No announcement yet.

Individual cylinder trims getting me down

  • Filter
  • Time
  • Show
Clear All
new posts

  • Individual cylinder trims getting me down

    I think I need help in finding a new Engine Management System. I currently have an all aem(series2) setup. the last component I purchased for my 2.0L Honda build is aem's 4 channel wideband controller. my specific goals for this component is to enable individual cylinder fuel trims in feedback, or closed loop operation. in aem tuner the individual cylinder fuel trims are a 2-d map that lets me adjust injector duty cycle vs. rpm but in a fixed manner. this will allow me to compensate for the allowable differences in cylinder compression or injector flowrates from one cylinder to the next, but will not help the other 3 cylinders from leaning out should 1 injector hang open or 1 spark plug not fire. upon consultation with aem, i am told they do not have any controller that supports 4 channels of o2 feedback. and do not have any intensions of creating or supporting this feature in the future that they know of. so, I need to steer myself towards a new ems. one that will support 4 channels of o2 feedback. not just monitoring. I am looking to tie each injector to a different o2 sensor and have each cylinder trim itself. anybody here have any suggestions on which brand or specific ems will or might support this feature???

  • #2
    IOW O2 sensors are not fast enough? (Just as I thought.)

    So are they fast enough to react to a single cycle (2 crank rotations)? Not that I think they can be mounted in each header, and I suspect with a newish engine (4 valves per cyl etc) you would not want long headers...(??)

    The only method I know for what you want is ionic. O2 sensors just don't cut it - tho I may be out of date re O2 sensor technology (not that afaik the basics have changed - ergo, still useless).


    • #3
      aem uses bosch 4.2 uego o2 sensors. they are among the best on the market. and they never need a free air calibration. the 4 channel wideband controller is designed and built for individual cylinder fuel trims, so It doesn't have to operate or switch at a higher speed than any other o2 application would. it will trim at the same speed as always, just localized to 1 cylinder. it comes with tall and finned o2 bungs to tig weld into exhaust header primaries. it's just that usually this setup is used for very high horsepower applications and at the track only. and usually track cars don't even use o2 feedback.


      • #4
        So are you saying they have an output reaction time of (say) 10mS or less?


        • #5
          what I am saying is moving an 02 sensor from the collector to a primary and having it read 1 cylinder does not require the o2 sensor to have a faster reaction time. i'm not looking to trim individual combustion events, just individual cylinder fuel trims. in feedback. so if 1 fuel injector fails, the ecu doesn't dangerously lean out the other 3 cylinders trying to compensate.


          • #6
            Yeah, but what I was saying was that afaik, O2 sensors were not fast enough. You said they were but later that you could not trim properly which sounded to me like a timing/delay issue.

            So what is the reaction time of the 4.2 uego 02 sensors?

            BTW - when I say DIY EFI, I mean DIY programming etc - not the mere DIY tuning. Hence fully configurable as you want. If that can use existing hardware, fine, but my DIY EFI/EMS was a full build.


            • #7
              megasquirt is so much less than aem's series 2 ems'. I did my homework. yes you can solder chips to a board, but in the end, the aem solution has better features, a more intuitive software suite, is more reliable, and 3 times as much money. proof of this is everybody I have told I want to switch to a different ems asks me the same question. "how much do you want for your aem ems?" I have one friend who wants to trade me for his hondata s300, one friend wants to trade me his 3rd gen megasquirt. I have seen homemade or home assembled megasquirt systems, and imho they are like a radio shack build your own fm radio vs. a fully equipped sony home theater system. the entire ems-4 with connector and completely sealed (waterproof, shockproof) enclosure fits in the palm of my hand. the electronics inside are too small to come in a megasquirt kit. it is all surface mount stuff. not the silicone board with a million holes you get from megasquirt. to build all the features the ems-4 has into a megasquirt, it would be huge. the built in 8mb datalogger, the usb 2.0 connection, the canbus, sequential injector and coil drivers, 8 g.p.i.o.'s with 4 of them supporting pwm. and though I am not positive on this, I am relatively sure that megasquirt will not drive an engine to 25,600rpm. I will never spool my engine to those speeds but it is so comforting to know that even when I am screaming my motor to 9,000rpm, the ecu is not on the verge and beginning to skew injector or coil triggering. anybody that knows of aem and has something else is always and only because of money. nobody says "because aem can't do this or that" or "because aem is unreliable" or any other complaints other than because the controllers are so expensive. and well as always in this life, you get what you pay for. and while I am currently unable to find the specific reaction time of the bosch wideband 4.2 or 4.9 sensor, some controllers sample the o2 reading every 2ms.

              AEM has plug n play ems systems that come with or you can download tune files for and from there it is a matter of tuning. the ems-4 is a universal standalone engine controller. it can be used on a single cylinder 2 stroke, it can be used on a v-twin, or a 3-cyl. geo, or a 4 cylinder, or even 6 or 8 cylinders with batch firing and waste spark. setting up this controller falls more under the category of programming than it does tuning. seriously. you have to tell it how many teeth are on the crank wheel and cam wheel, wether you want it to trigger off of rising or falling edge, wether each wheel is a missing tooth design, and what type of sensors they are(optical/hall or magnetic). then once that is set right, you have to tell it what injectors and coils you wanna use, and at how many teeth on the crank wheel you want each coil and injector to fire. before you can even start setting up a fuel or ignition map, you first have to configure all the breakpoints. so you really create your own maps. and the breakpoints don't have to be linear, so you can have more cells in the idle area if you wanna try to get the smoothest idle possible out of it. I strongly encourage you to download the latest version of aemtuner and just look around at/in it. the software is free. it is as diy as you'd ever wanna get.

              my problem isn't that my o2 sensor isn't trimming. it works fine. it is that I know that if an ignition coil arcs to ground rather than through the spark plug and that 1 cylinder doesn't fire, the 1 wideband I have downstream of the collector is gonna see rich and lean then other 3 cylinders out. this same thing will happen if an injector fails. or if 1 injector stops firing and is getting no fuel whatsoever, the ems will never see this or at best it will riched the other 3 cylinders a touch. I want feedback, or closed loop, or automatic trimming on each cylinder individually instead of looking at the exhaust of all 4 cylinders together. I want more precision in my engine's fueling than an average afr of all cylinders. I want to treat each cylinder like it is it's own engine.
              Last edited by AutoWiz; 01-20-2014, 06:45 PM.


              • #8
                Are you replying to me? If so I don't understand why you are repeating the obvious.
                And maps? Hey man, I'm into self tuning, Maps are things those primitive Haltechs have (all 32 of them LOL!).

                So, what is the response time of the Bosch 4.2?


                • #9
                  that is going to be determined by the guage/controller for the wideband. some controllers and I believe my ems-4 look at it every 2ms. as far as how fast the actual element can respond in relation to a given combustion event would greatly depend on cylinder volume vs. exhaust diameter and length up to the sensor and maybe exhaust backpressure should also be a part of that calculation to be accurate. yet another reason to move the o2 sensor to just a few inches off the head.


                  • #10
                    No. The "sense" time of the 4.2 - ie, how long between a "sample" until its equalisation or equilibrium chamber (or whatever they call it) has the appropriate reading ready. Since that depends of how far off stoich it is, let's assume it's at 14.5 or 14.2 (or whatever the 4.2 quotes) instead of 14.7, and let's not worry about how the sample is triggered (ie, how ever many mSecs after ignition).
                    I know EMSs can sample however often, but I want to know the 'reaction' time of a wideband like the 4.2 (which of course is slower than narrowbands assuming the same technology).

                    If it were 2ms and your ems-4 looks at it at that rate, I fail to see why you cannot trim appropriately. You should have all the info required. (Granted, IMO it's the same as a typical "self learning" process, and some consider that "impossible"...)


                    • #11
                      once more, the limitation of the sensor is the controller that drives the sensor and exhaust flow. or would you disagree that exhaust speed will impact how fast the o2 sensor sees a change? the smaller the diameter of the primaries and the closer to the cylinder head the o2 sensor is is how you increase reaction time. I feel like I need to point out to you that trying to use 1 o2 sensor downstream of the collector and hoping to measure o2 content from one specific cylinder is a loosing proposition, and could never be accurate in a real world setting. wether or not an o2 sensor is capable is irrelevant. its a great thought, but I think you're better off pushing ion or ionic sensing. and even if you had that, you still wouldn't be able to duplicate what I think you are talking about. o2 sensors have never measured individual combustion events, or even single combustion cycle for all cylinders afr. that is not how they work. the newer controllers sample the o2 sensor more frequently and in this manner provide for longer times spent in closed loop operation. but this is why it is so important to have a good tune, because anytime sudden changes are made to the throttle the first thing the ecu does is switch to open loop and stop using o2 sensor for feedback and just rely on fuel map, until airflow across the mass aif flow sensor goes stable, or rpm stabilizes(if running speed density). you would at the very least need to create some way of measuring exhaust speed with deadly precision. and to realize a calibration for timing airflow from the cylinder side of exhaust valve to the o2 sensor for 1 specific application would be a long and tedious process. and how do you compensate for the exhaust port building carbon and shrinking in size and effectively increasing exhaust speed on the affected cylinders? even if this could be created, even if, for the time and effort it would take to try to realize it, 4 o2 sensors aren't that much. and there is no more accurate way to measure the exhaust gas than to probe the primary, anyway.


                      • #12
                        also exhaust manifolds with equal length primaries are hard to come by and usually only used for turbo applications and not even available for a lot of applications due to lack of clearance or room under the hood for the extra tubing. with most v-8 longtube headers the front cylinders primaries are noticeably longer than the rear this can allow for overlap in exhaust gas where 2 cylinders exhaust hit the o2 sensor at the same time or closer or farther apart.


                        • #13
                          When measuring temperature, I want to know how fast the temperature senor reacts.
                          How fast the temperature is rising or how often and fast I read that temperature is a different issue.
                          Do you now understand what I am asking?

                          A more applicable example may be a DMM...
                          Its digital display could update every mSec but it doesn't (it's usually 2-5 times a sec).
                          It might update its digital or analog output (ie, serial interface or screen) after every sample.
                          It might sample every uSec (because maybe it takes 0.8uSec to sample and convert).
                          But a typical digital-display only might samples every half second.

                          My point being a typical DMM is no good for variable samples that are too fast - eg, voltage output from older dash voltage regulators (thermal "flasher cans"), O2 sensors (ignoring impedance!) etc....
                          As to its serial output, so what if it outputs at 1Mhz for a sub 100Hz signal? (IE - so what if my EMS samples every 1mS for a sensor that takes 50mS to update?)

                          So, do you have the sense/reaction time of the 4.2 or not? I'm not interested in how often it is sampled.

                          Again you have simply repeated the obvious and what I have said previously.
                          The reason I am asking is that it was you that said your O2 sensors or system were fast enough for trimming. (Whereas I did not think so.)
                          And yet you have troubles with individual trims which I took to be 'sampling' issues. Hence I was curious why you were having problems - whether to solve your problem or otherwise. (That may have been when I thought you were DIYing your EMS as opposed to an EMS with DIY tuning abilities.)

                          You also seemed content with using O2 as a control method despite the stuff you just wrote. [ There are ways of compensation for those factors using O2 PROVIDED the O2 sensors are fast enough, but as you say, why bother when ionic does it so much easier, cheaper, and completely with no sensors required (excluding the COP or splug add on). ]

                          What is the reaction or sense or (min) sample time of the Bosch 4.2 sensor?

                          PS - I missed your last, but that too is compensatable. Even dynamic headers (pipes) can be compensated for.
                          Last edited by OldSpark; 01-21-2014, 07:30 PM.


                          • #14
                            I want o2 as a control method for failures. to make my reliable setup even more reliable. otherwise, I have a good speed density tune established. I don't think ionic sensing has made it to the standalone controllers, yet. on an old school carb'd engine a misfire will make the motor spool up lopey and with less power. that same misfire on an efi engine will make the whole car buck and surge because the whole bank or whole engine's fueling is is being affected because all cylinders share 1 o2. we went from 1 coil to individual coils,. from 1 or 2 injector(s) to individual injectors for each cylinder. repeating this for o2 sensors is the next logical step, imho. for the specs you are looking for, you'd probably have to contact bosch. they also have a newer 4.9 sensor out that is more $$$ and doesn't work with most controllers and has some typ of ion pump or ion pumping feature or something. I do not know too much about these new 4.9 sensors. these are all the specs I can find on 4.2. the sensor can not operate apart from its controller. so any specs bosch did quote would be of the controller. NOT the ems. the wideband controller chip that drives the 6 wire sensor. for example, my ems takes a 0-5v input for oxygen sensor. but the 6 wire sensor does not generate that on it's own, nor does my ems have a controller for the heater ckt. so I have aem's "in-line" wideband controller and that drives the o2 sensor and sends a 0-5v signal to the ems. wideband gauges usually have a controller built into them and they have a 0-5v output for logging or connecting to ems.


                            • #15
                              Those pdfs are useless - there is no timing data.
                              The LT4 samples at 10mS intervals but that's not saying the LSUs operate at that speed.
                              I did have response data for the 4.2 etc years ago and they were far from adequate for per cycle feedback, hence I stuck to my old narrowband tuning method. But last I looked I couldn't find response time data (I lost my documentation) and hence why I was asking you.

                              And not that I know why carby is being compared to EFI or ignition misfires (many EFI's are bank injection and ignition at the wrong part of a cycle is rarely a big deal - generally only the "either side" mistiming is significant), but enuff on that.

                              I thought you had timing info since I said that IMO O2 was not fast enough but you had written elsewhere...
                              Originally posted by AutoWiz View Post
                              ... also my 4 aem uego wideband sensors and 4 channel wideband controller have no problems keeping up with individual cylinder fuel trimming.
                              - IOW you must have known the sensors were fast enough - not merely the controller sample frequency.
                              And tho that doesn't mean an EMS like your AEM EMS-4 can handle it, I've been referring to DIY EMSs whether 32 bit or hacked Delcos etc.

                              I thought you were using or intending to use O2 for your individual trims since you weren't using ionic, but if there is some other method...

                              IMO O2 sensing is way past its use-by date, especially for performance. IMO ionic is certainly the way to go - not that I've looked at it for about a decade, and - as you pointed out - that technology is now 20 years old.
                              All else you have said is well understood & known - I have long understand such basics. And things like sample delays and superposition have their solutions, but if the sampling is not fast enough (ignoring complex techniques - ie, you'd go ionic instead).