Edit: Adding TLDR since this ended up obscenely long
- Idle issues mostly fixed
- Start issues mostly
- AC fixed
- Off idle lean condition fixed
- Cold starts are still too rich, but it does start right up every time
- Cruise AFRs are very consistent
- WOT needs more logging but is mostly okay
- Still no VTAK
I fixed the hot start and AC problems, and the solution was really simple and kind of funny.
The idle control on these cars is more complex than you'd think, its a convoluted combination of IACV pulsewidth manipulation, ignition timing control, and fuel table switching and I had forgotten about the fuel table switching portion. Under certain conditions the car uses the open loop power enrichment fuel table, and for some reason it decided to use that for hot starts, whenever I tried to turn the AC on, and occasionally at seemingly random intervals. Since I have O2 feedback disabled and disabled power enrichment I hadn't really messed with that fuel table so it mostly just held garbage. So it would switch to the garbage tables then die for obvious reasons. Occasionally it would just idle really rich instead of at 14.5 like I have my tables set up for and looking at my old logs, it was because it was going into open loop idle there as well. Copy/pasting my tuned fuel table into the open loop table fixed these issues 100%.
I'm also working on some code tricks to manipulate the factory idle control algorithm into allowing a good amount of base idle air flow while still using timing to pull idle down to 800-900 when hot. FSM states idle should be set at 600rpm with the IACV unplugged, and I have mine set at about 1100, but the car will happily hot idle at 900 with idle timing control now.
Cold starts are still too rich, and I suspect this is because the fittings connecting my vacuum manifold to each intake runner are too small. They have maybe ~5mm ID so the amount of air the engine can suck through them is limited. It runs too rich when cold because the IACV can flow more air than the engine can consume at idle which artificially raises the measured manifold pressure, which raises calculated load, which results in too rich AFR. I made a restrictor to try to combat this issue and it sort of helped but I might need to restrict flow even further to make it work in its current configuration, I might be able to combat this issue in software for now but when I switch to the built motor I'm going to increase the size of the vacuum fittings to fix it properly, and maybe drill additional ports into the runners to make the idle air system completely separate from where I read manifold pressure.
Something to note is that the car will now happily free rev up and down slowly in neutral even with the lightweight flywheel. Before I'd really started tuning the low load areas of the fuel table with my tuning scripts it would have weird oscillations in RPM - I suspect due to the poorly tuned fuel tables causing wild variations in AFR and therefore torque causing RPM to wildly oscillate without touching the throttle.
As far as drivability goes, the leanness off idle is kind of working itself out as I log and tune more, I think it was just an issue in that section of the fuel table. I also tweaked the decel fuel cut parameters to kick in less often and less aggressively which has made the on/off throttle transitions much smoother as well. AFR control is pretty tight, there's a few places that need more fine tuning but its overall quite good. There was also an issue with the way I was injecting my own calculated "ITB load" value that would result in the original MAP-based load value being used occasionally and the engine would kind of "hiccup" a bit and the load would spike in my logs. This was due to a CEL-related function in the ECU that checks and clears error flags if the associated problem is no longer present, since the Alpha N flag is only present when there's a MAP error, that routine clears the AN flag and reverts to standard load calculations since there is no MAP fault. The way I handled this originally was to call that CEL function then immediately set the AN flag again since it just got cleared, however this left a small time delta between the time that function cleared the AN flag and the time my code could reset it, and if a crank signal interrupt came in during that time, the next load calculation would use the MAP sensor instead of the AN load value. The CEL function is baked into the CPU and can't be modified, but I was able to do some fancy things in my own code to keep all CEL functionality intact as well as remove that critical section.
Here's some log plots showing the current state of things. (yellow=AFR, dotted yellow=timing advance, pink=RPM, light blue=calculated load, red=throttle, green=manifold pressure) There's some spikes in the manifold pressure reading due to the way my logging code works but those aren't real.
This one is gentle cruising at 45mph or so. AFR holds very steady around 14.7-15, the variation is actually much less than you'd get in a stock setup with O2 feedback turned on as it will hunt up and down between low 14s and low 16s by design. The huge spike in AFR is decel fuel cutoff. The dip down to below 14 is slightly over-aggressive tip in fuel which I haven't really messed with yet other than cranking it up enough to eliminate huge lean spikes. This is a great example of the downside of purely AN fueling, miniscule TPS changes can cause large changes in manifold pressure and there's just no way to tune that well with only AN.
Here's one showing a near WOT pull on an on-ramp. AFR is more jagged here since I haven't tuned these regions of the table very finely yet and is also a bit too rich. The vertical white line shows the exact point my load calculation changes from MAP-based to TPS-based and as you can see the change is nearly perfectly seamless. On the left you can see coming off of decel fuel cut, as load rises my target AFR drops from mid 14s for cruising at low load to under 13 at high load. You can see some very fast lean spikes during shifts, high throttle tip in fueling still needs to be turned up more.
This log just shows some varied load conditions along with how the ECU responds, for the most part, steady state fueling is getting there, so soon I'll start working more on tip in settings which are mainly what the spikes are here.
This plot shows some moderate acceleration. Tip in is a bit too aggressive in these ranges as you can see by the AFR dips when getting back on the throttle after shifts. I might also lean out the fueling in these regions of the map since 13 might be a bit richer than needed here. Over on the left you can see a start and AFR leans out a tiny bit to mid 15s but it used to routinely hit 17-18 so its a huge improvement. The revision of my tune after this one stays in the 14s right off idle.
Overall, its driving really well, AFR control is much tighter now than it was with a [shop that I will not name] RDX injector tune on the stock manifold. I even hit the cruise control a few times to see what would happen and it works, sort of? If you're on the highway it will generally work okay but low speed cruising at like 40-50mph it will basically just alternate between 0% and 10% throttle instead of finding that 2% steady state point and you'll speed up and slow down ever so annoyingly. The throttle is still pretty touchy so I'll be updating the linkage again later on. If I change the resting angle of the arm the throttle cable connects to, I can gain some mechanical advantage at low throttle in in exchange for a slightly stiffer pedal at WOT. I'm planning on daily driving the NSX this week while I overhaul the rear suspension on my 540 so we'll see how that goes, should get lots of part throttle log data to fine tune even more.
If things continue to improve at this rate there is a very real possibility that steetable ITBs will be a virtually bolt on modification for stockish C30s in the near future. The ability to install an ITB kit, chip the ECU, flash a basemap that actually runs, then just make minor changes is virtually unheard of on any platform. I don't think I'd try this on any kit but the one @
A.S. Motorsport sells (also the billet manifold are s3xy compared to the cast options) though as the stock injector placement is critical, it allows the factory injection timing algorithms to work properly and as far as I can tell there isn't really a way to change them so trying to run a Toda or Hayward kit on this ECU might present problems because of that.
Oh, I'm using the RDX injector dead times that [MENTION=16951]sparky[/MENTION]
posted here 6 years ago as "RDX ACS 1.2" with excellent results so far.
All that said, I must give credit where it is due, since I'm just standing on the shoulders of giants. The countless hours I've poured into this over the past 3 years were only made possible by the work that sr5guy did and made freely available almost 7 years ago to the day. Without his extremely detailed ECU definition none of this would have been remotely possible.
. I have tried ordering a few different plastic clips for these rear interior pieces as several of mine are missing and broken but can’t find the right part number or place to order from. Any advice is greatly appreciated. 
