2004 R1150RT Wideband O2 Sensor Project (and AF-XIED for BMW)

Discussion in 'Vendors' started by roger 04 rt, Dec 13, 2011.

  1. roger 04 rt

    roger 04 rt Long timer

    Joined:
    Nov 9, 2011
    Oddometer:
    2,729
    Location:
    Massachusetts
    Hey PS,
    It's interesting how much careful attention is paid to valves and throttles, getting the air intake right, and how little any of us know about FI matching. During all the modes (e.g. Cold start, warm up, accretion and high power) where the Motronic is open loop and the AFR is less than 14:1ish, fuel injector matching, acceleration) several percent doesn't matter for the reasons we've been discussing.

    However during Closed Loop, when the average of the two cylinders is 14.7:1 (biased toward the richer cylinder a bit), the leaner cylinder can be in the 15s for AFR, creating added vibration, and Motronic induced closed loop surging. That surging could be especially true with an older, sluggish O2.

    Lots of good stuff to get into here. Sometime we should go over adaptation values and their effects. But that's another topic all together.
    RB
    #61
  2. roger 04 rt

    roger 04 rt Long timer

    Joined:
    Nov 9, 2011
    Oddometer:
    2,729
    Location:
    Massachusetts
    For anyone interested in RT spark timing, here is the ignition plot from a short ride I took in January.

    The left axis is degress for dwell and timing, and also degrees C for engine temperature. The right axis is RPM. The data was taken from my GS-911 and is for an '04RT. It looks like I might have made the ride with the Cold Start lever up all the time.

    [​IMG]
    #62
  3. roger 04 rt

    roger 04 rt Long timer

    Joined:
    Nov 9, 2011
    Oddometer:
    2,729
    Location:
    Massachusetts
    Idle dwell at 8 degrees. Even spread for most of the ride between 10 and 40 degrees , max advance is 43. Cold spark timing at 12 degrees advance.

    Notice that this spirited twisty ride was mostly under 4000 RPM. This is due to the added power I'm getting running closed loop fueling at 13.8:1 instead of stock 14.7:1.
    #63
  4. roger 04 rt

    roger 04 rt Long timer

    Joined:
    Nov 9, 2011
    Oddometer:
    2,729
    Location:
    Massachusetts
    Yesterday I was able to borrow an oscilloscope for a few hours. It was very interesting to see the raw waveforms being sent by the LC-1 (Innovate Motorsports) to the narrowband input on the Motronic ECU. The persistence of the scope I used wasn't long enough for photos so I'll just attempt to describe what I measured.

    As a baseline, if I was examining a functional narrowband sensor during Closed Loop operation, I would see a waveform that simply alternated between 100mV for half a second or so and then 900 mV for half a second or so, in a repeating cycle until the Closed Loop mode ended--usually due to a change of throttle. The Motronic would be creating this alternation by steadily increasing its injector pulse width until the O2 sensor jumped to 900mV. Then in would start slowly decreasing the pulse width until the O2 sensor voltage abruptly fell to 100mV--this is what creates Closed Loop operation. The frequency of this alternation would be about once per second or so. (As an aside, if the fueling table said that the average pulse width for 14.7:1 was, for example, 2.00 mS; but the average pulse width for Closed Loop was 2.08 mS; then the Motronic would "learn" an "adaptation value" of +4% and it would start there next time, it would also add 4% to similar Open Loop fueling. Also, in Closed Loop I might have seen a +/- 3-4% spread between richest and leanest values, for a 6 to 8% swing in fueling.

    What I learned by looking at the LC-1 real time waveforms (there are two analog outputs--one for the motronic and one for the gauge) was the following:

    1) As I have the LC-1 set up, the Motronic is able to alternate the mixture once per second, just as with the narrowband sensor--this is good, it means the LC-1 solution is compatible with the Motronic.

    2) In Closed Loop, the Motronic tries a new step every 30 to 50 mS (50 thousandths of a second).

    3) Each Motronic fueling step is about 1%. So if it started at, say, 2.00 mS it next tries 2.02 mS, then 2.04 and so on, same step size on the ramp down.

    4) The LC-1 is nearly instantaneous in its response. As a result, I observed a narrow spread of +/- 1% to +/- 2%. The stock narrowband O2 sensor is 2 to 4 times as large because it responds more slowly, which creates a variation that is probably felt as surging.

    5) Because the LC-1 is so fast, the 100 mV to 900 mV transition isn't a single jump from one voltage to another, it is a series of 100 or 200 mV steps. The Motronic handles this difference without a problem.

    Going back to point 4) has given me the idea that by using an LC-1, I might be able to run without surge or hesitation or weakness at AFRs above the 13.8 to 14.2 that I've tested so far. Next I will try 14.7:1 (stock AFR) and then 15.2:1 to see how those mixtures run. Maybe we can "create" some more gas mileage for those who aren't interested in the power increase that 13.8:1 brings.

    I will probably look at this some more next time I borrow a scope.

    RB
    #64
  5. roger 04 rt

    roger 04 rt Long timer

    Joined:
    Nov 9, 2011
    Oddometer:
    2,729
    Location:
    Massachusetts
    For anyone interested to read about air balancing vs fuel injector balancing vs richer mixtures. The Background on page 7 of this Fuel Injection Patent is interesting reading.

    It makes four key points:
    --Rough running under lean conditions is caused by power imbalances between cylinders
    --Air Imbalance is not the main cause of these power imbalances
    --Fuel injector differences directly result in power imbalances.
    --Fuel injector imbalances can be solved by carefully matching injectors, or by running mixtures richer than stoichiometric (14.7:1 AFR)

    The Wideband O2 project has taken the richer mixtures route so far. Although I'm considering getting a matched set of injectors and seeing how much the AFR can be pushed in the lean direction.
    #65
  6. Poolside

    Poolside Syndicated

    Joined:
    Apr 1, 2003
    Oddometer:
    11,887
    Location:
    Silicon Beach, CA
    <table border="0" cellpadding="5" width="650"><tr><td>Injector balance does make a difference on aircraft motors. Far more so than on a road vehicle. As a pilot you certainly know much about what's explained below. But the injector issue that causes a surging problem on the 1100/1150 isn't really related to the injectors. It relates to batch injection verses sequential injection.

    For a 'continuous flow' aircraft fuel injection system injector balance is very relevant. Injector flow rate makes a big difference in an aircraft continuous flow system because the injector is merely an orifice. Fuel is flowing 100% of the time, and at the same rate, for the full 720° of the combustion cycle. In an aircraft continuous injection system, all injectors are on, all the time. An electronic fuel injector is on only a small portion of the time.

    In addition, the motor in an aircraft at cruise is operating at a very high 'percentage of full power' setting. Certainly a very high power setting compared to that of a road vehicle at highway speeds, which is closer to 15%.

    Any injector imbalance on an aircraft motor is amplified by the the continuous flow rate, and the high power settings. Any imbalance error is multiplied 10 fold at least. It's a cumulative error, you know?

    Injector imbalance on a road vehicle at highway speeds or lower doesn't make too much of a difference. The low power settings and intermittent injection period limits the accumulated error. Injector imbalance makes a little difference in smoothness, but it isn't a source of surging.

    The oilhead boxer motor has 'batch injection'. Batch injection, especially on a two-cylinder motor, makes a significant contribution to surging.

    It's a little hard to explain. To summarize it's caused by the fuel charge arriving in two pieces, not one. That's what 'batch injection' is, multiple cylinders are injected as a batch. The result is the fuel for one cylinder arrives in the intake port at a different part of the 4-stroke combustion cycle than for the other cylinder. Which means that at the moment the respective intake valve opens, the fuel for that cylinder will be at a different degree of vaporization than the other. Unvaporized fuel does not combust. The difference in available combustible fuel radically changes the power level for that cycle.

    A pretty good explanation of the fuel vaporization topic is in this Wall-Adhered Fuel patent.

    How did this happen? Because the Motronic ECU on the 1100/1150 is a single cylinder device. Bosch called it Monotronic. Here's how the 'mono cylinder ECU' works on a two cylinder motor. As far as the ECU is concerned it is operating a single-cylinder motor, but one that's one-fourth the total displacement of the two cylinders, and turning at twice the RPM.

    But all of the above aside, and as you've experienced, if the fuel mixture is increased a little it cures a lot of ills.
    </td></tr></table>
    <BR>
    #66
  7. roger 04 rt

    roger 04 rt Long timer

    Joined:
    Nov 9, 2011
    Oddometer:
    2,729
    Location:
    Massachusetts
    PS, Thanks for your interest and thoughtful feedback. In the quote section below, I've added some thoughts and comments. And the Wall-Adhered Fuel is good reading, and especially relevant when the engine is cold. However, even when warm there can be puddling, with an average evaporation rate of the puddle that has to be taken into account by the FI system. My comments below assume a tuned and air-balanced engine. RB

    #67
  8. Poolside

    Poolside Syndicated

    Joined:
    Apr 1, 2003
    Oddometer:
    11,887
    Location:
    Silicon Beach, CA
    <table border="0" cellpadding="5" width="650"><tr><td>If you go ahead with locating matched injectors, have them matched at the pulsewidth of interest. You know, at the injector pulsewidth centered on the range of interest. The injector isn't open for very long when the motor is operated at low power settings. As you can imagine, low power settings equal short duration injector pulsewidths.

    The injector solenoid and valve mechanism takes time to open. That time period is known as injector dead time, or injector latency. An ECU accounts for this injector latency in the fuel calculations.

    The shorter the injector pulsewidth, the greater percentage of the pulsewidth is devoted to opening the valve, not flowing fuel. Consequently, the primary contributor of injector mismatch at short pulsewidths is the 'latency period' of the solenoid windings and valve mechanism.
    </td></tr></table>

    <BR>
    #68
  9. roger 04 rt

    roger 04 rt Long timer

    Joined:
    Nov 9, 2011
    Oddometer:
    2,729
    Location:
    Massachusetts
    Poolside, That's an excellent detail you've added here ... believe it or not I was posting the same thought on another site as you wrote this.

    At idle and low throttle, the total pulse command can be roughly 2-3 mS, injector turn on can be in the vicinity of a millisecond making it a large potential error. That led me to think I will ride the bike to the shop and see how the handle duty cycle. I also understand that the turn on time can vary with battery voltage.

    Can you recommend an excellent shop for this type of work in the Northeast?
    #69
  10. Poolside

    Poolside Syndicated

    Joined:
    Apr 1, 2003
    Oddometer:
    11,887
    Location:
    Silicon Beach, CA
    What type of work do you mean? Matching injectors? Or "handling duty cycle"?

    <BR>
    #70
  11. roger 04 rt

    roger 04 rt Long timer

    Joined:
    Nov 9, 2011
    Oddometer:
    2,729
    Location:
    Massachusetts
    Here's a curve of injector latency versus battery voltage

    [​IMG]
    #71
  12. roger 04 rt

    roger 04 rt Long timer

    Joined:
    Nov 9, 2011
    Oddometer:
    2,729
    Location:
    Massachusetts
    Here are some questions/comments from elsewhere. I hope that it's okay to include here and that some find it interesting and useful. It adds to what poolside and I were exchanging earlier.


    All good questions. My main reasons for showing that curve was to confirm the approximate turn on time, and to show that it varies with battery voltage. And of course I know as well as you that the higher and lower voltages on that chart are mostly irrelevant, although cranking voltages do drop, especially if you have a weak battery, I saw 11V the other day on mine (I need a new battery).

    I'm going to answer your questions/comments in the context of an R11xx engine that doesn't run its best at a 14.7:1 closed loop mixture unless things are really well balanced. That's why we get our valves to less than one thousandth, and balance our throttle bodies at multiple RPMs, with a precision differential instrument. I noted earlier that we give almost no attention to fueling imbalances which are just as important to total cylinder power balance. (Nor the O2 sensor)

    Your last point, if battery voltage variation was an issue the manufacturer would use a regulator, is valid. And it turns out the manufacture does regulate for battery voltage at the fuel injector. The way it is done is: the Motronic measures the battery voltage, and then looks up in a table what the "net dead time" is and adds that into the fueling calculation.

    I added the curve to this thread becasue it shows just how large the "net dead time" meaning the time that the Motronic must add can be--roughly 1 mS at 13.8 volts. At idle, the injector fuel pulse is in the vicinity of 2 mS and at 3500 RPM cruise, about 3.5 mS. So the net dead time is half the idle pulse and almost 1/3 of the 3500 RPM cruise. Any net dead time difference between the injectors will have a significant difference: a 0.1 mS difference would be 10% of the fuel on time at idle, and 4% at 3500 RPM cruise.

    Here is a picture that gives an even closer look at how they work.

    [​IMG]

    In the picture you can see that net dead time is comprised of:
    --delay to start turning on
    --time to ramp up
    --delay to stop
    --time to ramp down

    Another point: the 7 mS time is the total injector time in my example too since our Boxers fire the injectors twice per combustion cycle--half the fuel twice, with two dead time cycles. So my 3.5 mS per injection, doubled is very close to the example given above.

    So my points on injector imbalance are that there can be different flow rates and different net dead times, and both are important. And for those of us trying to balance our cylinder power AIR (valves and TB) are one side of the equation and FUEL accuracy is the other. Both are equally important, UNLESS you run a mixture richer than 14.7:1. If you run at 13.8, where all the European CO potentionmeter motorcycles were set, you are almost insensitive to small fueling differences, up to several percent. Then a good TB and Valve job gets it about right.

    More food for thought.
    RB
    #72
  13. roger 04 rt

    roger 04 rt Long timer

    Joined:
    Nov 9, 2011
    Oddometer:
    2,729
    Location:
    Massachusetts
    I know that a few of you have wondered what fuel mileage would be at the richer 13.8:1 AFR I'm running.

    Just ran my first gas mileage test using the LC-1 Wideband O2. All the driving was on local trips, no highway, up to 25 miles per trip, some as short as 5 miles, plenty of lights.

    I burned 2.48 gallons, for 106 miles. That's 42.7 MPG, give or take.

    I still plan to take a highway trip and see what the mileage is.
    #73
  14. twinsig

    twinsig Been here awhile

    Joined:
    Feb 23, 2012
    Oddometer:
    356
    Location:
    Middle Alabama
    Following this subject here & at BMWST.

    New owner of 02 1150RT & waiting on TB shaft kits (sucking air), will do my first valve/sync job very soon. Right now poor idle unless engine is cold, sometimes almost no idle when hot, annoying to say the least.
    I'm slightly anal about tuning & performance but no time nor the knowledge to do much about it except purchase the things it needs by those who solve the problems.

    I'll be back..
    #74
  15. roger 04 rt

    roger 04 rt Long timer

    Joined:
    Nov 9, 2011
    Oddometer:
    2,729
    Location:
    Massachusetts
    When youve fot it reasonably tuned and you're ready let me know, I have a loaner Boosterpug.

    For anyone who'd like to feel the effect without committing up front to something like the LC-1, I've persuaded Jens at BoosterPlug to loan me an 1100/1150 unit, for people who wanted to try the standard coding plug, unplugged O2 sensor and BoosterPlug configuration which gets AFRs to 4-6% richer (13.8 to 14.2 depending on fuel and specific bike tolerances). While Jens didn't design it for this application, he agrees that it will have the desired effect.
    #75
  16. HackyMoto

    HackyMoto TEAM USA GS TROPHY 2012

    Joined:
    Nov 19, 2002
    Oddometer:
    1,077
    Location:
    Oh it's you! Anchorage, Alaska 99507
    :thumb :lurk
    #76
  17. twinsig

    twinsig Been here awhile

    Joined:
    Feb 23, 2012
    Oddometer:
    356
    Location:
    Middle Alabama
    Thanks for the offer and I may take you up on it in the near future. Waiting on TB rebuild kits by Dan, recently did valve job, idles way better, runs way better/smoother. 2 months now and just tonight I "punched-it" really for the first time. Soon enough I will start wearing down that last 3/4" of tread on each edge of the rear tire!

    My name is Twinsig, and, I am an addict!
    #77
  18. roger 04 rt

    roger 04 rt Long timer

    Joined:
    Nov 9, 2011
    Oddometer:
    2,729
    Location:
    Massachusetts
    I've been running with an LC-1 and BoosterPlug (or it could be an IICE Air), kind of a belt and suspenders approach. The LC-1 for Closed Loop and the BoosterPlug for Open Loop. For some time I have wanted to see how the LC-1 and Motronic performed without the BP shifting the stock Open Loop AFR tables 6% richer, because I know that on the R1150RT Adaptation Values get built from Closed Loop and get applied to Open Loop. Confirming that the Motronic worked that way would mean that the BoosterPlug was a nice-to-have (since I own one) in my application, not a need-to-have.

    A key to these tests is a cold start, monitoring how the Motronic and LC-1 transition from cold (enriched and Open Loop) to warm (no afterstart or warmup enrichments, and Closed Loop). Therefore both tests were begun immediately after start and on the same course. (Note: In the upper chart, second half, I was modulating the throttle more than in the second half of the bottom chart. Otherwise, they are for the same course.)

    In the attached photo, you can see the first ride (top chart) after the Motronic is reset and with no BP; and then a day later, the second ride in the chart below. In the first ride after reset (in my theory, before the Adaptation Values referred to in the BMW Service manual can be rebuilt) you can see a lot of deviations from 13.8:1. If you look carefully on the left of the top chart you can see the mixture starting at about 12:1 and then watch what happens while the bike warms up at about the 3 minute mark (same point on the second day, bottom chart). You can also see the histogram on the right, that being the distribution of all AFRs during each ride.

    On the bottom graph, second ride from a cold start, you can see better adherence to 13.8 and a tighter histogram. (Ignore the spikes that go to the top of the chart, they are Overrun Fuel Cutoff.) It also gets to 13.8 right away, before the bike has warmed up and before it can be in Closed Loop, in the first minute or so. (Notice the dips to lower AFRs, that's where I'm accelerating and changing gears.) The only surprise in the lower chart is the blip to 14.7:1 (Open Loop target AFR of the reset Motronic) between 1:40 and 3:20 minutes. This was after a stop sign and reacceleration, before CL had started and perhaps (my deduction) in an area of the Adaptation Values table that had not built sufficient data yet.

    I plan to make many more rides and watch this closely and will have some more data over the coming weeks.
    RB

    In the photo, there are two horizonal dotted lines. The top line is 14.7:1 (Open Loop Table before Adaptation) and the bottom line is at 13.8:1 (My AFR target set by the LC-1).
    [​IMG]
    #78
  19. erkmania

    erkmania Last of the red hot left pipers

    Joined:
    Mar 19, 2004
    Oddometer:
    1,160
    Location:
    San Diego, CA
    I've only gotten this far in the thread so far. Please forgive redundancy if the repair has already been done.

    Baking soda and Super Glue combined should work just fine. Sprinkle the joint with the powder and then drip Super Glue onto the damaged area. Scoring the surfaces with sand paper should help. In years past, this technique worked on my damaged Honda F2 instrument cluster mounting tabs.

    Plastic welding with similar plastic will work, too.

    Cheers
    #79
  20. roger 04 rt

    roger 04 rt Long timer

    Joined:
    Nov 9, 2011
    Oddometer:
    2,729
    Location:
    Massachusetts
    Yesterday afternoon I disconnected the Wideband O2 sensor so that I could measure the "raw" or Open Loop, Air Fuel Ratio (AFR) of the Motronic MA 2.4., with Pink Coding Plug. And then measure the Open Loop effect of the BoosterPlug. The motorcycle is running 93 Octane E10 fuel, which is important since it has shifted the curves upward by about 4%, meaning running 4% leaner.

    In all the charts, there are two marker lines, one is at 14.7:1 (normal Closed Loop stoichiometric) and 13.8:1 (my richer Closed Loop target that creates more power). In the first photo there are two charts: the top chart is Open Loop, the lower Open Loop plus BoosterPlug.

    The top chart below shows Open Loop AFR, after the Motronic is reset, and from a Cold start. For the first three minutes, you can see the AFR moving from 13:1 to the low 15s as the bike oil warms up. If I were running pure gas, the warmup would be in the 14.7 range, but E10 leans out the mixture. Much of the cruising time, the AFR is between 15:1 and 16.5:1--fairly lean. Notice in the histogram inset chart that the bulk of AFRs are between 14:1 and 16:1, even including acceleration.

    The in the lower chart, the bike is warm, the Motronic is still reset but a BoosterPlug has been connected, dropping the air inlet temperature (AIT) signal to the Motronic by 20C. There is a similarly large spread of AFRs, but now they run between 13:1 and 15:1, about 6-7% richer. I was surprised at how much the AFR varies in Open Loop operation.

    [​IMG]

    For comparison, look at the next photo, the conditions are reset Motronic, no BoosterPlug and Closed Loop at 13.8:1. The upper run is the first test run after reset and the lower run is the second run after reset. Since the Motronic is only Closed Loop about half the time in these charts, and comparing these two runs to the first photo (Open Loop conditions), you see Adaptation, the learning of Closed Loop applied to Open Loop conditions, and in the second run looking at the even tighter spread of AFRs, further Adaptation.

    [​IMG]

    Some conclusions:

    1) E10 fuel results in leaner Open Loop operation
    2) Open Loop fueling results in a wide variation of AFRs
    3) The BoosterPlug does enrichen the fueling tables by 6-7%
    4) Closed Loop operation tightens the fueling range
    5) Closed Loop operation allows the Motronic to correct for E10, aging injectors and air filters, and variations in battery voltage and fuel pressure.
    6) The Motronic is a learning, adapting engine controller that takes what it measures in Closed Loop operation and applies it to adjust Open Loop fueling. (The mechanisms and degree could be examined further.)
    #80