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

BMW could claim, on an in warranty bike that your modification has affected the engine, exhaust, catalytic converter, drivetrain (by adding more power loading), etc., etc., etc. The same could be said for IAT modifiers, Techlustions, and PowerCommanders.

The reality is you're only adding the amount of fuel needed to make your bike run better. In my case with a 9 year-old R1150RT I don't have any concerns at all.

Update on the LC-1 on my 2001 R1150 GS:

The fuel injectors I had sent off for cleaning (RC Engineering) came back today. The report showed they had flow rates of 315 and 294 cc/min before cleaning; and 321 and 319 cc/min respectively after cleaning. The 21 cc/min difference between cylinders before cleaning seems like a lot, which I believe could account for the slight surge I still had after installing the RC-1.

So I installed the cleaned injectors and went for a ride, syncing the TBs after the engine was fully warmed up. The surging is now gone! I had forgotten just how smooth and torquey the GS is with the richer fueling. The LC-1 transformed the bike. I rode like a hooligan coming home – big smile on my face, 30 degree weather notwithstanding.

Thanks Roger for your help every step of the way. For me, this is a modification well worth doing.

That is a fairly large injector mismatch for a lean-fueled engine as yours was with the stock O2 sensor installed. When using the stock sensor I see AFRs move between 14.3:1 and 15.1. Since your cylinders had a 7% fueling imbalance so taking half that amount and adding another 3.5% to the 15.1:1 your leanest cylinder was at times in the vicinity of 15.6:1, a very lean mixture.

Also, around stoic, the fuel is converted to power at about half the difference of 7%, so one cylinder was producing 3-4% more or less power than the other. Your engine should be somewhat smoother now.

As you richened the mixture with your LC-1 but before you cleaned the injectors, you adding fuel and consuming the excess oxygen. As you do that your bike became less sensitive to the injector imbalance because once you've consumed all the oxygen the excess fuel can't be burned and the power differences diminish.

Glad it is working out for you.
RB

Hi guys,
A couple months ago I installed an LC-1 on my '04 R1150RT and finally got some good weather a couple days ago and went for a ride. Most of this info I have posted on the bmwsporttouring sight but for those of you that don't go there I will make a blanket statement: Even if you think your oilhead runs great or good enough install the LC-1 anyway, it's that simple. The difference in lower RPM torque and much smoother running is phenomenal. My '04 has about 42K miles on the odometer and most of those miles were put on ridding in 5th gear. Now, thanks to the LC-1 and all the data posted by Roger, 6th gear is completely usable.

I admit that in the beginning of these LC-1 threads I was a bit skeptical and had thoughts about snake oil bouncing around between my ears but seeing and feeling is believing. That's about the long and short of it.

Thanks Roger.

Since the LC-1 approach to richening the engine fuel to air mixture does so many good things but is a bit of work to install, I wanted to see if there was an easy way to use the stock Narrowband sensor and "pull" it a little richer than it is designed to run normally. It turns out that after some measurements, and tests, it is possible to do just that--get up to 6% richer fueling using the stock O2 sensor. Here's the story ...

A couple of months ago I installed a second bung in the exhaust of my R1150RT. I wanted to have both a Wideband O2 and a Narrowband O2 installed and running at the same time. The idea was to let the Motronic use the Narrowband in the usual way while I recorded the results of Motronic/Narrowband by monitoring the Wideband. In other words using the Wideband to spy on the Narrowband. Second Bung Install.

The Wideband showed that the Narrowband/Motronic pair kept the Closed Loop fuel right at lambda=1 (AFR 14.7:1).

Then I wanted to see if there was some kind of circuit that could be inserted in series with the Narrowband to richen the mixture without having to add an LC-1. The Narrowband sensor is well designed and has a big change of voltage right at Lambda=1. Just a bit leaner and its output drops to 100mV. Just a bit richer and its output jumps to 800mV.

The exact rich output voltage increases from about 700mV at 14.6:1 to about 900mV at 13.8:1. That is a small voltage change for a large mixture change, which means it is too small a change for the Motronic to work with.

Compounding matters, in the rich zone, from 700 to 900mV, the voltage changes as the exhaust gets hotter from higher engine load. Another way to say the same thing is that the voltage that corresponds, for example, to 14.1:1 (slightly rich) changes with engine loading. Compounding the problem further, the Motronic has a clever circuit that figures out if the voltages have been shifted. It uses that circuit to ignore simple shifts of the O2 signal.

I also built a test harness which allowed me to add circuits in series with the the stock O2. I tried a dozen different ideas, including some patented circuits from nightrider.com that work on Harley Davidsons. Nothing worked.

Over the last year, I've gotten to know the owner of nightrider, Steve Mullen. One of his Harley O2 richening products has a microprocessor inside. As designed for the Harleys, it didn't work either. But a couple weeks ago, we discussed and agreed on a different algorithm. Steve coded it up and sent me a new "chip" just for BMWs.

To make a long story short, the new "chip", with some other circuit changes, will pull the Narrowband sensor several percent into the rich zone. The way it works is that the microprocessor module measures the stock O2 sensor voltage, filters it to reduce noise, and then alters the voltage transfer function so that the signal it sends to the Motronic looks like a normal Narrowband sensor that is switching at a richer lambda (range is 0-6% richer).

As it is designed (proto with many extra wires, below) you unplug the stock O2 and plug this device between the two stock connectors. There is a ground wire to connect also. There is already power in the O2 sensor cable so no power connection or new fuse is required. The final product will be about 1/2" x 1" x 3" with two OEM connectors, plus one ground wire.

I have test ridden this circuit at 13.8, 14.1, 14.3 and 14.45 and will post some charts and other data tomorrow.

For those interested, PM me. We will build a couple of modules for pre-production trials for R1150XX with Motronic MA 2.4. Assuming everything works it will then go into production.

RB

Note: If the connectors are made compatible, this would work on any R1100, R1150, R1200 (2 needed).

Prototype test cable harness and small O2 processor device. Final product would be just the processor, two thin cables with OEM connectors and a ground wire—a simple plug 'n play solution.

Below is a plot of the O2 Circuit with Narrowband at the bottom of the chart, and an LC-1 at the top. The LC-1 is set to 13.8 (the red dotted line) but it would have the same look at any programmed AFR. The thing that jumps out is how solidly the Motronic locks the LC-1 onto the target AFR.

In the bottom chart the O2 Circuit & Narrowband was set to 13.8:1. The result was 14.0:1 at hot idle (the red line) and 13.6 (the blue line) at 80MPH cruise. Different loads resulted in different Closed Loop AFRs. Interestingly higher loads lead to richer mixtures which isn't terrible—some might even say it's a good feature. (Hotter exhaust makes the O2 sensor produce a given voltage at a richer mixture.)

This means the spread was ±0.2 AFR between idle and 80 mph and that was with the narrowband pulled a long way off its design point. There are some additional data sets below the charts. (Note that the first few minutes is very rich. That's the Open Loop Cold-Start Fueling.)



After two more test rides, I have a good data base at three settings. Setting 5, 7 and 8 are actual and setting 6 is estimated but I will ride it later. The data look pretty good. The curves for the various settings all look like the one above. Here are the different results:

S8: 13.8 ±0.2 AFR
S7: 14.1 ±0.2
S6: 14.3 ±0.15 (estimated)
S5: 14.45 ±0.125

It is my opinion that most bikes will run great at setting 6 or 7. An important point here is that when you richen closed loop AFR, the open loop fueling gets richer too, coming along for the ride through a process known as Adaptation.

As I said earlier, because of the heating, the mixtures get a little richer with higher loads. Looking at those numbers above, the lean end of the range occurs at idle and the rich end of the range happens at 80 mph.

There are still a few small adjustments to make to voltage levels and to the algorithm. Then a few pre-production units.

RB

Very interesting! If I'm understanding what you've posted, it appears the LC-1 is a bit more consistent than the new prototype, but most likely it isn't enough to make a noticeable difference. Did I get that right?

I was strongly considering an LC-1, but I may hold off a bit now...

So, this means keep the same Lambda sensor and just ad a bit of wire, connectors, a cheap processor and some new code? No LC1 or other stuff? Now you're on to something.

But, what about those of us who spend a lot of time at 90 MPH in summer heat?

You've got it.

I will likely leave the LC-1 on my own motorcycle. (Since I have two bungs to install O2 sensors I'll probably leave them both connected.) But if things keep progressing as they are the O2 circuit I've described will be a very good choice.

Reasons to use an LC-1 (and there are other products that so the same thing):
—precise control of lambda from 0.9 to 1.1 (gasoline AFR 13.2 to 16.2)
—can be set to stock fueling (gasoline AFR 14.7)
—sensor calibration cycle insures long term accuracy
—powerful diagnostic plotting of AFR. You can "see" the results of combustion.
—proven, reasonably priced, reliable product from a well-known supplier
—available today

Reasons to use an "new O2 circuit":
—simple installation (remove fuel tank, plug in device, connect ground wire, reassemble)
—uses existing O2 sensor (no need to remove muffler to install)
—approximate control of lambda from 0.96 to 1.0 (gasoline AFR 13.8 to 14.7)
—can be set to stock fueling

With either solution you should use the highest setting that gives good performance (meaning don't add more fuel than needed). My estimate is that will be about lambda at 0.94 (gasoline AFR 14.1:1)

Some of the benefits of richer mixtures:
—smoother, steadier cruise
—better low-RPM roll-on power
—often can run one higher gear (e.g. 6th instead of 5th)
—cooler exhaust temperature (stock setting at gasoline AFR of 14.7:1 is hottest setting, called peak EGT by pilots)

Other things you should consider in addition to regular maintenance:
—make sure your fuel pump, filter and in-tank hoses are in perfect condition
—run a tank using the recommended dose of Techron concentrate every 6 months
—send your injectors for professional cleaning and testing. If your injectors don't have their best spray pattern and aren't well-balanced your engine won't run its best.

This is one of those changes your bike will appreciate year round but especially in the summer. As you know, the way a pilot of a piston-engine aircraft finds the correct mixture is:

—climb to altitude at full-rich mixture, then level off and stabilize cruise
—reduce fuel flow and lean-out the mixture until the exhaust temperature is HIGHEST. This is about 14.7:1 right where our motorcycles run all the time.
—then for fastest cruise, increase fuel flow until the exhaust temperature drops by 75-100 degrees F
—or, for best economy, reduce fuel flow until the exhaust temperature drops by 75-100 degrees F

Your gas mileage may drop if you run richer mixtures. For each 2 percent you add to fueling, expect a 1% drop in gas mileage (and a 20% increase in fun . But it may not drop at all if you find yourself in a higher gear due to the better running engine, that's been my personal experience at 13.8 to 14.1.

Roger and others...

I noticed something interesting on my LC-1 equipped 2001 R1150GS. I have Lambda set at .94 (13.8:1 AFR). As long as I ride below 4000 rpms the engine hovers right around 13.8 on level going and modest uphills. However, above 4000 rpm the AFR goes up to 14.4 on level going while maintaining steady speed. If I increase throttle enough for the engine to pull a little, the AFR drops back to 13.8.

I've ridden several hundred miles, so I wouldn't think the Motronic is still learning. And even if it was, I would expect steady speed on level going to be running in closed loop, which should be 13.8. I've observed the above pattern in gears 1, 2 & 3. I'll need to get on a bigger highway to test it out in the upper gears.

Could it be that the Motronic is designed to lean out the mixture when it thinks the bike is cruising (steady throttle, above 4000 rpms)? I have no way of knowing for certain when the bike is in closed vs. open loop, since the only feedback I'm getting is the AFR gauge.

Hi Wally,
I've noticed that on my bike, in neutral, Closed Loop stops just above 2000 RPM. Perhaps lighter loads exit CL earlier than I've seen.

This could be a difference between Closed and Open Loop areas for the RT and GS models. I can tell you for sure that my RT remains in Closed Loop up to something like 6000 RPM and at speeds over a hundred miles per hour. I've got lots of plots and data showing the Closed Loop status (GS-911) and AFR (from the LC-1).

Can you connect a PC and get a running log of the data. (I strap my PC to the rear seat). You run with the serial cable connected and a program you got with the LC-1 called Logworks. That will tell us a lot.

If you can, try running the same test with the Pink Coding plug instaleled (30-87-87a) we could figure out whether it is coding plug related, or whether the difference is my MA 2.4; or whether it's something else.

Hi Wally, I went out and did some quick datalogging in first, second and third gear. The Motronic does go Open Loop at small TPS angles and RPMs above about 4K--but that's what your numbers 14.4 were saying.

If there is a slight acceleration my mixture at 4500 RPMs got richer than the closed loop target (acceleration enrichment) but with even the slightest deceleration in 3rd gear the mixture got slightly leaner (deceleration enleanment).

This provides a possible model for 4000-rpm-and-above, small-TPS-angle, driveability problems. The Motronic is a) Open Loop; b) in low gears with lots of deceleration/acceleration torque, and c) in some cases not too far from triggering the abruptness of Overrun Fuel Cutoff.

The interesting thing is that with my 13.8:1 Closed Loop AFRs I can see these effects in the data log but I don't experience the effect as a driveability problem. If I cut the mixture to stock 14.7:1 I can start to feel the effect as well as see it in the log.

BTW, since it was connected (testing proto 2 now) I made these runs using the prototype narrowband-shifting device I mentioned a few posts back. So it seems to be doing just as effective a job as the LC-1 (which I still prefer for datalogging).

Roger, I was away for a few days but now back home, and as soon as the weather clears I'll get out for more runs and will log data using LogWorks3.

By "small-TPS-angle" do you mean small differences up or down from current speed? My experience was at steady throttle above 4000 rpm on level going the AFR would drop to 14.4, and I experienced just what you describe, as if the Motronic cut off the the fuel completely but then decided not to, all in a split second. Not a big "surge", but detectable. Below 4000 rpm when the AFR is at 13.8 at steady throttle I didn't notice the problem.

I'm relieved to know you observed the same pattern on your RT - it assures me I wasn't just imagining this! Any guesses about why the Motronic is programmed this way?

By small TPS angles, I mean in absolute terms. I went back and looked, and in third gear at about 4500 RPMs the throttle is only open about 7.5 degrees, Overrun Cuttoff is only 5 degrees of throttle rotation below that. The other interesting note is that the injector duration is about equal to that at idle, only 2 mS, half of which is dead time.

I also looked through other data for the 3rd gear, 4500 RPM period. It was all open loop. And the injectors pulses started dropping in duration when I dropped the throttle from 8.32 degrees to 7.68 as reported by the motronic leading to injection times dropping from 2.3 mS to 2.18 then to 1.73 mS and the AFR going from slightly rich to slightly lean.

I had a chance to get out yesterday afternoon for a brief ride on the Interstate where I could get the bike into the upper gears. In 4th, 5th and 6th above 4000 rpm the AFR never got above 13.8, unlike the lower gears.

Does anyone have an idea why the Motronic apparently goes open loop (AFR 14.4) above 4000 rpm in the lower gears, but not in the upper gears? AFR stays right around 13.8 at steady speed on level going all of the time except above 4000 rpm in 1st, 2nd & 3rd!

I did observe a couple of times at higher rpm in the lower gears yesterday when the AFR dropped down to 13.8, but didn't get enough riding in to be able to replicate the conditons.

At 4000 rpm, small throttle angles, the 1150 appears to be open loop. In the higher gears because you're going faster, the engine has to produce more power and your throttle is more open. At higher throttle angles the Motronic applies closed loop control.

I've spoken to the guys at PC and asked them about the chart below which shows where they believe the Motronic is closed and open loop. According to them, it is their understanding of it but they can't say for certain that it is accurate. The dark area is closed loop and the white area is open loop.

Why BMW has designated the various areas open and closed is harder to work out but it is either because closed loop is unreliable at those RPMs and throttle angles, or because they always want either leaner or richer operating conditions that closed loop would allow.

Back when I began this project it was partly because when shifting up in the low gears, I experienced roughness when the engine was warmed up, right at the "top" of the shift. It now appears that this is an area of leaner open loop

As I mentioned earlier in the thread, I've been experimenting with some options for shifting lambda. A fully programmable, accurate solution, with the ability to datalog, is a Wideband O2 Sensor with a Wideband Controller like the Innovate Motorsports LC-1. (As a note, there are other manufacturers of this kind of product: Zeitronix, PLX, & wbO2. Also, I'm still working on a device that can pull a Narrowband O2 Sensor several percent to the rich side (this is looking pretty good).

The other day, I had someone build me a Wideband O2 Sensor (Bosch LSU 4.2), with a fixed-AFR shift Wideband Controller built into the cable (photo below, small controller not shown). At the moment it's not quite compatible with the Motronic MA 2.4 and it ends up running 9% richer than stock when it is set to 6% richer--still looking into it.

What was interesting was at 13.4:1-ish my bike's hot idle had increased to 1400 RPM from about 1200. I also noticed on a local trip to the Post Office I was in a higher gear (6th) than usual (5th). Like everyone, I shift by feel and was quite surprised that I was going 50 MPH at 2500 RPM.

While it's probably too rich (?), it seems the 1150RT likes its fuel.

RB

Roger, I've been following your work from the start and have to give you a huge attaboy and THANK YOU for making all this information available to us!

I have the Innovate LM-2 and I've used that to do similar work to yours, but I worked on the Xchallenge and not on one of these boxer bikes.
The LM-2 also has the narrow band sensor emulator output.
While playing with it replacing the stock NB sensor with the LM-2, my observations were very similar to yours regarding the ECU operating modes and effects of richer CL mode target AFR.

I've always wanted to replicate this work on my 1200GS but as you know it requires two sensors -> lot of money using the Innovate components.

Now I see you are working on the ultimate solution.
Forget the IAT tampering (people pay ridiculous money for those contraptions), forget overly complex and expensive PC.
Your effort to build a WB sensor based controller that gives a fixed shift in CL AFR is the way to go!
Let me give you all the encouragement to stay on it.

If you ever decide to make such device available in any form (anything from a bolt-on kit to just a set of documentation + required bin file) I would be extremely interested.

I've made my own IAT shift devices, used and programmed PowerCommander but what I've come to believe is the best way to go is what you are working on now.

Mikko