Announcement

Collapse
No announcement yet.

Lost in a sea of tuning... (Ostrich 2.0)

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    Lost in a sea of tuning... (Ostrich 2.0)

    So, I thought a stroker motor would be cool...
    Bimmerheads engine (used w/6K miles):
    2.9 Mahle motorsport stroker
    10:1 Compression ratio
    M52 Crank ( 84mm stroke)
    85mm Bore
    Super sport head with performance valve springs and 274 cam
    S50 Connecting rods
    ARP Rod bolts
    ARP Main studs
    ARP Head studs
    Nuke adjustable cam gear

    And it is, but it is not running well on the stock ECU.
    Rich at idle, lean under load, and cuts out at 3900 RPM @ WOT.

    So, I got Ostrich 2.0 and TunerPro RT. (How hard could it be? 4 or five inputs: Temp, AFM, CPS, O2, TPS and a few outputs: Injector pulse, Timing.)
    Oops. What are all those variables? How are they related? And what do they do?

    I have 7 tabs open at https://www.thirdgen.org/...

    I got the .bin and .xdf files.
    Got the USB driver installed to talk to the Ostrich 2.0.

    Seems like a recommended starting point is to set the "Injector Constant" since I have the M50 injectors instead of stock.
    0-280-150-715 Stock M20 325i flow rating is 14.5 pounds / 152.5 cc/min at 3.0bar/43.5ps
    M50 injector is 0280150415 180 cc/min at 43.7psi (3 bar)

    The "Injector Constant" is set to 10, so
    152.5 / 180 = 10 / X
    I set "Injector Constant" to 11.8 (The TunerPro rounded it up to 12 for me.)

    While I was there, I set "Rev Limit (primary)" from 6250 to 3250. It rounded it up to 3578 for me.

    I loaded the .bin into the Ostrich 2.0, setup the exhaust gas analyzer and went for a drive.

    Previously, idle was rich, and under load would be lean.
    Cruising on the freeway had AFR @ 14.5.

    Idle is no longer rich. It still goes lean under load. And cruising on the freeway AFR@ 14.9 - 15.5.
    And the rev limit does kick in at ~3500 RPM.

    It appears I was able to edit the .bin file and change the way the engine runs.

    Initially it seemed counter intuitive that scaling the injectors to larger flow would lean it out, but after thinking about it, the computer would shorten the injector pulse for the larger injectors, so less fuel.


    I have not figured out how to get data out of TunerPro RT. It seems I should have a .adx file, but I have not been able to find one for 173 computer.
    Does anyone have one they would share / sell? Or advice on how to log the data and review it.
    Also, any advice on what to look at next to modify would be a huge help.



    #2
    Maybe I am drowning in the sea...

    After more internet searching I found this statement:
    "... multiply by stock injector and divide by new injector ..."

    This says the constant should be smaller if you are using larger injectors. (This is the opposite of what I did yesterday.)
    152.5 * 10 / 180 = 8.4722

    So today I tried an injector constant of 9.
    The idle was not as rich.
    It seemed to pull better up the hill, and was less lean.
    But it still goes lean under load.
    On the freeway it maintains an AFR of 15.5.

    Do I need to disable the O2 sensor when adjusting fuel settings?

    Comment


      #3
      Honest recommendation: Just buy a tune from Sssquid Tuning. I have ZERO affiliation with Sssquid and gain nothing by recommending them. However, I have worked with them for multiple years on custom tunes and they do fantastic work...fully tuned idle, PT and WOT. Shoot them a message with info about your engine and they should be able to get back to you with a quote and lead time. You have an expensive engine, don't risk damaging it to save a little money.
      https://sssquid.com/v3/

      Do you need to disable O2 to make sense of fuel adjustments? Yes. There are almost a dozen pairs of constants in the BIN that need to be set to 0x80, and I do not think that those show in the XDFs online, but you can unplug the O2 sensor for a similar effect. I don't recall if missing O2 input causes the ECU to stop using certain maps that would otherwise need tuning.

      I've been where you are, and it is not going to work out. You can make the car run, but it won't run well compared to a proper tune. The XDFs available on the web are half-baked. If you have extensive tuning experience, then you may well be able to get on a dyno and optimize things, but if you do not have hundreds of hours of tuning experience and are not going to a dyno, sell the Ostrich, buy a pro tune and be done with it. A dyno tune will cost a hell of a lot more than one from Sssquid.

      For one, the injector constant in the Motronic 1.x ECUs does not do what you think it does, hence why it looks like things are corrected under certain conditions and not others. The best way I know of to adjust for injector changes is to leave the constant alone and scale the values in all of the fuel maps (idle, low-PT, high-PT, WOT). There are multiple copies of each, with some being used a lot more than others, but you need to hit them all. There are also base fuel maps and "correction" maps of the same size, with the latter not being corrected for injectors. Overall you have to adjust dozens of maps if you want to do a full fuel+spark tune, and WOT is by far the easiest to do since it is 2D. WOT is like 10% of a "proper" tune.

      I have looked through a pro's XDF that is actually correct, and it is very different than the ones on the web. Web ones are maybe 50% correct, which is enough to make a car run poorly if you get too deep into it since some things are flat-out wrong. Given the cost of the engine you have (I have a 2.1L Metric Mechanic M42, which is also not an inexpensive item), it is well worth a couple hundred extra dollars to get a proper tune in there. The car idles and responds to throttle inputs better than stock because of all the filter time constants and other non-map parameters that were adjusted.

      For reference, here's the level I went to in reverse engineering Motronic 1.7 back when I thought I'd bang out a tune myself. I am glad I got over that and just had Sssquid make a custom one for me. I ended up collaborating with them a bunch on the reverse engineering stuff, and I can safely say that they know what they are doing. Again, I have been through the full, correct XDF from Sssquid, and I would not even know where to start messing with most of it, which is why I left it to the pros.
      https://www.r3vlimited.com/board/for...se-engineering

      Comment


        #4
        Originally posted by bmwman91 View Post
        Honest recommendation: Just buy a tune from Sssquid Tuning. I have ZERO affiliation with Sssquid and gain nothing by recommending them. However, I have worked with them for multiple years on custom tunes and they do fantastic work...fully tuned idle, PT and WOT. Shoot them a message with info about your engine and they should be able to get back to you with a quote and lead time. You have an expensive engine, don't risk damaging it to save a little money.
        https://sssquid.com/v3/
        Thanks for the comment.
        I have reached out to SSSQuid.

        That was an interesting read. Very deep dive.

        Comment


          #5
          IT's not that difficult now that you are there.

          In TunerPro, you have the ability to emulate and data trace. In other words, you can see the block that the ECU is reading (will be highlighted yellow), make an adjustment, save to emulator and keep on driving.

          You need to activate emulation (TunerProRT only) in tools>emulation>enable emulation. There will be bleeping sound to let you know, and the emulation tab at the bottom of the screen will turn green.

          Open the map you would like to read. There is 5 icons at the top disk (save) "X" (close) graph (graph) scale (compare tunes) and finally "A" (read ECU).

          Now when you are emulating, you can see the actual cell the ECU is reading.

          I find the injector constant in the .xdf's are often wrong. I want to say you will have better luck using the baro correction if it's the public .xdf. Whatever number that is using isn't a true "k" injector constant.

          Realistically, you can just scale the maps themselves for 19lb injectors (reduce the numbers in cells to lean).

          The only part that's a pain when live tuning is the transition from low to high part throttles, and the WOT map (can only data trace one map at a time). It's best to have a friend drive with ya, or go to the dyno to get your upper throttle/wot and work your way back on the street.

          Too bad e30tech info is mostly gone, I started a thread there many moons ago and SSSquid was quite the forum poster back then.

          EDIT: Found it: http://e30techarchive.com/showthread.php&t=98922.htm



          john@m20guru.com
          Links:
          Transaction feedback: Here, here and here. Thanks :D

          Comment


            #6
            ForcedFirebird thank you for the instructions. I knew you could watch a trace, but when I get in the car, I forget to turn on the emulation, and then I didn't know about the "A" on the map.

            I went out yesterday with my son running the computer and we saw the trace. It was pinging in the HPT ingnition map at 2K RPM (3 rows) and 170 - 180 load. There is a lot of advance in those rows, and then it drops in the 3-4K rows.

            I reduced ignition timing there and higher (both axis) by *.85.
            And increased fueling in the corresponding cells in the HPT fuel map by *1.3.
            Will go test it shortly.

            I am finding it is hard to collect good data. We tried filming with a phone, but it was hard to see with the H&R sport springs bouncing all over. I think it will be good to figure out the logging / play back features. Is the logging able to track multiple maps?

            It is hard hard to remember what the changes are over the previous run with small changes to the map.

            Anyway, I am hoping to get rid of the lean conditions and bog at 4K RPM (I think the bog is from lean). At least until I can get a tune from SSSquid. (I was not able to recognize his username in your E30Tech tread.)

            Thanks again for your encouragement.

            Comment


              #7
              It is not lean at HPT maps. But, it doesn't pull so well with the reduced timing.
              And it still pings around 3K RPM, not as much, but it is still there.

              I wonder if the dip in rows of timing needs to be shifted up or down due to the different camshaft...

              I richening up a WOT1 fuel map, but then it did not get into that map accelerating on to the freeway, and it cut /bogged at 4K RPM.

              So many changes and results to keep track of...

              Comment


                #8
                Watch the AFM transfer maps and see if you notice any "blips" where the tracing jumps significantly. The wiper track in your AFM may have a worn spot in it, worth a look.

                The WOT maps should be reached when the throttle switch is closed. Check continuity for the TPS at the ECU connector.
                john@m20guru.com
                Links:
                Transaction feedback: Here, here and here. Thanks :D

                Comment


                  #9
                  I think I got the pinging at ~3KRPM to go away with reducing timing advance in that region and increasing fueling in all maps by 1.3.

                  It still goes lean sometimes. It is hard to correlate to the engine because I am using an exhaust gas sniffer. It takes ~20 seconds for it to register anything when I start the engine. So, results are delayed 20 seconds from engine output.

                  Any recommendations for a wide band O2 sensor that can output narrowband to the ECU and wide band to a meter, or computer?
                  It might be convenient to have it to the computer with TunerPro RT.

                  Comment


                    #10
                    Spartan 2, for sure. I have been running one for many years with great success, logging the wide-band output and running the ECU on the simulated narrow band output.
                    https://www.14point7.com/products/sp...a-controller-2

                    If anything, it is awesome because the LSU4.9 sensors are cheaper than the stock replacements lol.


                    Did Sssquid ever get back to you?

                    Comment


                      #11
                      bmwman91 how do you log the wideband O2 output?

                      SSSquid did get back to me. He had good feedback about the injectors and requested the part numbers of my injectors. When I bought the mustang injectors, they sent two different parts, but said they were the same. It turns out they actually have slightly different flow rates, and Jay says the have different startup times.
                      3x 0280155746 1275194 (208.6 cc/min @ 3bar)
                      3x 0280155710 F6VE-A5A (210 cc/min @ 3bar)

                      I still don't know what I am getting for the tune. (How do I know they aren't just increasing the rev limit and tweaking the WOT map?) And I don't know what questions to ask to clarify what I am getting... So, I am still thinking about getting a tune.


                      I did get more data. My Innovate LM-2 arrived today. I got the accessory that plugs into the tail pipe so I can use it until I get an extra bung welded in.

                      The response times are much faster than the exhaust gas analyzer. And so, it is easier to correlate the readings with vehicle performance.

                      I started working on the idle as it was showing lean. However I was not able to change the value. I increased the fueling for the idle fuel map until it was equal to the Low Part Throttle Fuel map, and it was still showing 16 -17 AFR.
                      I reset the idle fuel map to normal settings (115) and went for a test drive. I notice that off throttle always goes lean. (Innovate documentation says this is normal and it can mess up logging ...)

                      I have a hill that I can accelerate up to demonstrate the cut at ~3800 RPM. The WBO2 shows that I am not lean until I need to let off the accelerator. This is good that the engine is not lean under acceleration.

                      I switched back the stock tune, and it is still not lean on acceleration. But the cut at ~3800 RPM is more pronounced (it gets worse).
                      I think I need to work with the ignition map in that RPM range.

                      The Innovate LM-2 can take analog inputs and a Tach. I think the most useful would be the AFM signal. Where is a good place to get that signal? At the ECU connector? Is there a recommended way to tap into that?
                      Would it be interesting to log the injector output? (Would that work?)

                      Thanks for the feedback.

                      Comment


                        #12
                        Ha, yeah mix-n-match injectors are not really ideal.

                        Sssquid does not do half-assed nonsense like changing a couple of parameters and calling it a tune. Anyway, I understand if you are hesitant to spend cash since he is just some stranger as far as you are concerned, so you do what you gotta do. And, if you just want to have some fun learning and playing with data logging, then more power to you. Just try not to burn up a piston or wear out bearings with pinging, since it sounds like the engine is pretty unhappy with any sort of stock tunes (which still seems like an indication of a sensor or wiring issue).

                        Idle, just like with low-PT, high-PT and WOT, has numerous fuel and ignition maps. You likely need to change them all, although if you were able to identify the one(s) that are mainly being used with the Ostrich then I am not sure why idle is still so lean. You are tuning/logging with the engine fully warmed up and O2 feedback off / sensor unplugged?

                        My logging setup has been a mix of a totally custom engineered embedded acquisition system, and doing some after-hours borrowing of some nice signal acquisition systems from work. The custom logger system I had originally made a number of years ago is a little overkill and does not give real time data (stores and dumps it for later analysis), and it samples at 10000Hz as I was debugging my MAF conversion during development. The later stuff I used and still use is expensive National Instruments cDAQ hardware.

                        The LM-2 might be the most economical way to get basic data logging capability for a range of parameters, although the cost is going to be the same or a bit more than a pro tune I think. Getting at the signals is a function of how comfortable you are with electronics. The least invasive way to get at the AFM signal is to pull back the boot at the AFM and stuff a wire in next to the terminal for the output signal, and then snake it into the car. The Motronic connectors are not too hard to get apart, assuming the one for M1.3 is assembled similarly to the 88-pin one for M1.7. Same deal...once you get the terminal carrier out of the shell you can carefully stuff wires into the terminal slots, tape them to the wire that comes form the terminal so yours does not pull out, and then have fewer noise and routing issues than you would if running from under the hood. The shell does not come all the way off, you slide it up the big main wire loom so it is out of the way, leaving the terminal carrier exposed. In the case of my logging setup, I have a bunch of spare ECUs and I have straight up soldered taps directly onto the PCB at all of the various points to get the signals I want...one of the spare ECU covers has a hole drilled in it that I run a 9 conductor shielded cable out through lol.

                        Logging the injector control signal takes a little more care. Reading it on the wire that connects the ECU to the injector will require you to have a system that can handle 70-80V voltage spikes. The injectors are just little solenoids with a fair bit of inductance, and when the ECU shuts off the path to ground you end up with a flyback voltage spike. I am actually logging injector pulsewidth right now as part of an investigation with Sssquid into how lubrication affects running efficiency, and I tapped right into the logic level signals from the microcontroller in the ECU since those are just 5V TTL signals isolated from the inductive flyback on the output side. I am also getting my tach signal directly from the ECU main board, reading the TTL level signal which is ultimately sent to the instrument cluster. Logging for this uses a Teensy 3.2 ($20 microcontroller board that uses the Arduino programming environment, but is much much faster and more capable).

                        Comment


                          #13
                          Do these plugs look like the engine is running lean? And compression seems okay? Sent from my VS995 using Tapatalk
                          Last edited by Tinkerer007; 08-26-2020, 05:33 PM.

                          Comment


                            #14
                            Here is some more data.

                            I was trying to figure out why the LM-2 reads the exhaust as lean when it smells rich.
                            I calibrated the LM-2 O2 sensor, then installed it.
                            Started the engine. It starts out fine around 14.9, but as it warms up, it goes lean, to 15 - 17 AFR.
                            If I unplug the O2 sensor, the AFR goes to 14 -15.
                            I had the original O2 sensor from the donor car, so I installed that. Same result 15-17 AFR.

                            I measured the resistance from the ECU pin 29 to the chassis O2 connector at 98 Ohms. That seems like a lot to me. Could that be a problem? How do I find where the resistance is coming from?

                            Comment


                              #15
                              I didn't notice here, or the other thread, as to which camshaft you are running, but based on those cranking compression numbers, it's on the small side. If those numbers are accurate, you are trapping too much mass, even a fresh s50(2) is 210psi and requires premium fuel (with the benefit of variable cam timing).

                              Here is a good article explaining how cranking compression can give you some insight:

                              http://www.superchevy.com/how-to/816...e-compression/

                              Too Much of a Good Thing
                              If you're the classic hot rodder, then you're already thinking, "Hey, if some is good, then more is better!" True to form, if you do that, you will be disappointed. Higher cylinder pressure also means that you will need more octane to slow the burn rate in the cylinder. The limit for street engines seems to be around 200 psi of cranking pressure. Numbers higher than 200 psi create excessive cylinder pressure at low engine speeds, which turns the engine into a detonating monster. A classic example of this would be an 11:1 compression engine with a short-duration camshaft that features an early closing intake valve, such as 50 degrees ABDC. This would create excessive cranking pressure and an engine that would rattle its brains out on 92-octane fuel. Our current HT 383 engine actually tested at 195 psi with the original stock cam and later at 185 psi with the large hydraulic-roller cam and does not detonate.
                              john@m20guru.com
                              Links:
                              Transaction feedback: Here, here and here. Thanks :D

                              Comment

                              Working...
                              X