No announcement yet.

obdX and real time tuning

  • Filter
  • Time
  • Show
Clear All
new posts

  • obdX and real time tuning

    I've noticed a lot of enthusiasts here and on other sites and probably the world over want a powerful communication link with their vehicle's control modules. wether it's grabbing data to display on our carpc screens, or trying to gain control of outputs without running wires or even adjusting tunes. in a lot of cases you really can't fault the enthusiast, because we all got to make do with what we got or what is available. and the vast majority of what is going on in a given vehicle's electronic control systems are proprietary to the vehicle manufacturer. but here, in this forum on this website, here at there is no excuse. it seems as the majority of people here are very electronics savvy. so I am completely stumped and do not understand why people here will go through all the work and effort to gut the oem entertainment system out of the car and build and wire up a carpc and then limit in a BIG way how they integrate carpc into the vehicle by use of a generic or global obd2 scanner. I do not understand why there is not a plethora of carpc's attached to standalone controllers. understandably some solutions can be very pricey. like aem, or some solutions from edlebrock or other big names. but there are solutions designed meant and intended to cater to the diy'er. a perfect example of this is megasquirt. for a small handful of franklins they send you components and a circuit board. and you build your own engine controller. if you have a 4 cylinder then you solder 4 coil drivers and 4 injector drivers, if you are gonna run egr, then you build the ckt to control that as well. and from that point forward your carpc will be so integrated with the ecu that you can command different afr's, change ignition timing maps, idle speed or anything else on the fly while you drive from carpc touchscreen. without having to worry about how making a sensor lie might impact other areas of the engine controller. and without having to worry about hunting down vehicle specific communication protocols and other hard to come by info. these standalone or aftermarket controllers are designed for people like us. and to that end most of them have auto tuning features, or a side bar that pops up and explains in great detail about the field or menu or tab you just clicked on.

    obd2 was forced on the auto manufacturers. it was not their choice and they have been finding ways around it ever since. the latest thing is software updates. anybody with a newer vehicle that has not been back to the dealer in a year or two probably has outdated software in at least 1 module. most of the time there are tsb or service bullitens about this for certain symptoms that a car might experience. and it's big business because you can only get it from the dealer. even the mobile guys that ride around with a laptop or dealer specific scantool solution like the tech2 or drb3 even they have to buy that update from the dealer before they can flash it to your car. it is a bad joke the small datastream or low number of pids and all the codes that are only defined as "manufacturer specific code" I mean seriously, with your carpc connected to your car through generic/global solution if your electronic throttle body fails, you will never know why your vehicle is dead. because that p15xx code will just list as "manufacturer specific code" and how frustrating would that be dead on the side of the road with all that carpc hardware and not even have a clue as to whats wrong?

    I am dubbing my controller obdx because its diagnostic capabilities are advanced to say the least. with the engine off I can control all outputs, pulse coils and injectors individually at any hz I like to test part and ckt. if a cylinder should drop compression, I can lean out that 1 cylinder and advance or retard timing to just that 1 cylinder to try to smooth out the engine enough to make it home. these solutions have pc datalogging speeds at a much higher rate than anything you will ever get out of a obd2 datastream so datalogs are more precise and catch momentary afr spikes and burst knock. and with all the general purpose input/outputs or g.p.i.o.'s it can be an extension to your carpc and can aid in further integrations. instead of just pulling a code before I call a tow truck, I have a chance and opportunity to setup my own limp home tune to whatever is failing. if my fuel pressure drops I can radically change my fuel map and double or more the pulsewith to the injectors. if the engine develops a serious vacuum leak I can trim down or completely turn off the idle air control valve. all right there on the side of the road with carputer. if a sensor starts failing like ect or iat or even tps I can simply change its calibration so the value it is reading is close to the value it should be at. and believe it or not guys computer programmers or IT guys make better tuners than auto mechanics. that is a fact. is there anybody else here with a standalone solution? and if so what are you running?

  • #2
    You're not using ionic sensing are you?


    • #3
      no, no. I use traditional knock sensor. and I tune my car to it and have it properly calibrated and when my ems picks up knock, it pulls an equal amount of ignition timing. I know so many people delete the knock sensor, and even less know about newer tech like running current through the spark plug when it's not firing to detect detonation based on ion levels within the cylinder. I am impressed OS. but I do run optical cam and crank sensors for a much cleaner squarewave signal at higher rpm's and lower voltages. oh and my ecu is actually using a wideband o2 sensor instead of just having a guage or for logging only. my car stays in closed loop at wide open throttle and can accurately trim fuel when it is commanded to 12.8:1 afr for my n/a motor. with the individual cylinder fuel and ignition trims, I wish I did have an ion sensing setup.


      • #4
        Optical for a cleaner square wave? Hmmm - that reminds me of those that used to argue optic and Hall were "more accurate" than reluctor because optic & Hall were "digital" LOL! (As if reluctor varies with temperature and diffraction etc as do optic & Hall... )
        And those "lying sensor" things - like resistors or uPCs for AFMs etc. What - those vehicles did not use oxygen sensors? (Another money spinner for many.)

        But it was because you mentioned compression drops etc, and individually trimmed injectors and ignition timing, I assumed ionic. (Not that you need positional sensors for that. And I'd only use a crank sensor and let the smarts figure out if compression or exhaust, but my systems are envisaged as a complete add-on system to older vehicles with no significant machining nor mechanical work.)

        Not that I want to get into those old discussions again - that's long behind me.


        • #5
          well they are more accurate. or at least optical is. and especially when cranking the engine or with a dead alternator trying to drive the vehicle below 10v. that's no argument but proven fact. Chevrolet tried it for a few years on their lt engines with "optispark" but because it sat right below the water pump they were prone to failure from water intrusion and dirty lenses. my ecu will not tell me if I have reduced or no compression in a cylinder. I think I would use the starter for that diagnosis. or better my 2 channel lab scope and low amp probe around the battery cable and using the other channel to monitor cyl #1 coil so I can run a relative compression test by watching amp draw at each cylinders compression stroke while cranking and pinpoint cylinder without a single wrench or socket. I might also be alerted to weak compression with the 4 channel wideband controller seeing one cylinder going way rich. the individual ignition and fuel trims can be done by feedback(auto) or I can set trims. so because I only have 1 knock sensor, I do not have auto ignition trims on individual cylinders. and no matter how bad I want the new tech, I don't think my aem ems-4 supports ion or ionic sensing. think I would need aem's infinity controller for that.


          • #6
            Not that I want to repeat old arguments, but "proven fact"? One manufacturer's problems does not make it universal.
            The only issue a reluctor may have at slow speed is low voltage output, but that's more a function of its sensing circuitry's sensitivity. But sensing the 0V crossover (maybe up to +/- 0.6V for poorly designed circuits) requires little voltage unlike Hall & Optic modules which usually require 5V or higher. Plus a reluctor can be its own power source - ie, no battery or supply required for sensing and output pulse - whereas Halls and optic obviously can't.
            Accuracy - a reluctor is always more accurate. Whether that accuracy is taken advantage of is a different issue. (I have seen some pretty pathetic implementations.) Mechanically they all have the same rotary shaft variation, but only the reluctor is free from refraction etc issues though their modules (eg, Siemens HKZ series) are designed to minimise or eradicate those.

            Your single knock sensor should be able to monitor individual cylinders - that's a software & hardware issue, though certain sensor locations may simplify the process greatly. But a knock sensor has so many noise inputs - which ones do you ignore?. And the sensor itself is often bandwidth limited - certainly its typical associated circuitry is - and what frequencies are required? Does it differentiate between pre-ignition and (pre) detonation?
            But that's where ionic IMO is the solution. After all, knock detection itself does not reflect compression but a variety of factors (though like ionic, the knock waveform could be analysed, but let's keep it practical!).
            Same with starter motor - too many variables (bearings; starter commutation & non-linearities; engine temperature & mixtures).
            And O2 sensors - IMO too slow (not that I know modern 02 sensing times, but they were far from any sort of individual ignition analysis). And then there's the injector and splug issue.
            All in all - way too complicated to analyse all the above for meaningful results. From what I've heard, many consider ionic analysis complex enough, and that's a single sensor with limited inputs.

            Alas when I looked at ionic there was no commercial (nor production) product available - I was DIYing it - so I have no idea about this millennium's ionic sensing products (other than a revisit circa 2005).

            Not that I'm suggesting what way you go. IMO ionic is the only method that is currently practical. The others IMO are fraught with heaps more analysis (and brain pain!) that will yield less accurate results AND are not transferable to other engines.
            And I know many that have used novel techniques for the sake of it - ie, the fun of seeing how far they can go rather than having an effective end result (better fuel economy or power etc for worthwhile effort).

            But I'll bug out. I was only interested if you were using ionic sensing.
            Last edited by OldSpark; 11-25-2013, 10:44 PM.


            • #7
              saab uses ionic sensing and a few saturns did. also my 4 aem uego wideband sensors and 4 channel wideband controller have no problems keeping up with individual cylinder fuel trimming. the standalone ecu I have is pretty recent tech. it's an aem built ems-4. it will drive an engine all the way up to 25,600rpm. so I am sure it could handle figuring out which cylinder fired just before the knock started. but for the individual trims to work and be accurate they need to be tied to different sensors, because knock retard lasts for more than 1 ignition cycle, if all 4 cylinders were tied to the same knock sensor than all 4 cylinders would retard when the knock event happened. also 1 sensor does not have the resolution to see if the next cylinder down the line picks up knock or if it is the first cylinder knocking harder. all it can see is globally, there is detonation in the engine. if it were available in the aftermarket world I swear I would hunt down that ionic sensing OS. that would push my build over the top. but I digress, for I am quite certain that when it does hit the market, I won't be able to afford it, lol.

              and the only reason I haven't drilled and tapped my block and mounted a knock sensor at each cylinder is because to me that is a loosing proposition. I have an inline 4 cylinder, if let's say cylinder 2 only picks up detonation. maybe 3 or 4 degrees of knock retard. not damaging but substantial. then cylinders 1 and 3 will retard maybe 1-2 degrees because those cylinder's sensors will hear that noise, but maybe cylinder 4 won't hear it and will stay advanced and create a lope or imbalance to my motor.


              • #8
                As I wrote, that's a s/w issue, especially since the knock sensor itself does not control the retard.
                The one sensor can detect all events, but obviously dedicated sensors make the s/w (read: research and analysis) much easier since propagation delay isn't an issue and the noise of interest is more similar for each cylinder.

                I once proposed that one knock sensor would replace multi-sensors, but that was before ionic came along (ie, ionic should replace any knock sensor).


                • #9
                  well, the industry sure didn't go that route. most cars today have 2 or more knock sensors so they can better determine which half of the engine is knocking by comparing it to readings from another sensor(s). I mean, I log my raw knock sensor value and study it. I would love to know how to determine which cylinder is knocking vs. varying amounts of knock by any means other than timing it to which cylinder fired before the knock started. and past that point how to determine knock from any individual cylinder once it started. I think this is the reason this new tech is coming to light. also knock sensors don't explain why the knock. from over advanced timing or short on fuel. but I think this new tech will have that resolution. and I mean no scarcasim, OS, I would love to hear your thoughts on using one knock sensor for individual cylinder knock detection. currently that is the only lead I have to follow to achieve individual ignition trims in feedback(auto)


                  • #10
                    Well ain't the industry lazy! (LOL) Though for V engines I reckon I'd use one per side - it isn't worth the effort (and risk) using one for all.
                    But does that one per side determine which of the 3 or 4 (V6 or V8) is knocking? If so, they are doing as I suggested - analysing the signal (Fast Fourier Transforms, DSP, etc - as done for ionic (I presume)).
                    Ionic analysis is "all are the same" whereas a single knocker for multi cylinders adds extra dimensions (like sound delay; attenuation factor & spectrum; etc).
                    What a manufacturer chooses is a function of the usual factors - time to release, initial versus long term cost, etc. I could imagine initial release of one sensor per cylinder else "bank retard" etc of cylinders based on a single (per side or per engine) sensor. Then if that model (engine) proves successful, they do the analysis and programming to bring nett production costs down - ie, one sensor per side or engine whilst achieving individual cylinder response - to improve "power" acceleration or overall fuel economy (whatever the aim is).

                    The beauty of the knock sensor is that it is another closed loop system that enables backing off - eg, retard the timing - irrespective of what the preset program thinks the (non knocking acceptable) result is. It's like 02 sensing except that preventing knock prevents damage. [Or in some (ancient?) systems, allows maximum advance and hence best power/efficiency. ]
                    I'm surprised some choose to disconnect the knock sensor, but maybe they want acceleration even if for mere seconds (tho knocking IMO usually means a decrease in output). But I've seen enough ignorants fit bigger injectors expecting more power (or add AFM resistors or disconnect O2 sensors just as I used to see bigger carbs fitted to older vehicles; or baffles pulled out of exhausts etc WITHOUT additional compensations), and I remain ignorant of newer valid reasons (ie, I was into all this before SAAB even had their trionic system).
                    Of course (tr)ionic systems should achieve all the above. (Don't tell me I'm still ahead of the pack?)

                    BTW - the knock sensor won't determine why, but in conjunction with the O2 sensor (else ionic, but that does both) you can eliminate lean mixture as the cause (assuming O2 sensors are as quick as you say, and exhausts are properly tuned and mapped).
                    Last edited by OldSpark; 11-27-2013, 03:50 AM.


                    • #11
                      I think there are several things that are preventing people from just throwing away their OEM ECU and running an aftermarket solution. The biggest would be good ol Johnny Law. Not only is it illegal to do emissions systems modifications in many places across the USA, but you will lose the ability to drive your vehicle after the first inspection reveals that your vehicle no longer supports OBD2, and the plugin-in test at the inspection station fails.

                      Another reason is the high cost of entry, both in experience and in danger. To get megasquirt running on a modern vehicle you first have to know exactly what sensors your vehicle has and what voltage range they operate and report at. This includes crank/cam pickups, air/coolant temperature sensors, and throttle position at the very least. That information is not easily available to normal users, and typically requires either expensive shop manuals or a skilled hand at a logic analyzer and scope/mulitmeter. Once you have this information you have to cut off your OEM ECU connector (thus rendering your vehicle useless for the duration) to wire it into the megasquirt or other aftermarket controller. Oh, and don't forget about removing your exhaust to weld in a wideband o2 sensor bung into your exhaust. You can do it without, but it is neither recommended or smart to attempt.

                      Now that you have the aftermarket controller wired into your vehicle (assuming you've gotten it started and idling successfully) now you begin the month or two long process of tuning. This often requires quite a few drive-to-datalog, analyze, retune cycles. Even in a low power engine this can be quite dangerous, as setting fuel too lean or timing too high can VERY easily cause massive amounts of destruction in your engine.

                      Now look at the market for CarPC people. These are a bunch of DIY guys who use their cars every day and want a better information and entertainment setup. You can install it in a weekend, and it in no way negatively affects the driveability of the car unless occasionally running without music when your pc is down counts.

                      Aftermarket ECU's are a whole other level of complexity and really are pointed at a different market segment from the CarPC.

                      That being said, I run aftermarket ECU's in three out of my five vehicles. I'm an automotive enthusiast as well as being involved in the carpc hobby, which is awesome how the two overlap for me, but I have at least one stock vehicle that I can drive while the others are inoperable due to whatever issue they may be suffering at any particular time.

                      It sounds like you're going for designing your own ECU, have you checked out FreeEMS? ( It's an open source ECU (that I run in two of my vehicles). If you're looking for a great starting point to fiddle with, I highly recommend it as long as you understand the implications of do-it-yourself fuel injection systems.. If you're looking for an OEM replacement, spend the $10k on an AEM or equivalent. (Of course, then you wouldn't have access to the communications protocol)
                      "stop with the REINSTALLS, what do you think we got some lame-o installer!!!" - mitchjs
                      My Shop


                      • #12
                        Thanks for the diyefi update. I used to look at "diy-efi" until it became megasquirt whereupon the site died. But the last time I looked seriously at DIY EFI was 2005, and that was only because some newly graduated and automotive electronics industry employed self confessed expert made various claims about EMS - ie, what the could not do, or why certain things were done.

                        But I see DIY EFI very different other than the legal aspect, but that's why I keep to 1960's & early 1970s vehicles - in Australia they are "pre regulation" (aka pre-ADR) wrt to pollution etc. Hence I can take any modern engine, change to carbies or change EMS etc - provided I stay within capacity limits. If it was subject to ADRs (after ~1974), I wouldn't even be able to change the air filter type!

                        Firstly, like car audio & PCs etc, IMO you would never cut into OEM electrics. You would either use adaptors, or separate looms for more complex stuff like EMS.
                        Secondly, IMO you would never adapt an EMS/EFI to suit your sensors; you would get the sensors to match the EMS.
                        If it is a full DIY solution - ie, you are programming and constructing - then of course you can tailor the above to suit, but it might still be easier to change some components.

                        A classic example of the above is the common use of Delco systems. The typical scenario is to take the entire system from a suitable donor vehicle - ie, Delco sub-harness, sensors, AFM etc, distributor else ignition HEI, may be the injectors. The maps can be changed if required by various means. People can then have the entire EMS system for a few hundred dollars (based on Australian "Pick A Part" wreckers/recycler prices).

                        Of note, I knew several true achievers that pioneered engine enhancements etc. I watched as they pushed early manufacturers like Haltec, moved thru Autronic(?), "EMS", and whatever others I have forgotten. They all ended up using Delco systems (except for true leading edge stuff).
                        Since 2005 I've tried to convince others why some features show the limits or primitiveness of some (aftermarket) systems - eg, having 32 selectable maps (gained from dyno testing!). But I live in a country where dual-battery owners have been convinced they need dc-dc converters that can cost over $1,000 to "properly" or fully charge their 2nd or auxiliary batteries. As to the ability to understand EFI & EMS... [ Note - that typifies "public" web forums. Individuals can be very different, as can be private forums and groups. ]

                        Anyhow, my general suggestion is either Megasquirt, else Delco.

                        FYI - When I revisited a DIY EMS in 2005, I selected some 32 bit (ARM?) processor. I figured if my 1980's vintage distributorless ignition system (self learning etc) used a 68HC11, an equivalent EFI/EMS deserved an upgrade.
                        Last edited by OldSpark; 11-27-2013, 05:00 PM.


                        • #13
                          I do run aem in my AEM ACCORD. I run all aem. aem ems-4, aem epm(optical cam&crank), aem 4 channel coil driver, aem 4 channel wideband o2 controller, aem uego wideband o2, aem silkscreened full wiring harness with fuse and relay center, aem fuel rail, aem fuel pressure regulator, and aem tru-time cam gears. and I am only around $3,000 in aem products, thank you. I have a build thread here on the setup if you'd like to see. I would kill people to see what a $10,000 engine controller looks like. lol, that controller would have esp knock detection. and have a real human brain from some poor organ donor attached to it for $10,000. gosh. making sensors work is a no brainer. the vast majority of sensors work off 0-5v. and calibrating a sensor in the tuning software is easier than you think. and most tuning software is surprisingly user friendly. the tuning software usually helps and has a window to explain what each field is and does. and most even have base cal's that can be downloaded for different engines for sensors and cam and crank teeth count, just to get you started. i do not remember anybody actually plugging into my cars dlc for emissions. the way i remember it being done was with a 5 gas analyzer in the tailpipe. but with a wideband, a good cat, and a good tune there is no reason a standalone ecu can not pass emissions. even without egr. and even if the tuner were not able to setup an emission friendly tune, a little alcohol mixed with the fuel will help it burn clean. or at least that was an old trick from way back when they did still have emission testing in florida. also there is pretty much a forum for every year make and model car out there. where any and all information is readily available. you need to stop talking like we are not living in the information age. for my 5th gen Honda accord, i am a member of when i had a 1987 Honda accord i was a member of for the corvettes there is for bmw there is and i am sure there is a pinto forum out there somewhere, too. if you should be so lucky to own one of them. any information can be had for these vehicles in these forums. i do not think a standalone ecu is something people that have successfully built and installed a carpc should be intimidated by. and i think, nay, i KNOW from experience, that it gives the diy'er a great sense of pride everytime you turn the ignition key. it makes the whole car feel more like something you built, instead of something you bought. here are some screenshots from the aemtuner i am using.
                          Attached Files


                          • #14
                            there is work that is required to properly set one of these up. once you got everything figured out and are fluent with the tuning software the level of control you get is insane compared to tuning software for oem ecu's. the standalone's are so universal they can be used on ANYTHING. you turn on the coil and injector drivers you are gonna use, tell them what sensor they will use for feedback, then you tell the ecu at how many teeth on the crank wheel it fires each driver. it is simple math, but the ability to change that is huge. also being able to change breakpoints is incredible. on the corvettes at work i use hptuners. it is a more polished and better looking software, but doesn't allow the kind of control that the aftermarket aem controller gives. here are some screenshots of hptuners
                            Attached Files


                            • #15
                              Originally posted by OldSpark View Post
                              Thanks for the diyefi update. I used to look at "diy-efi" until it became megasquirt whereupon the site died.
                              Fwiw is not diy-efi. diy-efi died when the megasquirt folks killed it, and has been dead since.
                              "stop with the REINSTALLS, what do you think we got some lame-o installer!!!" - mitchjs
                              My Shop