1. Livestream NOW at 11AM PST / 2PM EST!
    Baldy Interviews + Noraly "Itchy Boots" from Youtube YouTube Live Chat here
    Dismiss Notice

Malfunctioning F800GS

Discussion in 'Parallel Universe' started by MotoLara, Jan 10, 2011.

  1. JoelWisman

    JoelWisman Long timer

    Joined:
    Apr 22, 2009
    Oddometer:
    1,392
    Location:
    STL, MO, USA
    Some of this is market conditions. I don't think there are more then a couple dozen F8's touring the world while thousands upon thousands spend there whole life in well developed countries.

    Simply put, there aren't that many F8's suffering from the lack of a knock sensor and it is quite possible that the bad press for the few that do and the fewer still that get low enough octane fuel to knock and the smaller yet percentage of those that are piloted by a rider that doesn't recognize knock and take appropriate actions to prevent failing the engine.

    Bad press is probably cheaper then installing a knock control system on all F8's made.

    Personally from a position of relative ignorance I firmly believe that BMW should come out with an F800 GSA that has stronger wheels, better suspension, a tougher fuel pump, bigger alternator, bigger fuel tank, and knock sensors.

    There are 3 types of knock detection on gasoline induction engines.

    1: Spark creation sensing. The higher ambient pressure is, the more voltage it takes to create a spark through a given electrode gap. This type of knock detection can only detect massive detonation and usually is allowed very limited timing control as it is very prone to false alarm from aging wet or fouled spark plugs.

    There is a rumor that the K7 platform DOES use this form of knock detection. I personally have been told by an FSE that the F8 DOES have this system and also told by a German engineer that it does NOT. If the F8 does have this system, keeping the spark plugs new and not fouled will improve this systems limited function. If the bike is runing rich, spark will flash over the carbon on the plug and never trigger knock detection. If the electrode gap is allowed to increase, the system can't sense the difference between knock and what it takes to fire the worn plugs and disables itself.


    2: Narrow band pizeo resonance sensors. This is the common system and what the R1200 uses.

    (A): A tuned pizeo is used as a knock sensor and is mounted fore and aft along the cylinders on 1 or more bosses on the motor. Since we only have 2 cylinders on 360 degree combustion cycles, one sensor would do just fine and there is an unused boss on the rear of the motor that is perfectly placed, almost as if BMW was considering adding this feature. Pizeos are tuned by how tightly they are torqued but the torque has to be precise for the knock sensor to work well, much more precise then the crap guns Germany uses to assemble the rest of the bike, so more time in production with a single item torqued with a strain gauge or at the least a well calibrated torque wrench which is what they use on the R1200GS.

    (B): You need a cam position sensor to tell you which cylinder is on the compression stroke when knock occurs. No problem, we have one of those on the F800.

    (C): You need a HIGH resolution crank position encoder to know EXACTLY where the piston is when knock occurs and it needs to be high resolution to see changes in acceleration of the crank to ferret out knock from a rock impacting the case. The R1200 uses the starter ring gear. We don't have a starter ring gear, we have a cheaper, lighter, more robust sprag clutch, so BMW would have to add more encoding notches in the flywheel and use a more expensive position sensor to see it.

    (D): A lot of simulation time to find knock patterns AND develop a response algorithm that meets demands while, and this is the hard part, still passing many markets super strict emissions standards.

    (E): A larger and more robust catalyst to cope with carbon loading during late stage knock control without plugging up or passing excessive emissions to the environment.


    3: Wide band knock sensors. These are great, but you aren't going to see them on motorcycles anytime soon. Only 3 manufactures have the capability of designing these systems and they charge a pretty penny to license their technology.


    There are 4 types of detonation inception. 1: spontaneous ignition induced by a non homogenous mixture. 2: Compression ignition induced by high combustion chamber temperatures. 3: Auto ignition induced by after glow. 4: pre-ignition produced by the presence of combustion as mixture is inducted.

    Here is an effective knock control algorithm I had a very small hand in developing. (actually I was just involved in figuring out why the damn system was triggering early on some engines)

    Knock occurring between 15 degrees before and after desired ignition. On next cycle retard ignition 5 degrees, if not present next cycle, go forward 5, repeated, if keeps occurring at forward position, write retard into long term map but check occasionally if it is still there forward.

    If 5 back doesn't do it on 3 consecutive cycles, retard 8, repeat as above but if present 4 consecutive cycles, retard 15 degrees. back and fourth as above, if still there 3 more consecutive cycles, increase injector time 5% and listen for 500ms

    still there 80% of cycles? increase injection time 10% and listen. Still there, shut down that cylinders injector every other cycle and flash the check engine light incase the operator is not savvy enough to notice his engine is running like shit, post excessive knock cylinder x code and snapshot engine parameters.

    If the knock is detected more then 15 degrees before desired ignition, we have after glow. I don't remember how the response to this kind of knock was different but it was different.

    Ok, theres the background on the subject of knock detection.

    Why is knock bad and how is it different from desirable ignition???

    Low fuel consumption and high efficiency is promoted by high combustion rates and optimal combustion patterns. Maximum combustion heat release should occur 5-10 degrees after top dead center. If most of the heat is released too early, wall heat losses are increased. Higher compression increases part load efficiency and maximum load power.

    Durring knock, instead of a slow flame front propagation from the spark plug to the furthest reaches of the combustion chamber (end gasses) pressure and heat instantaneously ignites all of the end gasses at once. This creates a shock wave that propagates at the speed of sound and local temperatures that can be more then 10 times higher then during normal combustion.

    In plain english, knock is an explosion rather then controlled combustion in the engine. It blows and melts holes in the piston and the shock waves destroy piston and crank bearings.


    There, thats about the limits of my knowledge. More to follow on the practical of how to notice it in the field and what to do about it.
  2. Bayner

    Bayner Long timer

    Joined:
    Mar 27, 2008
    Oddometer:
    2,372
    Location:
    Penticton, BC, Canada
    I'm pretty sure the throttle position sensor is being used. It serves this function in many bikes running without fuel injection for that matter. On a separate note I know several (myself included) who have had BMW (Canada) reprogram for lower octane fuels. My local dealer had no idea what I was talking about when I first asked them. They looked into it, and a few months later let me know they were now equipped to perform this procedure. I get the feeling it really just depends on who is on the other side of the counter on the day you ask about it.
  3. _cy_

    _cy_ Long timer

    Joined:
    Sep 25, 2011
    Oddometer:
    6,845
    Location:
    Tulsa, Oklahoma
    Joel, NICE writeup!! appears this topic is not new to you :rofl
  4. _cy_

    _cy_ Long timer

    Joined:
    Sep 25, 2011
    Oddometer:
    6,845
    Location:
    Tulsa, Oklahoma
    throttle position sensor cannot tell ECU if engine is under load and knocking. although throttle almost always moves before engine loads up before possible knock.
  5. JRWooden

    JRWooden Homeless motorcycle vagabond ... and ... loving it

    Joined:
    Jul 5, 2008
    Oddometer:
    6,244
    Location:
    The great state of confusion
    Thanks Joel, nice explaination!

    I'd be happy with putting a pizeo on my block that simply feed a counter with some simple algorythym that said ...
    hey if I hear what I think is a "knock" 5 times within a 1/10th of a second (whatever) I will light up the big red light on the dash and warn the hoser riding this bike that s/he needs to think hard about what's going on....
    Slow down, take out ear plugs, did they just refuel ... whatever ....

    Do the sensors cost a lot?
    I have a buddy that would help me build a detection circuit...
    Still if you have to have a very well calibrated torque wrench to install... that makes things harder.............

    (Memo to file ... change plugs every 24K miles if not sooner - just to be safe)
  6. Bayner

    Bayner Long timer

    Joined:
    Mar 27, 2008
    Oddometer:
    2,372
    Location:
    Penticton, BC, Canada
    Can the TPS report knocking to the ECU? -- definitely not. Does it report load? -- yes. Has been that way for many years in all manner of EFI programming.


    A knock sensor is usually around 100 bucks if I recall correctly. (Auto applications of course)
  7. JRWooden

    JRWooden Homeless motorcycle vagabond ... and ... loving it

    Joined:
    Jul 5, 2008
    Oddometer:
    6,244
    Location:
    The great state of confusion
    Bayner:
    That would be cool...but how does it report load?

    I thought that the TPS was just a glorified Potentiometer ... reporting the throttle position.
    The ECU can calculate the derivatives of of position and come up with the velocity / acceleration of the plate position,
    but how it can get the load on the engine from all of this?

    How, for example, would the ECU know the difference between me blipping the throttle in neutral waiting for the light to change vs. and me gunning it up the side of Mt. Everest? ...
    Does the ECU take this TPS Input and compare/contrast with other inputs some how??? :huh
  8. MotoLara

    MotoLara ADV rider wannabe

    Joined:
    Nov 27, 2009
    Oddometer:
    989
    Location:
    Salt Spring Island, British Columbia
    Joel, I don't know it's just me... but it sounds like you are repetitively insinuating that me and my partner are incompetent/incapable of recognizing knocking sounds. I apologize if it's just me but you are coming across as a 'I know it all and this would have never happened to me' :deal and I don't like it.

    As I said, the knocking noise was inconsistent (on and off, mostly off) and very mild (but unusual) and a few kms later my engine shut down. The knocking sound was very hard to notice with full-face helmets on + earplugs + the strong winds of the desert that day. The damaged was already done by the time we pulled over and put 3 gallons of the premium gas we had in our rotopax containers.

    I'm not sure if it was bad gas, I think we'll never know. What I know is that there is at least 02 other F8GS with very similar engine failures in Canada and the U.S. where gas isn't supposed to be an issue.

    Alberto.
  9. Bayner

    Bayner Long timer

    Joined:
    Mar 27, 2008
    Oddometer:
    2,372
    Location:
    Penticton, BC, Canada
    Okay, let me see if I can explain myself adequately. I guess first off I would say the TPS is a glorified potentiometer.
    As far as your engine management is concerned, there is no difference between you blipping the throttle at a light or doing it on your way up Everest (besides duration); it's reaction is exactly the same, that being to match the incoming air with an appropriate amount of fuel and lighting it off at the required time. It usually has other sensors to help out with this, such as a mass airflow meter, a manifold pressure sensor, some manner of detecting RPM and crank position to name a few. (I believe the bike just uses throttle position, RPM and some complex math with the occasional help of the oxygen sensor- but someone please correct me if I am mistaken).
    Think of it this way: your old automatic transmission used a vacuum dash pot to determine load on the engine. Why was this effective? Because as you open the throttle you begin moving more air and see a resultant drop in vacuum. The same can be said for the dashpot on a distributor used for adjusting timing. There is a distinct correlation between engine load and throttle position.
    A change in load when the throttle is closed is handled by the idle air control motor to prevent stalling or over revving. In this case the system is simply responding to maintain a programmed desired idle.
  10. JRWooden

    JRWooden Homeless motorcycle vagabond ... and ... loving it

    Joined:
    Jul 5, 2008
    Oddometer:
    6,244
    Location:
    The great state of confusion
    Bayner:
    I probably misunderstood you ... I would agree that clever ECU programming -- using the TPS input and lots of other stuff could guesstimate engine load with some reasonable accuracy........... I was just thinking that TPS by its self would not be a very good indicator...

    In the case of the automatic transmission the vacuum kickdown worked fairly well because of some other stuff going on ... the car was in 2nd gear or higher (so blipping the throttle in neutral/park didn't count...), some serious throttle was being applied (1st derivative of TPS was sizable otherwise the vacuum wouldn't be low),
    so this in turn implied that the car was moving at a decent velocity and accelerating briskly .. and also the engine was running in an RPM range where dropping back a gear would be ok most of the time (known/assumed because the centrifugal Governor in the tranny had not kicked in)

    All good :1drink
  11. itsatdm

    itsatdm Long timer

    Joined:
    Jun 19, 2004
    Oddometer:
    5,440
    Location:
    Nor Ca.
    The bike certainly needs high octane fuel because of both its CR and lean fueling map. How Triumph manages to use regular grade fuel with its 12 to 1 ratio is something I would like to know.

    After suffering a bout of low rpm detonation the other day, I am an advocate of anything that keeps the fuel injectors clean.

    We sometimes think these injectors work or not and that is not how it happens. They normally get varnished or covered with other residue over a period of time. The ECU is holding them open for a set period of time to meet the conditions the ECU is reading from its sensors against what it is stored in its logarithms. The expected amount of fuel is not being delivered because the holes are partially closed or it is not being delivered in the pattern as designed. Both can result in detonation. Contaminated gas speeds up the process.

    Ethanol gets a bad rap for this and maybe it does have a corrosive effect on metal bits in the tank, but it actually has a cooling effect on the motor. By volume it does not produce the energy of gas so besides its evaporative ability, it actually decreases CR.

    The one A/F ratio I have seen on a dyno'd bike had a range of A/F ratios of 17 to 14.2 to 1. I think that is the entire range possible. Pretty lean in my opinion and the motor is not going tolerate some malfunction of the delivery system under load or at idle either in my experience.

    Short of knock sensor coupled with the ability to retard timing I don't know an answer. BMW certainly is unlikely to redo the fueling map.

    As the bike seems to run well when all the parts are working, I will just stay alert to make that happen. It was a broken off piece of the T on the bottom of the air box to the hose connected to the right throttle assembly that start this drama.

    Couple of other little tidbits I discovered on the net. First a BMW800 Service alert dated 2008, that says ethanol up to 10% volume is suitable to this bike. Second is from the California Air Resource Board, stating that the percentage of Ethanol in its reformulated gas is 5.7%.
  12. JRWooden

    JRWooden Homeless motorcycle vagabond ... and ... loving it

    Joined:
    Jul 5, 2008
    Oddometer:
    6,244
    Location:
    The great state of confusion
    Gary:

    How much help do you think one of the several devices that fool the ECU/ZFE into thinking the ambient air is 20 degrees or so cooler would be in all of this?

    (I have an IICE Air on order :wink: )
  13. Bayner

    Bayner Long timer

    Joined:
    Mar 27, 2008
    Oddometer:
    2,372
    Location:
    Penticton, BC, Canada
    The IICE air works great in mine. Definitely smoother initial throttle response. Can't speak to fuel economy yet as it is still too cold here to get comparable numbers. I think you'll like the improvement.

    Going back to the TPS discussion, I guess I would sum it up by saying that it's the rider that makes the TPS suitable for determining load because we will quickly notice a drop in RPM or decide we want to increase speed and modulate the throttle accordingly. It make more sense when expressed that way I think.
  14. itsatdm

    itsatdm Long timer

    Joined:
    Jun 19, 2004
    Oddometer:
    5,440
    Location:
    Nor Ca.
    They work as described but they do not solve this issue, because they are not supplying a richer fuel mixture all the time.

    The oxygen sensor will always cause the ECU to try to return the AF mixture to the ratio it has been programmed to deliver(assume 14.7 for low load riding). If load or the throttle input indicates more fuel is needed, the ECU will do that (apparently up to 14.2), but as the load decreases or speed catches up to what is cranked at the throttle, the oxygen sensor will indicate that to the ECU and it will revert back to its programmed AF for those conditions.

    The module programmers will have to speak up, but I am sure the same happens with these temp modules. The difference is the ECU will assume it is colder than it is, so a little more fuel is pumped in. It still goes back to its programmed fuel ratio when it can. The only difference is when you crank the throttle, it may go to 13.6 instead of 14.2 ratio and there is some lag time needed for the oxy sender to read the exhaust and change its current level that makes the ECU revert back to program.

    What I saw on a dyno video, was a A/F ratio of 17 to one at idle. Crack the throttle and it richened up quickly, reaching something around 14.7 by 2,200 rpm. Works good on the street, but trying to ride slowly between idle and 2,200rpm, it still has to be fairly lean or switching back and forth between too lean and way too lean if your wrist is weak over the bumps. That is what causes the twitchy throttle, IMO.
  15. JoelWisman

    JoelWisman Long timer

    Joined:
    Apr 22, 2009
    Oddometer:
    1,392
    Location:
    STL, MO, USA
    The recent posts here make perfect sense to me and I agree almost 100% But I will add a couple of things.

    Jrwooden. Your thinking of this load thing in the same way a lot of people do and it shows great intelligence but is wrong.

    The ecu is always nearly perfect in its estimation of load. If it wasn't, the engine would die from too lean or too rich of mixture. The ecu is something like 99.99% accurate at judging load.

    Load is not the product of anything that occurs because of hills or gears, but rather the product of energy input less internal losses.

    When the throttle blipped in neutral, at each given RPM the load the engine experiences is precisely the same as it would at the same throttle opening in gear at the same RPM. In any gear or neutral it's all the same.

    Holding the throttle in the same position and starting at say 3000 rpm in 3rd gear, the load will equal x acceleration. In neutral that load is exactly the same except since it isn't coupled to the rear tire, the load will instead rapidly accelerate the engine till redline is reached and the injectors shut down.

    Load=throttle opening at x rpm and is not influenced by anything external. The external simply influences weather that load will rapidly accelerate the engine in neutral, slowly accelerate the whole bike, loose ground and decelerate the whole bike and so on.

    It can be hard to understand this if you have believed otherwise, but you are a super smart guy and if you root around the web or texts I think you will start to think about load otherwise quickly.


    betitolara,

    I'm not insinuating anything about your competence or value as a human :)

    I recognize knock, both at high rpm and low because in my career I have heard it over and over. I also understand the fundamentals of when and why it occurs so know when to listen.

    H/S low octane knock doesn't stand out, but once you have heard it over and over because of a career or life experience, say with a souped up 60's American car during the change over to unleaded fuel, it's very easy to recognize.

    Without that background it would be super unlikely that even the brightest people with the best hearing would recognize H/S knock.

    You have to hear it and be familiar with it, then it just clicks and you will hear it frequently cause even on vehicles with knock sensors, actually ESPECIALLY on vehicles with knock sensors, H/S knock occurs lightly for brief periods all the time. I hear engines knock all the time, and yes I have heard and recognized H/S knock from F8's while helmeted with a balkalava or whatever those things are called in rain storms while doing 100MPH.

    So, no, what happened to you likely wouldn't have happened to me.

    BUT, once again, few people are career technicians that have also built many HP engines and heard knock all the time.

    Your inability to recognize it, assuming thats what happened to your engine is COMPLETELY understandable and the same would happen to many others and is almost guaranteed if age or loud sound has dampened your ability to hear above 14khz, like it has with most over 35.

    I'm not blaming you at all and if I were going to build a bike to go around the world I would defiantly include a knock sensor for this and other reasons.

    As to these problems occurring in first world countries. Bad gas can happen anywhere. I pulled apart an F8 engine that died from low octane fuel with under 250 total miles right here in St Louis USA. In that case the fuel was bad because it had high alcohol content to increase octane, all of which had evaporated because of the dealer leaving the fuel in the bike which had a bad tank vent valve for over a year.

    Not your fault or incompetence, rather a motorcycle design that is not suitable to run on bad gas without damage unless you are one of the few that is really familiar with knock.

    I wish I had a high quality microphone and an F8 that i felt comfortable inducing H/S knock in, cause some practice in a knock "simulator" would make you immune to damaging an engine through knock.
  16. Steveman

    Steveman Been here awhile

    Joined:
    Sep 14, 2011
    Oddometer:
    471
    Location:
    Austria
    Sorry, seems I missed that post, but will go and read it. Thanks!
  17. Steveman

    Steveman Been here awhile

    Joined:
    Sep 14, 2011
    Oddometer:
    471
    Location:
    Austria
    Ouch, thats bad. I thought AFR unit can alter the ignition curve, but I'm obviously wrong. Maybe one could get in touch with Halltech or other companies doing injection systems tuning....
    I think the only practicable solution is altering the map according to the fuel quality, everything else seems very complicated...
  18. roadspirit

    roadspirit souvlaki for breakfast

    Joined:
    Feb 1, 2011
    Oddometer:
    200
    Location:
    Athens, Greece
    :ear
    that would be interesting :)
  19. Midnullarbor

    Midnullarbor Been here awhile

    Joined:
    Oct 13, 2010
    Oddometer:
    828
    I hope that Steveman's comments, about avoiding low-compression pistons, was a statement made very much tongue-in-cheek. [ Post #360 ]

    The root cause of the problem is too much heat . . . from too high a compression ratio, relative to the limits imposed by the [trashy] fuel.
    The oxygen molecules and the hydrocarbon molecules wish to tango together ~ and it is difficult to prevent that happening, once we have brought them together in the grand "Carnaval" dance held in the combustion chamber.

    Yes, a margin of success can be achieved as a chaperone, in delaying the inevitable "fireworks" combustion of the Peruvian fuel/air mixture . . . but that margin is very slight, when the fuel & air are "hot to trot".
    Retarding the spark timing, altering the fuel injection, or other electronic trickery . . . can only achieve a small amount. What would that achievable margin be . . . perhaps the equivalent of 5 or 6 octane points? Not so very much, in view of the potential problem we are facing.
    (And even where successful, the fuel consumption ~ thermodynamic efficiency of the power-stroke ~ is definitely not optimal : it will do for an emergency period, but is undesirable as a long-term approach.)

    A hopeful 6 octane point safety margin is good ~ but probably (and evidently) was not enough to prevent Betitolara's catastrophe.

    Again, what can be done?
    For current F800GS owners, the aftermarket re-mapping plus the carrying of gallons of octane booster [literally gallons, for a big trip into the "backwoods"] for the use on those occasions when a hair-trigger suspicion dictates . . . is all rather unsatisfactory . . . or worse : futile.
    And the fundamental approach : the fitting of low-compression pistons . . . is a major /expensive undertaking ~ and many riders would consider it impractical for a one-off big trip.
    Perhaps more practical to buy/hire a different bike altogether !

    But for future owners : for bikes not yet in the production/delivery pipeline, it would be very little expense at all to specify and fit low-comp pistons [and the appropriate mapping].
    Just one more option box to tick, when pre-ordering.

    I am appalled by Joel's comment/estimate that true adventure-riding F800GS owners are just "one-percenters" of the total fleet ~ please say it isn't so ! [But that would help to explain BMW's inertia on the detonation issue.]
    Thanks also, Joel, for the extra tech information . . . I hadn't realised [in my Dunning-Kruger-ish way] that the piezo-type knock-sensors were so "touchy" regarding their installation.

    Perhaps if the market for hard-core advrider F800GS's is as small as Joel's gloomy statement indicates . . . then there had best be only one sub-category of engine option box to tick : an even lower piston C.R. [say 10.5:1] combined with one piezo-electric sensor [and ECU modified].
    That might well add several hundred dollars to the purchase price . . . but might be attractive to a reasonable number of potential owners, who reckon it a lot cheaper than a longer and costlier unplanned extension of their holiday in South America [. . . or wherever].
  20. LukasM

    LukasM Long timer

    Joined:
    Nov 15, 2006
    Oddometer:
    5,511
    Location:
    On a RTW ride - currently touring the U.S.
    How expensive could it really be for BMW to put a knock-sensor and ignition retardation on the bike? The way cheaper F650GS single has it, and gets away with 11.5:1 compression using the shittiest gas anywhere in the world.

    Since there are no aftermarket FI options right now that implement one, what are possible measures to reduce compression, realizing that none of them are really practical unless you enjoy taking your engine apart? :D

    Increase total chamber volume by:
    - Using thicker head gasket (or doubled stock ones)
    - Machining reliefs into pistons
    - Opening up combustion chamber and valve pocket (ideally combined with some porting)
    - Anything else?