Forum

[Thack]

Just to clarify something: open-loop running doesn't automatically mean running rich. It just means the ECU uses an internal map to choose the injector duration, rather than using feedback from the lambda sensor to fine tune it. It can still use the engine temperature sensor to decide when it needs to run richer (for cold starting). When the engine is cold it uses a table with longer injector durations for a given throttle opening.



This is exactly how the Power Commander works - it is open loop all the time because it requires you to disconnect the lambda sensor completely. Indeed, lots of bikes from the 90s had permanently open-loop injection and no lambda sensors.



I mention this because I'm going to experiment today with running my EFi permanently open-loop to see what happens. If they've done a good job, it should run just fine, but without necessarily meeting the emission regs. It all depends on how much effort they've put into the default maps which the engine uses when it has detected a fault with the lambda sensor.



I will report back in a few hours! (Assuming I'm not walking home pushing a seized B5.......)

[Thack]

Hey, it works brilliantly! Admittedly I wasn't able to take it above 60mph, but I did manage to give it a few bursts of full throttle and it feels absolutely fine.



In fact, to be specific: the fuelling feels perfect throughout the range, with no noticeable flat spots or hesitations. The acceleration is fine - not noticeably different from closed-loop running. Interestingly, I think it might be running a shade richer than before, as the exhaust note might just be slightly softer (less "blatty"), but I'm not very sure about this and think it might just be imaginary.



Best of all is the tickover; it is steady, reliable, and shows none of the varying and slowing down that it does in closed-loop mode. It hasn't done any of those occasional stalls just as you come to a halt, either, but it's early days just yet and that might still be there.



My conclusion is that there are no fundamental problems with our EFi systems. Whoever developed the default "open-loop" map did a fine job, as far as I can see. Perhaps Keihin developed that map?



However, I'm confident that there IS something wrong (or, rather, less than perfect") when running in closed-loop mode. I wonder if Keihin left it to the Indian's to produce the closed-loop map; or maybe Keihin did it, but something has changed since then with the production vehicles. I'm vaguely suspicious of the temperature sensor, but it's only a gut feeling and I need to do a lot more research yet.



If you are feeling adventurous, I strongly recommend that you try it on your bike. I just disconnected the 4-way connector for the lambda sensor - it's located under the tank, near the front, and you can just manage it without taking the tank off (I wrapped some tape over the connector with the visible connections). If you do, perhaps you could report back your findings?



As I say, it's too early to come to a firm conclusion as I really need to put more miles on it, but I am cautiously optimistic that we can improve the behaviour of the engine without having to shell out a load of dosh. I'm going to leave it open-loop for now and will continue to report back.

[Scalyback]

I will be bust with a visiting friend from overseas, but if I get the odd hour or two, I will try on thunderbolt, my EFI.





Will be neat if the tickover is constant.

[Scalyback]

BUSY NOT BUST



gimmie edit, gimmie edit

[Thack]

Scaly, have you tried changing the plug? See the other thread: "The Americans...."

[Thack]

Scaly, when I get chance I'm planning to stick some instrumentation around my B5 engine to see if I can observe what is going on when the idle speed varies like it does (including stopping altogether). I'm certain we can fix this without having to put up with the permanently-lit MIL lamp which unplugging the lambda sensor gives us. However, rather than "poke and hope" I want to see if I can properly understand the underlying cause before coming up with a fix.



If you are willing to try yours open-loop, and let us know how it went, that would be great. Mine works beautifully, but I'm the first to admit that a sample size of one is no good for making decisions.