Fuel Injection Open Loop and Closed Loop
Written Aug ’06. Probably best to read Fuel Injection A Brief Piece On How It Actually Works first. And, if you don’t find what you want below, Google is your friend. That or the legendary Bosch Automotive Handbook, the densest, most tightly packaged collection of a staggering amount of information, specifications and data you’re ever likely to come across. If you’re an engineer or the like and haven’t seen it I suggest you have a look. Not cheap, but even a quick flick through will overwhelm you so comprehensively you just know it is money well spent. Or, in my case, gratitude to the two Bosch people who have supplied me with editions 4 and 6 over the years for free Werner (with a 916 then R1150GS) here in Australia and Torbjorn (with a 900M) from the diesel side of things in Sweden.
Plus the Tech Edge site http://wbo2.com/default.htm , like the Bosch book, has lots of specific sensor and other controller and logger info that is well over my head.
Much of the practical information comes from my BMW training, where they tell you more than anyone else does, and experience. BMW have been running all new model bikes with Closed Loop systems in Australia since the R1200C and K1200RS were launched in ’97, and since ’93 I believe in the USA with all 4V boxers R1100RS, GS, R, RT and 4 cylinder K1100 RS and LT.
Also keep in mind the closed loop thing is not a conspiracy or attempt to piss you off personally. It’s a very sound system with a very clear aim reduction of harmful exhaust emissions. It’s not going to go away. But, as with most things, you don’t need to understand it completely. Feel free to jump down toward the bottom where I get on to the practical implications for tuning if you don’t want the full discussion and usual amount of crapping on.
As ever, this is all to the best of my knowledge, and writing this report really made my brain hurt. If anyone notices any mistakes, please let me know.
The fuel injection systems fitted to all the bikes we now sell are, by definition, digital control systems. They perform a set function based on rules and information stored in the ECU, with all the freedom digital control systems allow. Just like any other digital control system, they can be ‘open loop’ or ‘closed loop’. These two terms refer to the individual system’s ability to measure the result of what it is making happen (feedback) and modify what it is doing based on that feedback.
Open loop simply means there is no feedback of the result to the ECU. In our case, it means there is no sensing or measuring of the exhaust gas to see how the bike is running. The fuel injected is determined by the RPM and throttle position, derived from fuel injector pulse width numbers stored in the fuel maps, and is trimmed for environmental conditions due to air temperature, air pressure and engine temperature.
Closed loop means there is feedback of the result to the ECU. In our case, it means there is sensing or measuring of the exhaust gas to see how the bike is running. This sensing is done by a probe (a galvanic cell) which generates a voltage based on the gas around it. These probes are referred to as Oxygen sensors, Lambda sensors, O2 sensors, Exhaust Gas sensors and probably a few other names as well. I’ll call it a Lambda sensor. Why Lambda you ask?
Lambda is a term used to describe the composition of the exhaust gas of combustion. Also called the Excess Air Factor, it is a ratio of the difference between the actual air/fuel ratio and the chemically correct air/fuel ratio. The chemically correct air/fuel ratio is also known as the stoichiometric air/fuel ratio, and is given by the chemical equation for combustion of a fuel in air. If combustion takes place at the chemically correct stoichiometric ratio, the air/fuel ratio for petrol is 14.7:1 and the Lambda number is 1.0. The 14.7:1 bit means that for chemically correct combustion you need 14.7 times as much air (by weight) as you do fuel (again, by weight). By volume it’s around 9,500 litres of air per litre of petrol.
If there is excess fuel (rich) then the Lambda number is less than 1.0. If there is excess oxygen (lean) then the Lambda number is greater than 1.0. So the Lambda number is the actual air/fuel ratio divided by the chemically correct air/fuel ratio.
You’ll notice I’ve avoided using the terms ‘ideal’ or ‘optimum’ or something like that. This is because 14.7:1 is not ideal, etc. It’s just a number defining a chemically correct process in an ideal situation. In reality, the composition of the petrol is not just one hydrocarbon, it’s a combination of quite a few and the chemically correct ratio varies with that. Plus, air is mostly nitrogen, and this nitrogen comes along for the ride and gets mixed up in the products. And combustion is very rarely ideal anyway.
The products of the ideal combustion are CO2 (carbon dioxide) and H2O (water). In reality, the products are CO2, H2O, CO (carbon monoxide), HC (hydrocarbons, unburnt fuel), O2 (oxygen, unused) and NOx (Oxides of Nitrogen). Lots of nasty stuff, which we’ll talk more about later.
Lambda and air/fuel ratio numbers are the same thing expressed differently for petrol, Lambda 0.9 = 14.7 x 0.9 = 13.2.
Now, back to the Lambda sensor itself. These sensors come in two main types Narrow Band and Wide Band.
A Narrow Band Lambda Sensor generates a high voltage if it is in an atmosphere lacking oxygen. If the atmosphere has excess oxygen, a low voltage is generated. In use, a rich mixture gives a voltage of 0.8 1.0 volts. A lean mixture gives a voltage of 0.2 0.0 volts. The graph below taken from the Tech Edge web site shows the output more clearly. As you can see, there is very little variation on either side of rich or lean, but as it goes from rich to lean the voltage output changes from 0.8 volts to 0.2 volts almost instantly. Note the Lambda range at the bottom of the graph is 0.98 to 1.02 the transition is very quick.
This means, in practical terms, that the Narrow Band Lambda Sensor is a yes/no type indicator. Yes the exhaust gas shows a rich mixture/no it doesn’t. Simple as that. It has no real use as an indication of mixture apart from rich or lean.
Narrow Band sensors have been around for some time and come in two types - heated and non heated. They can also have one to four wires. Initially, they were one wire the output voltage, using the vehicle components as the ground path. Then they went to two wires for the signal with a dedicated ground wire, which gave more consistency and less noise to the output. When the first heated probes appeared they were made as 3 wire, being one signal, one heater, one ground, or 4 wire one signal, signal earth, heater, and heater earth. These days 4 wire sensors are pretty much the standard.
The heater is to warm the sensor up. The sensors don’t work very well until they are hot, and with the emissions laws increasingly reducing the time after start up that a vehicle must start complying, the heater is needed to get the sensor working quickly.
The Wide Band Lambda Sensor differs from the Narrow Band Lambda Sensor quite markedly in its construction and operation. It works by using a supplied voltage to create a chemical reaction that then creates a current based on the mixture of the gas. I don’t understand any more than that, but then I don’t need to. The output of these sensors is non linear, but varies over a very wide range of lambda numbers, from very rich to very lean. As such, they are very useful for tuning. The graph below taken from the Tech Edge web site shows a wide band sensor output. The Lambda number is across the bottom, output in mA, not Volts, up the LH side.
The Wide Band Lambda Sensor comes in two types that I know of. Firstly the older style ‘Bosch Motorsport’ (BMS) sensor with four wires that Duane Mitchell has used for years with his Gold Motec and the Corsetec Lambda meters. Secondly the later ‘Universal Exhaust Gas Oxygen’ (UEGO) sensor with 5 wires that is now coming into wide use with the cheaper Lambda meter/data logger boxes like the LM-1, Dynojet Wideband and local Tech Edge to name a few. These UEGO sensors are also being fitted OEM to many cars in place of the narrow band sensors. How much use is made of their potential I don’t know.
According to one of the engineers at Motec, the UEGO style sensor is much smarter than the older BMS in that it can tell you when it’s going bad or is too hot, etc. The too hot bit is very relevant, as the BMS can handle much more heat we often mount them 2 or 3 inches from the exhaust port depending on how much room we have. Whereas I’ve read that the UEGO sensor needs to be around 450mm (18”) from the exhaust port to avoid overheating. While this might be fine in car applications where that’s in the collector anyway, on a bike we usually have headers that aren’t 18” long, so running a probe per header for main and offset fuel mapping with these sensors probably isn’t possible unless you cool the sensor somewhat or have a lot of time to stop for a rest when it gets too hot. As yet I haven’t used one, but I have a Tech Edge unit Duane has sent me that I’ll hopefully get to use on the 750M when I’m tuning the carbs. Duane did have this UEGO probe in a bike header next to a BMS probe to compare their outputs in the same situation, but it fairly quickly got too hot and stopped playing.
Like the Narrow Band sensors the Wide Band sensors are also heated.
Why Closed Loop?
The main reason for running engines closed loop is for the efficient operation of 3 way Catalytic Convertors. The main reason for running Catalytic Convertors is to chemically change the constituents of the exhaust gas from nasty to not so nasty (because it’s never going to be nice). Catalytic Convertors are very natty pieces of gear, and are responsible for allowing performance to return to vehicles in this age of vehicle emissions control.
Originally, Catalytic Convertors appeared as ‘2 way’, meaning they cleaned up two gases only, CO and HC. While this was good, it left NOx (the various Oxides of Nitrogen) untouched. NOx combined with HC makes up smog, and on their own are responsible for acid rain. The problem with this is that NOx is formed when combustion goes over a certain temperature, so to reduce NOx the temperature of combustion needed to be reduced. But temperature means pressure and lowering the pressure of combustion is fundamentally opposite to good power and efficiency. This is why, in the 70’s and early ‘80s, car engines had lower compression ratios, wacky cams with longer overlap to give natural EGR (exhaust gas recirculation) and generally crappy performance.
When the ‘3 way’ Catalytic Convertor came along and dealt with NOx everyone was well happy again, as this meant combustion temperature and pressure could once again go up. This is why late ‘80s car engines (especially large ones) tend to go a hell of a lot better than those from the early ‘80s. In real terms, 3 way Catalytic Convertors are probably the most important internal combustion engine development alongside digital electronic engine management, especially in terms of emissions control. They can remove around 98% of all harmful exhaust constituents (although they turn any carbon based gas in to CO2) and are the technology that gives rise to things like the SAAB ad campaign of the ‘90s where they claimed that, in a London traffic jam, the gas coming out of their engines was cleaner than the air going into them. In Mother Nature’s battle against the insidious, cancerous virus called the Human Race the 3 way Catalytic Convertor is one for her.
But, for a 3 way Catalytic Convertor to work well, it needs to reach a certain operating temperature. The richer mixtures of warm up help here. Then, once it is hot, it likes the exhaust gas composition to cycle from excess oxygen to lack of oxygen. Just a little variation either side of stoichiometric. Luckily, the output of a Narrow Band Lambda sensor is just what is needed.
The Lambda sensor senses the rich or lean state of the exhaust gas and tells the ECU either rich or lean. The ECU reacts to make the mixture the opposite of what it is. If the mixture is lean, the ECU extends the injector pulse widths to make it rich. It does this in a series of steps, as it has no idea how lean the mixture is. Then, once the mixture has gone rich the ECU shortens the injector pulse widths to make it lean again. This cycle is repeated every second or so, and is by definition a control system using feedback from its output to modify the input. And this is what we call Closed Loop.
Because of the time delay needed to get the air/fuel ratio cycling around stoichiometric, an accelerating engine can’t be effectively controlled in this way using Narrow Band sensors. This is why most Closed Loop systems operate only when the throttle opening is constant and the rate of RPM change is very low. In addition to this, the reason for this system in the first place is to comply with emissions laws. So the system is configured to pass the regulated test and otherwise let the engine work as well as it can.
As a side note, some of the engine management systems these days are smart enough to work out when the vehicle is being subjected to a test and go into ‘test mode’, running as required to pass the test. Several large diesel truck engine manufacturers were caught doing this by the USA EPA a few years ago. Who, being the kind of ruthless bureaucrats they are (and need to be), simply bought forward by a few years the new emissions laws for trucks. Which was about the nastiest and most simplistically effective punishment they could hand out. The engine manufacturers won’t be trying that shit again anytime soon.
Anyway, to the testing procedure. The tests themselves require a vehicle to be driven or ridden on a load dyno and for the vehicle to follow a pre-determined speed/time cycle. In reality, it’s like playing a video game I saw a test being done at the Victorian EPA testing station back when I was at uni in ’91. In that case, the driver watched a monitor, on which there was two lines. They had to keep a dot in between the two lines by accelerating or decelerating as required. There are also portions of the test where the vehicle sits idling. The actual test cycle, maximum speed reached during the test procedure and the duration of the test varies from country to country as you’d expect, but they appear to be becoming more uniform. Probably as European Union emissions regulation is now finally catching up to the USA and between them they’re the biggest markets in the world for most manufacturers. Regions like India and Asia are huge markets by themselves, and may differ quite a bit.
I’m not sure how the tests allow for differences in vehicle aerodynamic properties, engine size, etc, which will affect how much load (especially as a percentage of the vehicles potential maximum) the tested vehicle will experience.
Application in use
All of the above test and practical restrictions mean that the Closed Loop systems usually operate at low throttle openings (below 20%) and below 50 to 60% of max RPM. Above those throttle and RPM points the system goes back to Open Loop operation, running off the fuel maps stored in the ECU.
The RPM at which the Closed Loop / Open Loop switch happens is generally determined by the maximum speed reached during the test cycle, how hard the vehicle has to work during said test and the chosen gearing for the vehicle. This is why changing final drive ratios (sprocket sizes) is considered breaking the law it changes the vehicle form how it was homologated.
The throttle position at which the Closed Loop / Open Loop switch happens is generally determined by the relationship between engine size and the size of the throttle bodies, and by how much load (and therefore throttle angle) the test requires the engine to provide. Cruising at 100km/h on a 900 1000 Ducati 2V engine with 45mm throttles will see a throttle angle around 10 degrees, whereas on a 4V of the same capacity with 50mm throttles will be around 8 degrees. Not a big difference, but as all the Ducati 2V engines run the same 45mm throttles you can see the difference 620cc will give as compared to 1000cc.
To show this a little more clearly I’ll show a fuel map with approximated switching values, with the Closed Loop section in green and the Open Loop section in blue. This is a 916 std fuel map, but it’ll do just fine for now. As you can see, the idle is also controlled by the Closed Loop system. The logic is simple - RPM and throttle inside green zone: Closed Loop, RPM and throttle outside green zone: Open Loop.
Another thing to remember is that the Closed Loop system is only activated once the engine is up to a predetermined temperature. On the air cooled BMW boxers, this is 80 degrees C. On the Ducati Sport 1000 I’ve been playing with this is 95 degrees C. I’m not sure what it is on the liquid cooled models, but I expect it would around 65 to 70 degrees C, certainly under the thermostat opening temperature.
Adaption (or self tuning)
The Closed Loop system is not only used for instantaneous mixture adjustment at constant throttle /RPM conditions. The changes the ECU makes to the injector pulse width to reach the cycling mixture as compared to the map pulse width is stored in what is commonly called an ‘Adaption’ table. This table is then used to trim the map value in future. Instead of having to work its way to the cycling mixture point in steps of incremental pulse width change, the ECU will start where it ended last time. The constantly updated adaption table is the self tuning facility talked about with Closed Loop systems, but it is often misunderstood.
As far as I understand it, the self tuning will only occur in the green area on the above map, so it only happens at cruise, overrun and idle. And the self tuning system aims for a Lambda of 1, as this is the only thing the system can measure rich or lean, not degrees of with a Narrow Band Lambda sensor. Lambda 1 means an air/fuel ratio of 14.7:1, corresponding to an idle mixture around 0.5% CO. Generally, best idle on the twins comes at 3 to 6% CO and best cruise comes at an air/fuel ratio of 13.5 to 14:1. So the self tuning function makes them leaner than desired all round, and does nothing for higher throttle and RPM areas of the map where you can get large changes in fuelling with modifications to inlet and exhaust systems.
One benefit of the adaption table system is that it allows for variations between engines as they are produced and also allows for engine and component wear and tear, such as an old fuel filter that’s getting blocked, that sort of thing. Although this is limited by the layout and make up of the engine’s inlet and exhaust systems. On the older BMW twins, with only one Lambda sensor and one fuel map (no offset or second cylinder map) the Lambda sensor sees the combined exhaust gas. Meaning you could have one cylinder lean and the other rich, with the Lambda sensor picking up the stronger of the mixtures at any time. This can cause problems. The new ‘hex head’ 1200 series twins GS, ST, RT, S have a fuel map and Lambda sensor per cylinder.
The Closed Loop Ducatis, while all having a main fuel map and an offset fuel map, vary a bit in Lambda sensor location the ST3 has the Lambda sensor in the front header pipe, the S2R1000 and S4Rs have the Lambda sensor in the Catalytic Convertor box, where it reads both cylinders as the gases mix, and the Sport Classic series have a small connector tube with who knows what flow characteristics.
Clearing the adaption table
To clear the adaption table, generally all you need to do is disconnect the battery for a few minutes. This shuts off the constat power supply to the ECU and wipes the RAM where the adaption table are stored. With the BMW diagnostic software, this is also done whenever logged fault codes are cleared. Not sure about the Ducati Marelli systems they don’t tell us stuff like that, but I assume it’s similar.
Adaption in use
When I was playing with my R1150R, I had the ECU on top of the fuel tank in a tank bag. This was so I could swap eproms without removing the tank, etc, to get to the ECU. I would disconnect the ECU, remove it from the tank bag and replace the eprom. So with every eprom change I’d have a zeroed adaption table. In the mornings on my ride to work when I’d just swapped an eprom, this meant I could clearly feel when the bike went Closed Loop. Before it went Closed Loop it ran as sweet as could be. This is because, as far as I can make out, the base map is generally on the rich side (like any normal map would be) and it gets adapted leaner to Lambda 1.
If I was riding at the time when the engine reached 80 degrees C and went Closed Loop it suddenly went all doughy, and gave that characteristic “someone’s gently holding the bike back” feeling. If I was stopped at a set of traffic lights (it always happened around the same intersection) the idle dropped and became a bit raggedy.
If I didn’t change the eprom or zero the adaption table before the next commute it would start out a bit doughy before becoming more doughy when it went Closed Loop, and after a couple of commutes I couldn’t tell when it went Closed Loop anymore, as there was very little difference.
The speed with which the adaption table sets itself depends on how much time you spend around each throttle and RPM break point on the adaption table. If you don’t commute on the bike then it will take quite a while, and only happen at Throttle/RPM points you use. If you commute, it will happen much faster as I found, as you use much more of the Closed Loop area of the map when commuting in typical peak hour traffic and far more often. I’m not sure if the adaption table uses the same break points as the base fuel map or not I’d expect it to, but it may be a much coarser map with less break points and a more blanket type correction.
If you fit a new pipe or air filter, etc, you need to allow some time for the adaption table to set itself accordingly, during which time it may surge (more?) or backfire on overrun, etc. I’m told Ducati are quoting a certain number of starts, which is one way of doing it I guess. We generally found a couple of hundred km of tooling around on the BMWs was enough.
Adaption and tuning
The adaption table, by definition, constantly modifies the fuelling in the Closed Loop area to give a very close approximation of the fuelling required for a mixture of Lambda 1 (14.7:1). This means that if you make any mods to change the air/fuel ratio in the Closed Loop area it will be overridden, sooner or later, by the Closed Loop system. So if you fit a PC3, Dobeck or Dimsport add on box and tune the whole fuel map the Closed Loop system it will return the tuning to how it was previously in the Closed Loop area. But, this is expected, so it’s not a big deal as such. Or, if you knew where the Closed Loop throttle and RPM cut offs were, you could just stay out of that range.
My theory (which is untested) on how to work out these cut offs out is by using a load dyno. You hold the bike at a constant RPM 3,000 would be a good start and monitor the Lambda sensor output signal voltage using a scope or analogue multimeter (digital may not be fast enough, although you could use the AC voltage function) when holding a constant throttle position. You’ll need some way of reading throttle angle too either thru the PC3 or Dimsport or using the Ducati Mathesis or DDS or Technoresearch diagnostic system.
The signal voltage will switch from 0.2 to 0.8 to 0.2 volts repeatedly when it’s running Closed Loop. If it switches, open the throttle a couple more degrees and hold it there. The hold time would need to be at least a few seconds. If it still switches, open a couple more degrees, etc. When the signal stops switching from 0.2 to 0.8 to 0.2 volts repeatedly you know the system is running in Open Loop and the throttle angle you have is what you need to know.
Now you’d do the same thing, but with varying RPM. Open the throttle to a certain angle (less then the switch throttle angle found during the previous test) and starting at 3,000 or so RPM lower the load to raise and hold the RPM in steps or 250 to 500 or so. Again, by watching the Lambda sensor output signal voltage, you’ll know whether the engine is running Closed Loop or Open Loop. Quite simple, and should only take a few minutes for an experienced operator.
Then you’d just tune the Open Loop section of the map as you would have previously. You should be able to use an older model PC3 to do this without any problems, regardless of what Dynojet say about fitments.
Removing the Lambda Sensor
Go straight to the source and purge the bad guy. What you encounter here is how the system reacts to having the Lambda sensor disconnected. We’ve done it on a couple of Ducati models (Sport 1000, ST3) with varying results. The idle mixture on the Sport 1000 went to about 8% CO, which is quite rich, but not a huge change in % terms. It still idled just fine. The ST3 went to over 10% CO and idled quite poorly.
Neither had the adaption table reset though, which may have an impact. And both were done with the engine running as far as I can remember. Maybe we need to try some more stuff, as procedure or sequence could also influence the result. Best to start from cold with sensor disconnected with ignition off and battery disconnected to zero the adaption table.
Whether or not you can tune the idle mixture with the air bleeds I’m not sure. I tried with the Sport 1000, but the result was a bit inconsistent. As these bikes also have an idle control system, keeping the idle at a predetermined speed for a given engine temperature, winding out the air bleeds shouldn’t necessarily cause the idle to increase.
Whether or not disconnecting the Lambda sensor itself causes any ECU wackiness is another point of question, with many people talking about “Limp Mode”. I’ve seen enough BMW with the Lambda sensor disconnected to think it doesn’t. The ECU will simply see an open circuit in the Lambda sensor and log a fault. Then I expect it will use the base fuel map (or maps) and adaption table (if not zeroed) and go from there, maybe adding a factor of some sort as well.
Feedback from Ducati S2R1000 Monster owners appears to indicate that disconnecting the sensor and zeroing the adaptions works just fine, without any real change in fuel usage as well. Some owners of 1100 and 1150 series BMW found a slight increase in fuel use, but no other issues. This increase in fuel usage I’d expect, as the base fuel maps are, in my experience, a little richer than you’d have them for Open Loop operation.
So, to get back to the base map for the sake of consistency, you need to clear the adaption table. Once that’s done you have a typical open loop system. To safeguard the Lambda sensor in case you want to ever refit it (resale time) it’s best to remove it and store it in a clean environment away from stuff that can contaminate it, such as silicon sealants. You plug the hole in the header pipes with an appropriate threaded plug M18 x 1.5 I think.
Some people remove the Lambda sensor from the pipe and zip-tie it to the frame, etc. This means the sensor will still be active and reading excess oxygen, enrichening the Closed Loop area as much as it can before reaching the preset limit in the software, at which point the ECU will log a fault such as “Lambda signal high limit lean” or the like. This enrichening isn’t necessarily good as many think, it just makes the mixture richer. Best to remove and disconnect the Lambda sensor and then zero the adaption table.
The “Limp Mode” thing is something that has a big effect when a principle sensor, such as the TPS fails. I sent Duane this report before I published it to get his thoughts/corrections/approval and this was his reply on the Limp Mode thing, cut from his email. I’ve included it not so much because it’s directly relevant to this report, but it’s some info you wouldn’t normally see and I know some of you like this stuff:
The Marelli ECU uses RPM to look up a reference TPS value (varies on 5.9 models from 3 degrees at idle to 85 degrees at about 6000 rpm, depending on model.).
Also Marelli ECU don't really have a limp mode - they always run the same software loop, just substitute default values for bad sensors. Marelli ECU compensate for bad temp sensors on a cold start by using the other sensor (ie use CTS for air temp up to 23 degrees, use air temp for CTS up to 23 degrees) so they can still do cold start enrichment because it's principally a map which uses the cold temp and an rpm counter to wind down the fuel trim (not in the mapping program).The cold enrichment maps start at +50% or more for the first 200 or 300 revolutions (depending on the cold temp).
Tuning with an add on box PC3, Dobeck, Dimsport once you’ve removed the Lambda sensor and zeroed the adaptions should be the same as any other Open Loop bike.
Why does my bike surge?
The eleventy billion dollar question. I don’t know. Personally I think that bikes will surge more noticeably under Closed Loop due to their quite high power/weight ratios and the effect of that on vehicle response. Most cars are much less responsive to small variations, so they mask this better. The car I drive, a little ’00 Nissan Pulsar, does surge in some situations, but it’s very, very minor. Most people wouldn’t notice it.
Also, bike engines are all single throttle per cylinder, whereas the car engines are generally fed from a single (or pair of) throttle body. The BMW M3 (I think) and maybe some other high end stuff would be the only exceptions. This I would expect to add the balance/synchronisation issues that single plenum engines don’t have. Although single plenum systems are not always equal flow systems anyway. So this may or may not have an influence.
Car engine management systems are much higher spec than bike stuff as well the emissions requirements are much more stringent. So the smarter management (and direct airflow measurement) may make some difference.
Plus it can be an owner thing to some extent. I’ve ridden lots of BMW that allegedly surge - some do, some aren’t that bad. But, I’ve also ridden BMW that allegedly don’t surge and thought they were quite bad. So there’s definitely an element of owner perception to it, as there also is to Open Loop bikes.
BMW fixed much of it on the 4V boxer twins by adding a second spark plug, offset at the bottom of the chamber. Made a very big difference, especially to the R1150GS and R engines in my experience. They also had revised mapping that you accessed via a coding plug change for the single spark R1150 RS and RT models that made a big difference to them.
I really don’t understand how this works though. In that I don’t see how you can change Closed Loop logic to fix an issue like this, given it’s just a simple feedback system that oscillates around a consistant (and unchangeable) switch point. Maybe it’s an ignition mapping thing, I don’t know. You can certainly get away with a leaner mixture by running more ignition advance, but that does have limits.
When I was playing with my R1150R I fitted some bigger cams to it, which Duane did an eprom for. With the R1150R based eprom it surged noticeably. With an R1100S based eprom it was (like most R1100S) almost surge free. So there is the capacity for software changes to help things, but I don’t understand how. And, of course, the companies responsible will never tell you.
Likewise, Ducati have released a service bulletin for the ’06 ST3 with an ECU replacement for these bikes that are experiencing surging and other low speed poor running symptoms. I do know that, as a general rule, the non Closed Loop 5.9M ECU Ducati maps tend to have not that much spark advance at lower throttle openings. I could fit the Mathesis up to a Closed Loop ST3 and go out and log the spark advance using road test function, but it’s somewhat down my list of stuff I need to do.
That’s about all I can think of to say. Hopefully it’s been easy to understand the points I’ve tried to get across. It’s just a basic system with basic feedback, and once you understand what the system is trying to do and when it’s fairly easy to see the effects and to relate them to what you may be feeling riding the bike. Doing something about it may not be so easy however, but that will never stop the intrepid home tuners out there or the internet speculation, of which I’m a part.