EDIT: THIS EXPLANATION IS INCORRECT. PLEASE LOOK FURTHER INTO THE THREAD FOR A CORRECTED EXPLANATION. The error was due to an omission in the PC III Installation document that has since been discovered. Further, I have been in touch with Dynojet who have provided good detailed answers to some questions on the connection of their Wideband sensor to the Motronic that is not in any of their documentation.
Another way to add fueling to an 1150 (or 1100) is by using a PowerCommander III, block diagram below. This product includes a Wideband O2 sensor and an Add-On Fuel Table computer. It is a standard product, that can easily be added to an 1100 or 1150. Many believe that it is necessary to "Dyno Tune" the Powercommander after installation, but you could richen the mixture without having to go to that expense.
One reasonable way to use the PC III would be to install it along with its Wideband O2 sensor, use the included software to program the O2 sensor Lambda to 13.8, and fill in the fuel table with 6% everywhere, perhaps tapering off toward the high RPMs and high TPS angles. In this mode it would be similar to the LC-1 and somewhat easier to install.
With the O2 sensor programmed to 13.8, the Powercommander claims to operate its own proprietary Closed Loop program (see the shaded Closed Loop area in the second chart). That said, major parts of the Motronic are disabled when the PC III is installed, particularly the Adaptive ability. Powercommander makes no claim of Closed Loop adaptation within its module. Also, the Limp function of the Motronic is enabled when the O2 sensor input is disconnected. This means the pulse stream going into the PC III will have the 10% variation I wrote about a couple posts back.
The PC III looks pretty straight-forward to install but you do have to disconnect the O2, add two connectors to each injector, and install a piggy-back onto the TPS connector. So I would say it's fairly invasive to the Motronic system--if that matters to you. The PC gets its throttle position information from the TPS piggy-back but doesn't seem to have a TPS learn function like the Motronic. The PC III gets its RPM information from the injectors by measuring the frequency of fueling pulses. That works well but during Overrun Fuel Cutoff, the PC III doesn't have an RPM input. I think that might be a nit, but I mention it.
At its list price of $495.95, it's about three times the cost of an LC-1 but it is a Plug and Play solution. In terms of software, it's not clear to me whether the PC III has the logging capability of the LC-1 which I see as an important diagnostic tool. All the AFR charts of this thread have been made using that capability. It does come with a good suite of software for populating and managing the fuel tables should you wish to adjust them.
A further note on using the PC III for richening:
I mentioned that a simple way to use the PC III would be to program the Wideband O2 that comes with it to (say) 13.8:1 and fill in the fuel table with 6% (maybe tapering to a lower number at the high RPMs and high TPSs). However, 6% is only the right number if you're running pure gasoline.
Since I haven't seen any documentation on the PC III's Closed Loop capability and haven't measured it; and since the Motronic's Closed Loop ability (and therefore its Adaptation capability) is disabled by a PC III; I believe that the fuel table should have 10% added for motorcycles that run E10 fuel. That would be 4% for E10 and (say) 6% for the richer Lambda setting on the Wideband O2.
Another method for richening the mixture is through the use of a Dobeck Engineering product. I'm aware of a couple: one disables Closed Loop (Techlusion) and the other allows Closed Loop in part of the RPM range (Electronic Jet Kit). Both products use the same technical approach of monitoring the injector electrical pulse but not intercepting it. By measuring the time between injector pulses and their inection-on-time, the Dobeck technology can cleverly approximate the Load and RPM of the motorcycle. Neither product employs a Wideband O2 sensor like the LC-1 or PC III (although a gen 4 version does use a Wideband O2, I don't see it listed for the 1100s or 1150s).
The rider can then add fuel (lengthen the pulses) using three/four rotary dials (or plus/minus buttons) that allow different amounts of fuel to be added in cruise, acceleration or WOT. For me, the challenge of this approach is you can't specify how much fuel to add with a concrete setting (e.g. you can't specify 4% more than Closed Loop), and their is no Closed Loop monitoring of the result of the adjustment. However, with skill, it can be made to work--provided that Closed Loop is disabled. With Closed Loop disabled, the fueling additions are made to the Limp Home pulse train with its 10% fueling variation that I showed a few posts back.
Looking at the block diagram below, the Narrowband sensor gets connected to the Techlusion but there is no indication in the documentation that suggests the O2 sensor signal is used to calculate the fueling addition:
I've been doing some further research on the Powercommander III for Oilheads and now have a much different view of how the PC and Motronic work together. My prior view that the PC didn't drive the Motronic's O2 sensor input was wrong and was due to a lack of documentation. But that said, here is an update for those interested:
I've heard from Dynoject and although I'd like some further clarifications, I don't think I'll hear more. The product was designed nearly 10 years ago and as you would expect the designers have moved on.
[INDENT]" the testing was done so long ago I may not be able to answer your question thoroughly."[/INDENT]
The BMW Powercommander III USB Wideband is essentially a Wideband Commander coupled to a Powercommander III USB.
The essence of what I've heard fro Dynojet is:
[INDENT]"The connection to the stock ECU narrow band input is tied to our Wideband sensor. We are able to offset the narrowband signal based on our wideband input."[/INDENT]
This is how the Innovate LC-1 works, although the LC-1 is a later, better performing design with data logging capability.
The BMW Powercommander and Motronic runs closed loop and develops Adaptation values everywhere the Motronic would on its own. That's because the Wideband signal, converted to narrowband format, is "Always connected". The shaded area on the PC fuel map (see photos several posts back) is just a guess on PC's part about where the Motronic is in closed loop.
[INDENT]"The highlighted area is what we have interpreted as the closed loop area of the stock ECU."[/INDENT]
I have many measurements that show the closed loop area of the Motronic is up to 62.5% throttle and up to 6250 RPM. I think we should expect that it is closed loop everywhere below those numbers.
I did some digging and found letters from Dynoject to Harley forums where there was a clamor for this capability. Here is part of what the design manager wrote:
[INDENT]"Let us start with why the BMW uses a wide band O2 sensor as part of the unit. The bike already has a "closed loop" circuit as part of the OEM injection system. It does not "auto map" the entire rpm/throttle position range of the fuel map. Generally speaking, the closed loop system only adjusts the fuel curve below 40% throttle. Above that the system is "open loop". The new Wide Band BMW unit only controls the stock "closed loop" area. Outside of that the bike is mapped in the normal fashion, on the dyno.
We would actually prefer not to maintain the closed loop section. Due to the design of the OEM injection system it is not possible to bypass it as we do with other models. Closed Loop systems are not the "magic" that most people believe they are. There are a number of problems that keep it from being the best choice for high performance applications."[/INDENT]
This all means that the PC III for USB works very differently on an Oilhead than on any other motorcycle most Dyno tuners work on. It also means that the WOT "pulls" aren't likely to provide an optimal tuning since most dyno tuners don't seem to understand the interaction with Closed Loop Adaptation Values. There's no reason that they should be familiar since this product works differently than most every other PC they would work with.
With this information, it is now pretty clear to me how the Powercommander and Motronic work together. My plan is to update the block and show a recommended fuel map for implementation. I am a lot more positive on the Powercommander as a tool now than before. It can be implemented on the majority of BMW Oilheads with NO Dyno tuning.
If anyone reading this would like to loan me a PC for a couple weeks I'd like to run some tests. In the meantime I'm going to try and buy one used. Then later I'll resell it, all set up for installation.
[I]Earlier in this thread I presented an incorrect explanation of how the Power Commander III USB with Wideband O2 operated when integrated with the Motronic MA 2.4. The reason for this error was an omission in the Power Commander installation manual for R1150RT. Here is the correct explanation, please refer to the diagrams at the end of the thread.[/I]
The Power Commmander III with Wideband O2 for BMW R1150 and R1100 is actually two different functions in one package. One function is that of the Power Commander III USB an Add-On Fuel matrix of values according to RPM and TPS. The second function is much like that of the Innovate Motorsports LC-1. Rather than recapitulate the Add-On Fuel function you can read about it here: [URL="http://www.powercommander.com/powercommander/Products/PowerCommander/powercommander_iii_usb.aspx"]Power Commander III USB[/URL].
The second function, Wideband O2, functions just the same as an LC-1 Wideband O2 controller. You can read about that in detail earlier in the thread. In short, you replace the stock Narrowband O2 sensor that has a fixed transition point at 14.7:1 AFR (actually Lambda = 1, meaning that theoretically all the oxygen has been consumed) for a Bosch LSU 4.2 Wideband O2 sensor which has a programmable transition point. By selecting a Lambda value less than 1, the Motronic will automatically richen the mixture in the Closed and in the Open loop fueling calculation. I'll rephrase that: if you drop Lambda to 0.94 the Motronic will run Closed Loop at 13.8:1 which is 6% richer than stock 14.7:1. Then through the magic of the Motronic's Adaptation Values capability it will also add fuel to all Open Loop fueling calculations after a "learning" period.
The "learning" period takes some time and is not quite perfect because you have to drive your motorcycle at steady throttle for enough time in Closed Loop at a range of throttle openings and RPMs for the Motronic to "learn" the full adaptation map. It is likely that there are fewer Adaptation Values than cells in the Base Fueling Table of the Motronic, which means that there is likely a coarser correction. But I've measured it, and Adaptation works effectively, it just takes a while. The bigger the shift in Lambda that you're making, the longer it seems to take to "Adapt".
There is a way around the "learning" time. In short, find a way to either a) add a percentage of time to the pulse coming from the Motronic; or b) increase the fuel pressure so that the pulses coming from the Motronic deliver a percentage more fuel than stock. Since the Innovate Motorsports does not have an Add-On Fuel map I added a Fuel Pressure regulator and boosted the fuel pressure. However, if you use a PC III USB, you can add the fuel through the fuel table function.
Looking at the sample Add-On fuel tables below, if you fill in the fuel matrix with a 6% addition in every cell at 60% throttle and below, and then reduce to 4% more fuel at 80% throttle and 2% fuel at 100% throttle you will have a sound starting point. Note that this approach acknowledges the great work that BMW has done in their design of the Base Fuel Map in the Motronic; it just adds a proportional percentage to account for the amount of richness you want to add. I'm sure this seems like a simplistic use of the Add-On Fuel function of the PC III but it will work and you won't need to add a BoosterPlug or Fuel Pressure regulator.
I'll end with a short comparison of LC-1 and PC III. Both are technically sound methods for controlled richening of the Oilhead's mixture, leading to much better driveability and a bit more power in the mid-band (2000-5000 RPM) which it achieves by moving up the AFR vs Power Curve earlier in the thread. (There will not be an increase in WOT horsepower because the BMW fuel tables are already near Best Power Mixture there.)
--Lower cost: $170 for the LC-1, $395 for PC III (although I've seen PC III for $285)
--Calibration function for the Wideband O2 sensor
--Datalogging of the realtime stream of O2 readings
--AFR gauge included in purchase price. It is an add-on for the PC III.
Advantages PC III:
--Plug & Play: The PC III has all the connectors you need to plug it in out of the box. For the LC-1 you need to wire your stock O2 sensor connector and also power leads to its cable.
--Built in Add-On fuel capability. If you use the LC-1, you either wait for Adaptation, add a BoosterPlug which shifts fueling a fixed 6% or, as I did, increase the fuel pressure. A good fuel pressure regulator costs around $100.
I've been doing some more research on the Power Commander so I'd like to update the Notes section but can't go back to edit it. Here are the notes with updates, 5) and 6):
1) I am trying to get a PC III w/Wideband but have not run one yet.
2) The fuel table on the PC III allows you to enter percent increase/decrease. Typical injectors have a 1 mS dead (on/off time), I don't know if the dead time is taken into account by the PC III.
3) At 7250 RPM the injectors fire every 8.3 mS. The longest injection pulse that I've seen is 8.2 mS. This points out that you can't add 5% to the longest pulses with making them longer than the frequency of rotation in some cases. (However, the vast majority of the time injection pulses are less than 4 mS.)
4) I don't think there's a problem but I can't tell how the PC III responds to Overrun Fuel Cutoff when no pulses occur (i.e. the PC III isn't getting any engine speed info during that time.)
5) Because the Powercommander gets its +12V from the injectors/fuel pump and since the fuel pump goes off after a couple seconds, Dynojet recommends hitting the starter button before the fuel pump cycles off.
6) The Powercommander DOES take injector dead-time into account when adding/subtracting fuel.
I know you went with an adjustable fuel pressure regulator on your bike Roger. To allow flexibility in testing I assume. One could always go with the K bike 3.5 (50.7 PSI) bar regulator which is around $85 new so probably half that used. I like this method simply because the part fits right into the 1100 or 1150 without any plumbing changes needed and is known to work just fine. You do have to lift the tail to get in there though!
That's right, I used a variable FPR so I could try different Lambdas and different pressures. And you're right, if you can get access to the FPR putting in a 3.5 bar regulator gets you close enough for up to 8-10% enrichment as long as you don't mind waiting for the Motronic to adapt itself to your target Lambda. 4-6% enrichment would be a piece of cake. As you said, no extra plumbing, just a swap.
An update and some new info:
Lately I've moved my Lambda setting from 0.94 to 0.92 which is another couple percent richer. I did this because I noticed that just as the bike was finishing warm-up (while riding) I liked the cruise performance just a bit better than once it went Closed Loop. The biggest difference between a minute before to after Closed Loop is a few percent more fuel. I'm not sure just where I'll stop adding fuel.
There are a couple IC engine performance curves I've found that you might find interesting. Although NOT specifically for an Oilhead, interesting nonetheless.The first is HP vs AFR at various power levels:
The second is flame front speed as a function of mixture (note that it is Fuel Air ratio not AFR). This shows pretty clearly that as you richen the mixture, you speed up combustion. It looks to me to be the equivalent of 2-3 degrees more advance for an AFR of 13.8 compared with 14.7, at 3000 RPM. At 6000 rpm it might be 4-5 degrees if I'm doing the math right. Because of this effect I have been paying close attention to any hint of knock. There doesn't seem to be any probably because even though the timing advances some, knock resistance of richer mixtures is better.
So the bottom line is that when you richen the mixture several percent you're probably getting a triple benefit: a bit more torque from the extra fuel, a bit more torque from the more advanced (relatively) timing, and better efficiency due to better running and therefore into higher gears several hundred RPM sooner than at 14.7:1.
This is some great stuff Roger. Thanks!
A veritable RPM trifecta! :laugh
As I mentioned earlier in the thread, I realized several months ago that a GS-911 can be used as a personal Dynomometer. The key things you have to know are the vehicle's weight including rider (~800 lbs.) and aerodynamics (Cd=0.66, Frontal Area 8 sq-ft), gear ratios, and rear tire diameter (25.59" for PR2 in motion). You log the GS-911 to a data file, import it into Excel, run some math on the numbers and you have Torque, Horsepower and Acceleration. Working out how many feet/second of velocity are created per RPM in a gear is tedious but not difficult (0.01956667 for 4th gear).
It has been my belief based on the feel of my 2004 R1150RT that it has gained torque in the lower RPMs from running Closed Loop at 13.5:1 and fuel pressure at 52 psi. Spend some time looking at the charts below. The Excel chart is a scatter plot of 6 test runs that I took in 4th gear, no wind, 3 in one direction on the highway, 3 in the opposite. Some runs were better than others but I used them all. There are significant torque gains below 3500 RPM, especially between 2000 and 2500--this tracks my driving experience. My RT produces about 55 lb-ft at 2500 RPM compared to 45 lb-ft on Ron's sample. Note: The numbers I measured are actually 5% higher but, as with a Dyno, I reduced them by a factor of 0.956 to account for weather conditions.
Between 3000 RPM and 5000, my bike accelerates at about 12 ft/s??. Anything over 10 ft/s?? gives you reasonable acceleration. In fourth gear that requires 48 lb-ft of engine torque, the curves on my RT show that at 2000 RPM. There are also gains at higher RPMs.
While I believe my data is accurate, the one thing I would say is that I was careful to get the run prepped for low RPMs by running along at idle in 4th or 5th gear, then crank open the throttle. I did not run quite to the Rev Limiter, choosing to let go of the throttle at around 7000 RPM. So my numbers might be a bit better due to the care of low RPM starting but I don't think that explains it all.
Enjoy the curve, I've got some more coming.
Stock bike from Ron Hankison's web site.
In the interest of documenting what the Motronic does under various conditions, I've added a plot of fuel injection time and spark advance logged during the WOT runs I made to produce the torque curve posted above.
(Below the advance/injection plot I've added a summary of the GS-911 data log for anyone interested in what's being collected. A typical test run has several thousand lines of data that gets imported to an excel spreadsheet. The first two columns of time and RPM are used to calculate acceleration, and from that torque and HP.)
The conditions for the WOT runs were: shift the bike into fourth gear; running along at 1500 RPM; and then open the throttle fully until about 7000 RPM.
Looking at the plots below a couple things stand out. On the injection time plot you can see the Motronic fire off 7 mS shots of fuel right away to get the acceleration goingÔÇövery rich, and keep in mind that each injector fires twice per combustion cycle. Then from 2200 to 4000 RPM, still WOT, the pulses drop to about 5.8 mSÔÇöstill rich but a little less. From 4000 RPM to 5500 the injection time lengthens to 7.5 mS which is nearly on continuously at 7000 RPM. Between 5500 and 7000 RPM the engine is putting out 70 - 94 HP so the Motronic is keeping the mixture very rich. AFR on my bike is in the low to mid 12s for these WOT runs; by comparison the stock setting is in the 13s.
Looking at the Spark Advance plot, under the Wide Open Throttle conditions of this series of measurements, the Motronic is conservative with Spark Advance, holding it at 20 degrees or less until 5500 RPM. By comparison, under partial throttle cruising conditions the Spark Advance gets as high as 43 degrees.
A couple posts ago I showed the GS-911 data which is the source for the Dyno calculations that I've made. If you look at the first column there is a six digit number in milliseconds, and in the second column the RPM reported by the Motronic. By taking any two adjacent RPM differences and dividing it by the difference between the times for those two RPM values, you get a direct measure of acceleration. The formula for calculating the acceleration in feet per second-squared is:
[b]Acceleration = (RPM2-RPM1)/(Time2-Time1)*19.6852[/b] ... ( 19.6852 is equal to 1000 times the 4th gear drivetrain ratio including transmission, final drive and tire diameter)
In the chart below you can see the actual acceleration I measured for my R1150RT (the Blue Wideband line), five runs in fourth gear. These were 5 Wide Open Throttle runs after the Motronic had adapted itself fully to the Lambda = 0.92 setting (gasoline AFR 13.5:1). This acceleration data was the basis for the Torque curve I showed a few posts back. 12 feet per second-squared means the motorcycle accelerates by 8 MPH each second.
For the stock bike (Red line) I used the torque curves I showed earlier, and boosted them about 5% for the temperature and pressure conditions of the day; then calculated the acceleration that torque curve would produce in fourth gear. If you ignore the red (stock) line altogether, the Wideband acceleration is very good from 2000 RPM on up. And at 2400 RPM (11 fps2) the acceleration is nearly as good as at 5000 RPM (11.8 fps2). The boxer doesn't have to be a bike that only performs well at high RPMs.
Comparing the Red and Blue lines, there is a significant rate of acceleration advantage in favor of the Wideband curve between 2000 and 3000 RPM, which is what I experience while riding.
There's a saying, "There are three great lies ... lies, damned lies and statistics". That said, I'm pretty certain with all the measurements made and data collected, that the Oilheads run better with a bit more fuel. And given the number of comments you can find about bikes with Catalytic Converters running lean and the negative effects of that, it shouldn't be a surprise that a few percent more fuel improves them measureably.
I've added a second O2 bung so that I can use the LC-1 Wideband sensor at the same time that the narrowband stock sensor is installed and feeding the Motronic.
The exhaust is back on with the Wideband sensor in the position near the catalytic converter, and the Narrowband sensor mounted in the new bung near the clamp. By connecting the Narrowband sensor to the Motronic and using the LC-1 as a recorder, I'm already getting data on the Narrowband/Motronic combo.
The first things that jump out at me:
ÔÇöThe stock Narrowband sensor is much slower than the Wideband, no question. That means the Motronic takes a lot longer to make corrections.
ÔÇöThe AFR spread is much larger with the Narrowband sensor. The larger spread plus the slower response shows fueling patterns that take [U]seconds[/U] to change direction. Even in the quick data I've taken I've seen the mixture take 4 seconds to go from 15.4 to 14.2 while in closed loop. This is a long time and a fairly large change of mixture. The mixture change is large enough and slow enough that I'm sure it could be felt as su----g.
I'm going to do some riding with it tomorrow if the rain hold off and then post some plots and photos.
This is really interesting.
- Do you think the slow response time is just due to the O2 sensor probe itself or is the dampening purposeful? By this I mean they have purposely slowed down the response using software delays (used to call them no ops) in the AFR algorithm or is the hardware itself just really slow.
- The surging I've experienced on both my '96 and '00 RTs was not really a slow response but rather a hunting for a correct setting that is never really found type of action. Not sure this correlates to what you are seeing now but I don't know enough about the control loop to really judge this.
[QUOTE=Happy Wanderer;849454]This is really interesting.
- Do you think the slow response time is just due to the O2 sensor probe itself or is the dampening purposeful? By this I mean they have purposely slowed down the response using software delays (used to call them no ops) in the AFR algorithm or is the hardware itself just really slow.
- The surging I've experienced on both my '96 and '00 RTs was not really a slow response but rather a hunting for a correct setting that is never really found type of action. Not sure this correlates to what you are seeing now but I don't know enough about the control loop to really judge this.[/QUOTE]
I've only had the motor running in the garage for 15 minutes so I don't want to get ahead of myself. The slower response is comparing closed loop fueling plots. When the LC-1 is used as the input to the Motronic the rich/lean loop is quicker and the mixture spread is smaller than when the Narrowband O2 is feeding the Motronic. So I would say it is due to the characteristics of the Narrowband sensor mainly.
It may be that the slower sensor allows more lean operation time which is recharging the catalytic converter's oxygen storage elements. It may be very good for the cat but not so good for the rider.
This speed and mixture spread is probably why my bike runs better at 14.7 with the Wideband than is does with the narrowband 14.7 sensor.
If I can, I'd like to capture some "surging" by running the stock O2 and recording it with the LC-1.
I've seen the average AFR get to 15.4 during closed loop today in my short test. By average I mean the combo of both clinders. If the injectors were mismatched by 4% the leaner cylinder would be 2% leaner than the average and its peak AFR would be 15.7:1ÔÇövery leanÔÇöthese numbers seem likely to be very representative of most bikes.
I think the "hunting" you mentioned is what I'm seeing as a seconds long change of mixture, not sure, more work to be done,
I picked up my exhaust Friday afternoon after a local welder with TIG capability cut a hole an inch and a half in front of the stock bung, and then welded in a stainless steel bung.
Next everything was put back together with the Wideband O2 in the usual bung and a stock Narrowband Bosch O2 in the new location. The Narrowband O2 got connected to the Motronic. In order to return the bike to stock operation the fuel pressure boosting regulator was removed from the fuel return line. While the Wideband O2 was removed from the exhaust and in free-air, the LC-1 was recalibratedÔÇöa nice feature it has.
Everything worked and the LC-1 and GS-911 started recording data on the stock setup. The LC-1 is a great tool for analyzing the stock sensor, coupled with the GS-911 you can clearly see what the engine is doing with mixture. Several test rides confirmed what I reported earlier, that the high/low AFR range during closed loop is between the low 14s:1 into fairly lean territory in the low 15s:1. Any injector mismatch would widen the range. The average AFR in Closed Loop was 14.7:1 as expected.
My test rides yesterday did remind me how differently the R1150 runs on the Narrowband sensor. When fully warmed up, with a light load in 1st, 2nd or 3rd gear in the 3000 to 3500 RPM range the motor feels on the edge of stumbling. On a slight downward grade in those gears/RPMs I could feel a light surging. I also found that I was shifting at higher RPMs than with the Wideband connected to the Motronic. Later when I pulled into the garage after the rides I could "smell" a hot-exhaust odor.
The richer mixtures (13.8:1) I had been running with the Wideand O2 connected for most of the past year improved driveability, softened the response to throttle and added muscle between 2000 and 3500 RPM. The richer mixtures were also reducing exhaust temperature. When I have time I will try and measure the temperature difference Wideband vs the stock O2.
Over the next few days I plan to see if the Narrowband sensor can be nudged to several tenths richer AFRÔÇöwith a first target of 14.2:1. I'll report how that goes.