name:
    Sam
    Phil, so failure occurs when the weld goes into tension rather than shear (torque)?
     
  8. bpu699

    bpu699 F1 World Champ
    Owner Silver Subscribed

    Dec 9, 2003
    16,206
    wisconsin/chicago
    Full Name:
    bo
    Can you outline the fastest way to inspect the diff? My motor is still out, major done, just working on cosmetics now...

    Sounds like the clutch has to come out? More? Does this then require more oil seals to be changed? Require more parts to be bought?

    I know you can do it in an hour and a half... But for a newbie, these things seem to take much much longer...

    Bo
     
  9. godabitibi

    godabitibi F1 Veteran
    Silver Subscribed

    Jan 11, 2012
    6,252
    Papineauville, Quebec
    Full Name:
    Claude Laforest
    If your engine is out it's time to check it. Yes clutch need to go out. Then the rear cover and the side cover. I think you can check with side cover only removed after the lay shaft is out from the rear. That would save to remove the large rear cover and flywheel.
     
  10. carguy

    carguy F1 Rookie

    Oct 30, 2002
    3,401
    Alabama (was Mich.)
    Full Name:
    Jeff
    Regarding the point made by Phil about "lift off throttle" or hard deceleration while in gear, during these events the pinion gear is pushed away from the ring gear. This results in a lot of stress on the differential/transmission casing to contain everything. I do not think the differential weld gets a tension load though, it does of course get a torsional load in the opposite direction of normal driving. Perhaps over time and forward motion the weld gets accustomed to torsional force in one direction, but looses strength in the other direction? Is this possible?

    Sam I applaud you for putting together that fishbone diagram. I think we have some smart and experienced folks here who can add some value to the various inputs.
     
  11. qwazipsycho

    qwazipsycho Formula 3

    Oct 30, 2004
    1,176
    Utah
    Full Name:
    Scott
    #186 qwazipsycho, Jun 16, 2016
    Last edited by a moderator: Sep 7, 2017
    Yeah, if engine is out already this is a no-brainer. At least check it. If everything checks out, it's only going to cost you time, a bit of gasket sealer, and new gear lube.

    Assuming all you're into is a major, you've got the engine still in the sub-frame and everything intact, and exhaust disconnected at the headers. You have easy access to everything. I don't know how you have things supported now but you're going to disconnect the back two engine mounts from the sub-frame. You'll need to plan on how you'll support the back of the engine while not blocking access to the sides of the gearbox. Here's a couple of pics of how I did it when my engine was out. I had it all sitting on a platform that made it easy to just put a small scissor jack under the gearbox so I could lift it up. You'll need to do this so you can clear the subframe when removing the diff.

    As far as the process for inspecting the diff from where you are, I'll try to list the steps from memory:

    Drain gearbox.
    Remove starter bolts and all nuts around clutch (bell) housing.
    Remove drop gears/clutch housing just like a transmission - "jiggle n pull"
    Be careful as the quill shift is near the bottom and will be the last thing to clear as you remove.
    Also remember not to tilt forward or you'll pour gearlube out. Lay it on it's backside.
    Remove clutch assembly. (Keep parts in order, it's balanced that way. Note paint marks as guide.)
    Remove flywheel. (Note paint mark to balance with clutch assembly)
    Remove Two large vertical bolts through engine mounts. Here's where you need to support the back of the engine. You will want to loosen front engine mounts as well to allow slack.
    Slightly lift the back of the engine until mounts don't touch.
    Remove the left side engine mount from the back cover - 4 nuts.
    Remove all the nuts and bolts holding on the back cover. There are 14 nuts and 2 vertical bolts screwed into the block.
    Since the engine is out, you'll be able to lift the back of the engine high enough to get everything to clear the subframe. (This isn't possible while in the car and requires stud removal.)
    Remove the two bolts holding on the splash shield and remove the splash shield and standoffs.
    At this point, you will be able to see the diff. Don't be fooled. You can't tell if there's damage yet.
    Disconnect the left and right drive (CV) shafts at the diff flanges. (This sucks!) Using a hammer and some type of "chisel" bend all the bolt locks back on the bolts holding the CV shaft to the diff flange. To break the CV flange bolts loose, put the wheel lug bolts back in without the wheels and use a long pry bar between them for leverage. Cover the ends of the CV joint with a baggie and rubber bands to keep from smearing grease everywhere.
    Remove the left rear exhaust header.
    Remove diff flange bolts in the center of the flanges and remove the flanges. These can be tough to break loose if they're original.
    Remove all nuts and 1 bolt holding on left and right side covers. The bolt is vertical into the block, inside left side cover. There is also an O ring on this bolt. It sucks. Silicone it upon reinstallation.
    At this point all you do is slide the side covers outward. There will be at least 1 shim under the right side cover. Use a rubber mallet and beat on the left side coverfrom the inside. It's awkward but it works.
    Now the diff is laying in the gearbox. Lift the back of the engine until the diff will clear the subframe. Remove and inspect.

    Also check bearings and races
    seals and O rings on the side covers
    gear wear on crown and pinion
    quill shaft condition
    clutch and flywheel wear
    pilot bearing
    TO bearing and seals
    rear main crank seal
    input shaft seal - VERY IMPORTANT

    If your diff looks good, you may want to have it x-ray'd or magnafluxed or something just to be sure.

    Replace what needs replaced and if your diff looks good reassemble.

    Geez that was longer than I thought.
    Image Unavailable, Please Login
    Image Unavailable, Please Login
    Image Unavailable, Please Login
     
  12. 2dinos

    2dinos F1 Rookie

    Jan 13, 2007
    2,772
    Would you rather keep the carrier that's been debated on when it will fail and how much damage it will do?


    After seeing all the carnage posted - I couldn't get that thing out of there fast enough! Good riddance. Maybe if the readers & posters would have experienced this somewhere. My Fiat X1/9 carrier failed while backing out of a parking space. Going 2 MPH, zero warning, and the whole trans-axle was garbage. I think it cost me ~$150 and a day of work to get a new gearbox from a wrecker, overhaul it, and off the jack stands.

    With the Ferrari!
    50 phone calls all over the planet / endless research / financial justification and planning / months of worry dealing with a down car.
    Just the thought of it. No thanks.

    j.m.h.o.
     
  13. Melvok

    Melvok F1 World Champ
    Owner Silver Subscribed

    Jul 25, 2008
    14,101
    Amersfoort, The Netherlands, Europe.
    Full Name:
    Mel
    Absolute GOOD Q !

    F factory did repair all gb's and included oem diffs long time before the 3 other upgrades arrived...
     
  14. Melvok

    Melvok F1 World Champ
    Owner Silver Subscribed

    Jul 25, 2008
    14,101
    Amersfoort, The Netherlands, Europe.
    Full Name:
    Mel
    Thanx for all conclusions up to now Sam ..

    I myself have found at least 12 gearboxes stacked away in crates of carbreakers yards.

    And have seen many, many photos of cars taken apart in F.I. Eurospares. (have a look in their site)

    Hundreds in all I think ...

    Many, many owners did not want to reveal that their gb/diff/C&P was broken ... they thought that other people would call them bad drivers ...

    I heared so many people talking about that and this was confirmed by several specialized gb shop owners also ...

    There were far more gb problems than we dare to believe ... it was hidden in secracy ... because ....

    So: go for sure and put in an upgraded diff is my humble advise ! Or make a cash reserve of K$ 20 ...
     
  15. turbo-joe

    turbo-joe F1 Veteran

    Apr 6, 2008
    8,866
    southwest germany and thailand
    Full Name:
    romano schwabel
    you are sure mel that 20 k will last when you make it really right and the bearbox itself is not broken?
     
  16. Melvok

    Melvok F1 World Champ
    Owner Silver Subscribed

    Jul 25, 2008
    14,101
    Amersfoort, The Netherlands, Europe.
    Full Name:
    Mel
    Yes, you are right Romano, that should be around the price in Euro's .... :D

    And of course, the F factory is happy to refurbish your oem gb for KEuro 25 + VAT + freight. (Not included the de-install and install ...)

    Guess what diff will be in the gb then ? .... Yes, a oem one ... that will break again after 20.000 miles ...
     
  17. bpu699

    bpu699 F1 World Champ
    Owner Silver Subscribed

    Dec 9, 2003
    16,206
    wisconsin/chicago
    Full Name:
    bo
    I see a lot of folks on here advocating replacement of the diff just "in case." I suspect the folks on this board are way more proactive and worried about their cars than the general ferrari owner...

    How are dealers handling this?

    If you take your pristine tr for its service, and it drives fine, are dealers recommending the diff be changed to the tune of $6-7000?

    Would the average ferrari owner, who isnt on this board regularly with 5000+ posts want the diff changed? Or, would they assume it was a rip-off?

    Changing the diff is a great idea. A great $3500(if you do it yourself) to $7000 idea if you pay a dealer...

    Thats a lot of dough for a "what if."

    Some enterprisng ferrari guy should come up with a nondestructive, in situ, test for the diff that can be done...

    Scope?
    Fiberoptic?
    Tuning frequency? (ie. cracked items resonate at a different frequency)

    Why wouldnt a tuning frequency test work? Hit it with a tuning fork, and see what a noncracked one, vs a cracking one, sounds like...

    And if per chance this works, I hereby claim ownership to the idea and want a 5% royalty :).

    Why, would that not work?

    Bo
     
  18. Melvok

    Melvok F1 World Champ
    Owner Silver Subscribed

    Jul 25, 2008
    14,101
    Amersfoort, The Netherlands, Europe.
    Full Name:
    Mel
    Too many Q's in one post :D

    Lost of TR owners are not on Fchat still, not aware that this super Forum exists ... believe me !

    If they were, they could read of these problems and if they see the risk of not changing diff could cost such a fortune, I am almost sure they would think of exchanging the diff.

    If only they knew ...

    F dealers sometimes hesitate to install non-oem parts ... One client/friend had to take out his TR and bring it to another ...

    What would YOU do Bo?
     
  19. bpu699

    bpu699 F1 World Champ
    Owner Silver Subscribed

    Dec 9, 2003
    16,206
    wisconsin/chicago
    Full Name:
    bo
    Well, I plan to examine the diff in situ and if it looks ok, leave it in. I wish I could say that an extra $4000 or so is in the budget for preventative maintenance, but like for most, it just ain't so. Thats a lot of money, for something that "might happen." When there is a 90% plus chance it won't, at least in the time frame of several years and the next 5000 miles.

    Honestly, I would bet most owners would say the same...

    My car has 16,000 miles... perhaps if it had 70,000 I would feel more impending doom.

    That said, it would be nice to find a way to evaluate the diff in the car. I still think a frequency analyzer would work great... Any engineers on here?
     
  20. ozziindaus

    ozziindaus F1 Veteran
    Silver Subscribed

    Aug 16, 2012
    9,799
    Detroit
    Full Name:
    Sam

    Bo, I like the way you're thinking. This is similar to how accelerated fatigue tests work but they are typically performed on a bench for a more controlled environment with less noise. The "tuning fork" is replaced with accelerometers to measure resonance or even just a change in frequency due to fracture.

    I would love to share some past studies I have performed to predict the fatigue life of laser welds but unfortunately I'm limited by NDA's but I can tell you that your approach is not far fetched. The only issues I would have with an application based test or "in situ" as you say, are as mentioned earlier, that is, noise factors or unintended/uncontrolled outputs.

    To eliminate these factors as much as possible, you'd have to;

    1. Test a healthy diff over a long period of time and monitor its output.
    2. Visually examine the part to confirm no cracks have initiated. This forms the baseline.
    3. Continue monitoring the same part until a change of frequency occurs.
    4. Visually examine the part again for fracture if this is indeed the best way to confirm a failure

    Here is the dilemma.

    (a) If a crack has initiated, then you have determined your threshold amplitude or resonance frequency depending on your test method but it may only apply to this particular diff due to so many other unknown variables. Also, if you have confirmed a visual crack, then you're likely to change the diff and therefore the test has paid off but this assumes you were able to catch it at precisely the right time since there is little to no evidence that a cracked diff has a known period of running time before catastrophic failure.

    (b) If no cracks have formed, then something else must explain the change of frequency (noise factors) and as mentioned before, there may be plenty of those to eliminate before you reach your answer.

    In the industry, we take a different approach. Rather than monitoring for failure after design freeze, we induce the failure and design for infinite life. This obviously was not done for this particular diff/application.

    And you're right. I wish there was an easy and reliable way to test for cracking before the "dreaded" failure occurs, if it ever does. The money spent as a precautionary measure may be all we have until we understand the true root cause and implement a solution if one even exists short of changing the diff out.
     
  21. bpu699

    bpu699 F1 World Champ
    Owner Silver Subscribed

    Dec 9, 2003
    16,206
    wisconsin/chicago
    Full Name:
    bo
    Why can't it be simpler?

    Take a 4000hz tuning fork, apply to 3-4 brand new diff. Record frequency given off... That's normal.

    Take a diff that has a fracture. Do the same. That result represents complete failure...

    Now, you have a range...

    Add in some standard deviation assumptions...

    Let's say normal resonates a t frequency "x." As long as you are within 10% of that, you are low risk. I.e., no large fracture line that effects the diff...

    If the result is half "x," you are screwed...

    This could work...

    Same technique as seeing if tile is stuck to cement, if a glass sculpture/ vase is busted, or a violin is solid or delaminating...

    Whack the tile with a fist. High pitched knock, it's attached. Thump, and it's not.

    Take a crystal vase. Tap it. No fracture, it resonates. Fracture, it doesn't.

    Violin... Cracked seam? Doesn't resonate...

    There has to be a way to find a fracture, without complete disassembly...
     
  22. ozziindaus

    ozziindaus F1 Veteran
    Silver Subscribed

    Aug 16, 2012
    9,799
    Detroit
    Full Name:
    Sam
    I understand what you're getting at but let's be clear on a few assumptions.

    1. Are ALL the diffs installed?
    2. Are the boundary conditions all the same? Oil levels, oil viscosity, friction plate pressure, gear lash etc......sure there are more factors that directly affect vibration response.
    3. Can you detect a response difference between the good and bad?
    4. If there is a distinguishable difference, does this necessarily mean it is good?
    5. What is the uncertainty between a part that is marginal and total failure?

    3. 4. and 5. concern me the most and here's why.

    Assumption 3.

    In order to have 95% confidence (Alpha risk = 5%) that you can distinguish a good from bad diff, your response must be separated by >2 standard deviations otherwise you run the risk of failing good parts.. This also assumes Normal distribution. Alternatively, we can run a 1 tail t-tests assuming cracked diffs ALWAYS have a lower frequency and vise versa.

    Assumption 4.

    Worse than assumption 3. is failing to detect bad parts. This is referred to as Beta risk.

    The best way to combat 3. and 4. is to determine your sample size through Power-of-Test. I wont get into the details but this step is critical in providing the level of confidence in your analysis moving forward otherwise you're taking on a lot of risk. Cross you fingers that you don't have to run hundreds of samples to gain confidence. All depends on the separation of response between good and bad diffs but don't assume a few samples will be enough.

    Assumption 5.

    This assumes a cracked diff can survive X amount of miles. We've been told that very few have been confirmed to have cracks but have yet to completely fail. Obvious question would be, how lucky do you have to be to find your diff in THIS volatile transitional phase? Unless you're monitoring in real time and have a known window of safety, then you're still running the risk of being caught off guard.

    So I'm not saying it cannot be done, just saying it's not going to be that straight forward especially if you want a high level of certainty. But if you can prove to me that there is a significant margin between the vibration responses of good and bad diffs, then I'll be happy to work the statistics out with you.
     
  23. bpu699

    bpu699 F1 World Champ
    Owner Silver Subscribed

    Dec 9, 2003
    16,206
    wisconsin/chicago
    Full Name:
    bo
    Like you said, its all about stats. I think of it in terms of probability, sensitivity, and specificity...

    If you have an owner that would never change his diff unless it broke, don't check.
    If you have an owner that will replace it preventatively without a financial whimper, don't check.

    For all others, the risk/reward of a test, though not perfect, has value.

    If the test is sensitive enough to find any issues, the fact that it isn't specific is less important.

    IE. A very sensitive test would catch 90% plus of failing diffs, but miss 10%. Most folks could live with that. There will always be that false negative rate.

    Using the tuning fork example, it might be difficult to create at test thats very specific. IE. If the diff isnt broken, it will show it isn't broken. The test may show false positives due to any of the issues you mention. There will always be variation.

    But, given those constraints. There is still value. Pretest probability matters too... how often does a problem occur? 100% chance? 1% chance? Based on what I gather here, risk for any given year is 1-3% at best.

    A test thats very sensitive (ie. DIFF resonates at a certain frequency when new, and your diff does the same)... confers 90% confidence you are ok...

    If the test isnt normal (your diff resonates aa a different frequency), then all that means is you need to investigate further. IE. Take the diff out. Or replace it.

    No test is perfect. And for those that can swing $4000 with no issues, this is all pointless. But for most, $4000 is $4000.

    Kind of like doing an eval on a poor running motor. You don't just jump in and rebuild the whole motor if it starts acting up. You don't just take off cylinders. You might do a leakdown or compression test first.

    If the test is normal, its not a cylinder/piston issue/valve. Test is very SENSITIVE.

    If the test is abnormal, it could, or could not, be serious. Could be a high false positive rate (misaligned timing, carbon, operator error, etc, etc)...

    For such an expensive part, some type of testing prior to just replacing it, seem warranted...

    This seems pretty doable... Anyone have a way to check frequency of a tuning for against a normal installed diff, vs a failed installed diff?

    I am not enough of an amateur engineer to see this through, but I bet someone can run with it.

    Bo
     
  24. timeckart

    timeckart Formula Junior

    Apr 9, 2015
    557
    Wildeck in Germany
    Full Name:
    Tim Eckart
    #199 timeckart, Jun 22, 2016
    Last edited by a moderator: Sep 7, 2017
  25. timeckart

    timeckart Formula Junior

    Apr 9, 2015
    557
    Wildeck in Germany
    Full Name:
    Tim Eckart
    #200 timeckart, Jun 22, 2016
    Last edited by a moderator: Sep 7, 2017

Share This Page