Tuning the NSX 3.0L Using the OEM ECU

[mskrotzki]

John, perhaps you could help me out. Following on:

If you look in some of the reverse engineering spreadsheets I've created you'll see that the EGR status bit is in the same data byte as the power enrichment status bit, so to get EGR status all you need to do is add an entry to the ADX that looks at this bit, using the power enrichment status entry as an example. I'll probably just include an EGR flag by default in the "beta" release.

So I added these to the ADX file:


But a logging run I took this morning did not respond. Never once did the EGR flag turn on, nor does it show up in the csv dump of the .xdl file.

Bythe way, the colors on the dashboard correspond to the colors on the monitoringgraph -- makes it easier to understand whats happening at a glance.


FWIW,during this lovely COVID 19 crisis, I fooled around with Excel -- which I nevermuch used before -- and got into conditional cell functions. I came upwith this:

<!--[if !supportLineBreakNewLine]-->
Againthe colors all correspond to the dashboard and the monitoring graphs, more ofless. I am looking to see if I can pick out the hysteresis functions andanything else of interest.
There is nothing in here that is not in thegraphs, but looking at the same thing in different ways can sometimes beuseful.

My final ignition timing swings wildly and Isuspect the graph does not reflect reality, and I'd appreciate your thoughts.

Happy New Year and Stay Healthy
<!--[endif]-->

 

[mskrotzki]

Woops, forgot to add the Color Key
View attachment 168135

 

[MotorMouth93]

Is the EGR flag enabled? It's disabled by default in all binaries that I upload.

The ADX entry looks okay at a glance.

 

[nsx_nordic]

Using that device sounds like a terrible idea. It looks like it tricks the ECU into thinking the EGR is present and functional, but if it's not actually there you'll be running EGR fuel and timing maps without exhaust gases being cycled through the intake which could present a potentially dangerous situation.

The ECU has separate timing and fuel maps for EGR, flipping the EGR system switch in ECU binary disables these maps so engine runs as if it were never there, no additional changes needed, Honda did all the work for us.

Just removing the EGR with no ECU changes disables the system and reverts to the standard fuel and timing maps as well, but also sets a CEL. Perfectly safe to run that way, but then the CEL will be on and you won't be able to tell if/when another CEL-worthy issue pops up later. The only danger here is not being able to tell that the CEL is on for some other issue, so I wouldn't recommend doing this.

Plenty of other Honda engines such as the B-series engines in various Civics and Integras don't run EGR at all and do just fine, the downsides to removing it amount to slightly worse emissions.

Like I said before, the correct way to delete EGR is to chip the ECU and flash a modified ROM with EGR disabled. I've done a few NSX ECUs already so maybe I should start offering this as a service in the future if there's interest.

So thats mean you are able to delete the EGR from the ecu and flash a modified rom with egr disabled ?
Are you able to do it for me? :smile:

 

[mskrotzki]

Is the EGR flag enabled? It's disabled by default in all binaries that I upload.

The ADX entry looks okay at a glance.

Yes, the EGR System & Valve Lift Sensor Switch has the Set Box checked.

 

[MotorMouth93]

I'll check the disassembly to make sure I didn't make a typo for which bit it is.

The other option is that the car just isn't engaging EGR due to some parameter requirement not being met. The range where it can operate is pretty narrow.

 

[mskrotzki]

I'll check the disassembly to make sure I didn't make a typo for which bit it is.

The other option is that the car just isn't engaging EGR due to some parameter requirement not being met. The range where it can operate is pretty narrow.

Thanks, John. I'm looking forward to getting that switch working.

In the meantime during this miserable COVID-19 hibernation I’ve been analyzing over 250,000 lines of logging information. First take-away: Most of it is incredibly boring. Second: tuning is way more of an art than a science. With the complexity of what happens in an internal combustion engine, this should not have surprised me.

In any event, I’m attaching a 5300 line excel spreadsheet made up of selected lines from my logging as well as three of Jeremiah’s logs he uploaded to Google docs. http://www.nsxprime.com/forum/showthread.php/211416-Tuning-the-NSX-3-0L-Using-the-OEM-ECU/page5 post 114. The bin I’m using is NSXRDXNOV2020.bin and the XDF is nsx_v1_alpha.xdf. Using the color codes from the original logging graphs I’ve color coded cell entries that are interesting: Changes in AFR – either lean, a little rich, somewhat rich, crazy rich; pulse widths as they become significantly large and other factors. Both RPM and Load are gradient colored, allowing me (and you) to shrink the sheet to 10 or 12% and look at long stretches of logging and then hone in on interesting spots. Included on the spread sheet is a complete color code. The cell colors are controlled by the conditional formatting function in the styles tab on the home page of Excel. These are easily changed. Here is the analysis:

https://docs.google.com/spreadsheets/d/1AlcKv7z0LZO22Hq9tvNIYBtyNMLEBDip/edit#gid=2076496531


There is no new information generated beyond what you can see in the TunerPro logging graphs – as demonstrated on lines 150 – 230, where I have placed the matching TP graph next to the Excel lines. For me, at least, the spreadsheet smacks me between the eyes more powerfully. I think it also gives insight into transients that can be ignored. I really don’t know just how accurate the sensors are, so I look at all of this with a degree of skepticism.

I also entered three of Jeremiah’s 8/3/2020 logs using 3%, 5% and 8% alterations. These are early logs, and much has been improved since. The limited 5% table, while the smallest, looks the best. The greatest take away from this preliminary analysis, is the PE Map #3 is somewhat rich, but care must be taken to determine transients. I want to go out later today into the mountains and climb some 5 to 7% grades at a steady RPM and see where (if) the AFRs smooth out, and how long it takes. It is difficult to determine a lot on level ground without going too fast too soon.

After investigating more about injector latency issues, hysteresis settings, interplay of barometric pressure, intake air and coolant temperature and hearing comments from others, I am going to start altering entries in the PE Map 3 fuel tables, slowly and with extensive notes. Then I’ll see what, if any, changes can be detected in these logs and go on from there. I’m still in a simplified mode, as with a brand new, not-yet-broken-in engine, I have not gotten into either Map 4 or 5, the WOT and VETC maps.

The final step will no doubt have to be chassis dyno testing and tuning. I’ll keep you posted.

 

[MotorMouth93]

I checked the ECU ROM code again and the bit for EGR is correct, if you aren't seeing it enable while driving then it's either not using EGR for whatever reason or the ADX is setup incorrectly.

 

[mskrotzki]

Street Tuning Update

Time for a short update on my street tuning adventure for anyone who is interested. I have made several street tuning runs on Sante Fe Blvd between 8th Avenue and Dartmouth Street in Denver. It has an advantage of making a U-turn at Dartmouth on the left turn signal and being able to do WOT run returning home. (I can also check for cops on the way down :smile:. I can stop on the way back and check that I got the data and save it before returning home. I won't reveal how many times I got back to my office only to realize the run did not get recorded and saved properly. The whole time, garage to garage is somewhat less than 30 minutes.

I started out the project with [MENTION=36380]Moto[/MENTION]rMouth's bin file for RDX injectors which I believe he made by upgrading the factory bin by ~7%. Here are the results of those first runs


The red triangles are AFR and the blue ones are PW. The yellow line is 14.7AFR and the orange one is 12.75 AFR (or at least they are supposed to be.) The data is only for Table 3, Power Enhancement, and there are about 750 data points plotted. The top chart is the points as a function of RPM and the bottom one is the same data plotted as a function of Load. As you can see the AFR goes crazy rich at higher RPMs, approaching single digits. We want to be somewhat rich, but not this much.

Using this information I made a series of 6 iterations (some of which even got properly logged) and finally wound up with this set of data:


As you can see, the AFR is getting close to a straight line right at 12.75. It gets a little richer around 6000 RPM, where it is about to change to Fuel Table 5, VTEC - WOT. With only ~800 miles on my new engine, and still running on dinasaur oil, I have not ventured into this territory yet. I'll change to Mobil 1 at ~3000 miles. The AFR clusters around 3300RPM are of little interest, although I have made a bin file trying to make some of them richer -- I have not made a logging run to see the effect.

Comments/critiques are welcome. I hope this may be of interest to other Prime members.

Mark

 

[mskrotzki]

Uploaded my Application to Analyze TunerPro Logs

I've just uploaded an application to [MENTION=33247]MotorMouth93[/MENTION]'s Google Docs Page that quickly analyzes TunerPro logs. You can get it here:


https://drive.google.com/drive/folders/1JIpy1YxEH27n2_a4UJa7LrPwDgSgEBFc?ths=true


It is a file in folder "TableAnalysis" called NSX Street Tuning Setup.exe.


It takes the csv file TunerPro produces and puts it into Excel -- after which it produces charts that you can manipulate to your heart's desire.


It puts 2 icons on your desktop, one leading to the application, the other a pair of documents describing it, along with a quickstart.


Let me know what you think.


Mark

 

[Honcho]

I've just uploaded an application to [MENTION=33247]MotorMouth93[/MENTION]'s Google Docs Page that quickly analyzes TunerPro logs. You can get it here:


https://drive.google.com/drive/folders/1JIpy1YxEH27n2_a4UJa7LrPwDgSgEBFc?ths=true


It is a file in folder "TableAnalysis" called NSX Street Tuning Setup.exe.


It takes the csv file TunerPro produces and puts it into Excel -- after which it produces charts that you can manipulate to your heart's desire.


It puts 2 icons on your desktop, one leading to the application, the other a pair of documents describing it, along with a quickstart.


Let me know what you think.


Mark

What a fantastic tool- thanks Mark! I've seen this in person on a laptop and it is quite useful for tuning. BTW, you image links aren't working.

 

[mskrotzki]

Those following this thread know that since I got my new SOS total engine rebuild I have not been able to enter VTEC. I’ve traced nearly every VTEC electrical circuit, disassembled and cleaned my spool valves, checked oil pressure and all the rest of the troubleshooting suggestions in the Service Manual. I have completed and distributed a handy side-of-the-road Excel application to analyze TunerPro logs for on the fly adjustments (not that I have made any yet). I have reread about all the information I have found on PGM-FI. I have pulled the Demon 2 emulator out of the ECU and reverted back to running on my original RDX chip. Yet there is no joy in VTEC land for this NSXer.

Going through the binary file that runs our ECU it appears that, among other parameters, for VTEC to engage the load calculation needs to be above ~81%. I’ve noticed for some time that I’ve never seen a load above ~80% in my location in Denver at 5300’ ASL. This morning I drove up toward Mt Evans to gain significant altitude, although the road up to 14,150’ is not yet open. I recorded pulls at 9000’ ASL and 10000’ASL. At 9000’ Load maxed out at ~66% and at 10000’ Load was ~63%.

If I understand it correctly Load and Volumetric Efficiency computations revolve around the amount of mixture that enters the combustion chamber relative to the theoretical amount available for said entry. The result is reported as a percentage, not an absolute compared to standard sea level. This HAS to be dependent upon altitude (barometric pressure) and there HAS to be a correction applied when the engine is 2 miles above sea level. I have been unsuccessful finding any correction factor in the bin file, but I understand there is quite a bit of the total code inaccessible to mere mortals (outside the Honda pantheon.)

In the 100,000 miles I had driven this car prior to blowing the radiator and rebuilding the engine VTEC ran perfectly even at the top of Mt Evans, all 14,150’ of it. Is there a sensor missing, or misreporting? Is the Baro chip on the ECU inactive? Running on the original chip in the original configuration eliminates a software glitch in my mind, but am I overlooking something?

Any and all thoughts and pointers welcome. I want to get to the bottom of this.

Thanks,

Mark

 

[MotorMouth93]

The "alpha" RDX binary is based on the first RDX tune shared by jvtec95 that was having issues staying in open loop, I tweaked the open loop parameters so it works a bit better, it's not just scaled stock tables as that wouldn't work at all. It's not a great tune by any means but it does work.

At this point, if you're willing to swap in stock injectors, I can send you my spare and unmodified ECU and recently serviced stock injectors to test and see if it's a hardware problem. DM me here or on Facebook and we can arrange shipping if that's something you want to try.

What you're seeing as far as load goes is about expected with the altitude. The "calculated load" the ECU uses is just manifold pressure that's been converted to a 0-255 value roughly representing 0-15psi. At 10,000 feet, maxing out at 63% is about right. One of the nice things about a MAP sensor is that barometric pressure compensations are built in, the ECU has a separate concept of "high" and "low" altitude that tweaks a few things but for the most part that's irrelevant to our tuning and barometric pressure compensations are handled by the MAP sensor.

VTEC activation isn't dependent on load as far as I know, it's mainly coolant temp, RPM, and a road speed threshold so you can't bang off the limiter in VTEC while standing still, where did the 81% come from? What is that parameter called? I can look into it in the actual program code and see what it does.

Edit: Oh and make sure to check the small plug on the opposite side of the ECU as the main plugs, those are the VTEC pressure sensor lines so if the connector is loose that could cause issues.

 

[mskrotzki]

Thanks for your reply. You are the guru of this thread.


I have the stock injectors (not cleaned) and chip, and could certainly swap them in. Changes to my current ECU are simply the addition of the pins to accept the leads from the Demon 2. Fur sure, the ECU could have been damaged with all the manipulations in and out of the car.


The 81% is the maximum Load I've seen here around Denver -- again what would be expected at 5300'.


Here are the TPS/Load calculations from the bin:





I'm not exactly sure what these tables mean, and hope you can give some guidance. I only I could drive to significantly lower elevation in less than 1/2 a day, I could get a better handle on this.

The small plug on the ECU nearly adjacent to the main relay are the points i have checked continuity to the spool valves. I is not loose (unfortunately).


What confuses me is I have pulled the Demon and replaced it with my original ProSpeed chip which I ran for at least 40,000 mile with no VTEC issue with the RDX injectors. Changing injectors to stock (dirty) injectors with the stock chip wouldn't seem informative. However running another ECU with the ProSpeed chip could eliminate/prove an issue with my current ECU. I wouldn't take more than 5 minutes to get that answer. I'll ask Paul and Spencer again for a very short term loan.

 

[MotorMouth93]

With the issues you've been seeing I wouldn't be surprised if something was damaged during the chipping process, hence suggesting a different ECU. If your ECU is indeed damaged I wouldn't expect any different behavior just swapping back to the Prospeed chip. From the ECUs perspective, the Demon is no different than the stock chip or Prospeed chip. The ECU asks for the contents of a memory address, and the chip (or Demon) places the data on the bus, and that is the extent of the interaction. Think of a chip like a hard drive in a computer, and think of the Demon like a network storage drive. The chip can only be accessed by the computer it's in (ECU), but other devices can also access and update data on the network storage device (your laptop with TunerPro).

Oh you're talking about the alpha N load calculation tables, those are only used under certain error conditions, such as a MAP sensor failure. Under normal circumstances they aren't used at all so I don't think they (or elevation) are the problem here. I've forcibly enabled alpha N before as an experiment using some special code patches I wrote and you'd know it, the car runs quite poorly and EXTREMELY rich.

You might try shutting off the EGR system as a test, in your logs I've seen the EGR flag pop up at weird times so I wonder if there's some EGR parameters that might not be set up properly to account for the injector change. It's kind of a shot in the dark but worth trying just in case there's some overlap in EGR engagement and VTEC that I haven't come across myself.

Load in the low 80s is about right for 5300 feet.

 

[mskrotzki]

You might try shutting off the EGR system as a test, in your logs I've seen the EGR flag pop up at weird times so I wonder if there's some EGR parameters that might not be set up properly to account for the injector change. It's kind of a shot in the dark but worth trying just in case there's some overlap in EGR engagement and VTEC that I haven't come across myself.

I'll definitely shut off EGR and see what happens in the next 24 hours and keep you posted. Thanks

FWIW, I've noticed that xdl files exported from TunerPro as csv files start sometimes start out with the VTEC flag set to ON. This has occurred on Jeremiah's files, and I think on yours also. On these files the VTEC flag gets correctly set during the run, frequently after venturing into PE territory. I have no idea what to make of it. Doubt if it is terribly important.

 

[Honcho]

I'll definitely shut off EGR and see what happens in the next 24 hours and keep you posted. Thanks

FWIW, I've noticed that xdl files exported from TunerPro as csv files start sometimes start out with the VTEC flag set to ON. This has occurred on Jeremiah's files, and I think on yours also. On these files the VTEC flag gets correctly set during the run, frequently after venturing into PE territory. I have no idea what to make of it. Doubt if it is terribly important.

Sounds like a bug in the VTEC code. But, Jeremiah and John both have working VTEC, so if anything it is just an annoyance.

Taking a flier here, but have you checked your neutral switch and its connection? The ECU will not allow VTEC if it thinks the car is in neutral. Might be something else to check...

 

[mskrotzki]

Sounds like a bug in the VTEC code. But, Jeremiah and John both have working VTEC, so if anything it is just an annoyance.

Taking a flier here, but have you checked your neutral switch and its connection? The ECU will not allow VTEC if it thinks the car is in neutral. Might be something else to check...

I am going out right now to (find and) check the neutral switch, shut off EGR and see what happens. I'll report back RSN.

 

[mskrotzki]

Drove out 6th Avenue for a few miles with EGR turned off in the bin file; no change in lack of VTEC. Drove back with EGR turned back on and, of course, no changes. FWIW, Driving back at least 25% of the trip was in EGR at nearly perfect 14.7 AFR. Analysis shows that EGR never turned on.

At this point, if you're willing to swap in stock injectors, I can send you my spare and unmodified ECU and recently serviced stock injectors to test and see if it's a hardware problem. DM me here or on Facebook and we can arrange shipping if that's something you want to try.

Testing the other day with my old chip in and the Demon 2 out, I think, takes the Demon 2 out of the equation -- there is nothing wrong with my Demon. John, does your spare ECU have the 4 pins required to install a Demon? If it does, then I could simply install my Demon without messing with the fuel system. If I get VTEC heaven then it is my ECU, without doubt. Let me know what you think.

Mark

 

[MotorMouth93]

My spare ECU is not chipped and I don't plan on chipping it, so the Demon or any aftermarket chip won't work with it.

 

[mskrotzki]

Test Harness pin equivalents to ECU pins

With all the messing around I have done chasing my VTEC difficulties, I believe the following pdf could be of help to people now and in the future. What I do know about C 480 I believe to apply to Connectors 154 - 157, but use at your own risk.

View attachment PHYSICAL PLUG CORRESPONDENCE.pdf

 

[mskrotzki]

Well, the damndest thing has happened. We took a road trip this past week to Gunnison, Durango, Chaco Canyon, Grants, NM and Albuquerque. Prior to leaving I was ~150 miles short of my third oil change to Mobil 1, so I changed it early. During the first part of the trip my VTEC started working-- and working very well. I never remember it pulling so hard -- especially at the altitudes on this trip. Aside from the changed oil and any new detergents introduced that freed up a clog in the valve train, I have no explanation. I got some 5 - 2 codes; rear gallery oil pressure not returned. There is a pressure switch, normally closed to ground that opens when sufficient pressure is sensed in the gallery. What the ECU does with this information is unknown to me. It seems like if might be a safety device to switch the engine between the power enhancement and VTEC tables, but I don't know. I do know that the engine pulls like mad all the way to 8K and sounds terrific. I was also getting a 1 code -- front O[SUB]2[/SUB] sensor CEL from time to time. I am staying out of VTEC until I figure this out, as it might have the rear bank out of VTEC while all is well in the front bank. It feels fine, but better safe than sorry.

My wide band O2 sensor decided not to respond and I decided not to mess with anything while 500 miles from home. I am anxious to clear up the 5 - 2 code and the WBO2 sensor and see what is happening with my fuel tables.

 

[Honcho]

Well, the damndest thing has happened. We took a road trip this past week to Gunnison, Durango, Chaco Canyon, Grants, NM and Albuquerque. Prior to leaving I was ~150 miles short of my third oil change to Mobil 1, so I changed it early. During the first part of the trip my VTEC started working-- and working very well. I never remember it pulling so hard -- especially at the altitudes on this trip. Aside from the changed oil and any new detergents introduced that freed up a clog in the valve train, I have no explanation. I got some 5 - 2 codes; rear gallery oil pressure not returned. There is a pressure switch, normally closed to ground that opens when sufficient pressure is sensed in the gallery. What the ECU does with this information is unknown to me. It seems like if might be a safety device to switch the engine between the power enhancement and VTEC tables, but I don't know. I do know that the engine pulls like mad all the way to 8K and sounds terrific. I was also getting a 1 code -- front O[SUB]2[/SUB] sensor CEL from time to time. I am staying out of VTEC until I figure this out, as it might have the rear bank out of VTEC while all is well in the front bank. It feels fine, but better safe than sorry.

My wide band O2 sensor decided not to respond and I decided not to mess with anything while 500 miles from home. I am anxious to clear up the 5 - 2 code and the WBO2 sensor and see what is happening with my fuel tables.

Mark-

What an interesting development! The code 5-2 is indeed for the Rear VTEC oil pressure switch. If there was debris inside the switch port preventing oil pressure from reaching the switch, then the ECU would definitely keep you out of VTEC, since it would think that the pressure changeover did not occur. I think you're on to something with the debris theory- The oil ports that feed the rocker shafts/VTEC pistons are quite small and even a small obstruction could cause a problem. In any event, I'm glad you've got VTEC working again. I wonder if it will show up properly in your logs now?

You should re-calibrate your wideband using the specified method (basically let it hang in open air and turn it on). You need to do that every few months to keep the accuracy.

 

[mskrotzki]

I went out this morning to check logging, and my WBO[SUB]2[/SUB] sensor is still not working. I'll take it out to let it "air out" this afternoon and see what happens. The car still went into VTEC with no trouble and only the 5 - 2 CEL on a pull through 2nd gear. It was fine in 1st gear -- likely not enough time in VTEC to register the CEL. There was no error code for the front O[SUB]2[/SUB] sensor. I'll check all the wiring and the pressure sensor on my rear spool valve assembly. Would I be able to borrow your rear bank spool valve for a quick test if it is readily available (on your workbench)?

I forgot to mention that in three fuel tank fill ups, I got 31.29 MPG (run to Durango on 2-lane highways), 29.27 MPG through Chaco Canyon (1st and 2nd gear for over 50 miles (if I ever go there again I will rent a car -- truly atrocious road)) and 29.17 MPG with mostly interstate driving @ ~85 MPH. I remember getting those figures what I first got the car but never again saw them after using Brian's tune for the RDX injectors. Now that I have reset the fuel tables to suit my particular car, they are back again.

 

[MotorMouth93]

There is a specific wideband calibration process that needs to be done, it's pretty easy just make sure you do the right steps. Its in the LC2 instruction sheet.