plot pulse width, afr map, mivec, and boost and see what that looks like. If pulse width follows afrmap then I would suspect what you are seeing is a rapid change in VE that requires a lot more fuel.

 

I've got one of those, bar the MIVEC logging - I'd switched that off again to keep the speed up. MIVEC actuals won't be significantly different to the previous graph though.

Here...





IPW does indeed follow AFRMAP.


I must say, this is starting to look like a half-decent setup!...

Spool-up mixtures look healthy. A little rich, but I'm guessing LTFT of +1.8% is mixed into the equation early on, before MAF Hz gets over the "LT Trim Control" threshold. Not messed with that area much yet, though.

Lean-spool is now starting a progressive enrichment right at the top of 3rd gear. It starts its work at about the time I'm hitting 7000rpm in that gear. edit: Oops, I was mistaken there - it's actually following the fuel map just fine!


Right now it's got a max enleanment of 0.4 to be "dialled out" over time. I can fine-tune it as I want, and balance fuel map with LS offsets until I'm happy.

First pull AFR is around 11.6:1 until 6500rpm.

On the track, I expect AFR to be around 11.2:1 once LS goes to sleep.

Timing is at 7* at peak, and touching 22* at redline.


It must be getting cleaner, as I feel the need for a gratuitous VD pic...





Definitely getting there! I'm now making only very small changes to the fuel map to fine-tune little dips, like the one at 7100rpm. So... OCD stuff.


Cheers!


Rich

 

did you try shifting the LS end rpm up from 6500?

 

I did, and it worked a treat.

In the previous ROM test, AFRMAP went 0.4 points richer in one big hit around 6500-6600.

Now, with "LS-end-rpm" moved up out of the way, AFRMAP doesn't drop off that cliff any more.

Note: I do have a mild top-end enrichment built into the fuel map anyway, so if LS starts to unwind at the same place, it'll look quite pronounced when viewing "first pull" results.

I'm keen to see how well-behaved this is after a lap or two on a track.

Rich

 

Here's a comparison graph before applying that LS end rpm tweak...




That big AFRMAP drop (at the highlighted point) no longer occurs.


Rich

 

Are you running with Lean Spool on now?

 

Yep, I am...

• "Lean Spool Stop RPM" now bumped up from 6600 to 8000.
• "Lean Spool AFR v Fuel Map AFR" dialled back so it only enleans by 0.4AFR (instead of the stock 1.3AFR).

When I posted earlier, I thought I was seeing LS start to wind back and enrich at the top of 3rd, but I was mistaken. AFRMAP has been consistently "fuel map + 0.4".

Street driving only so far. I may tweak stuff once I get some track logs.... but right now I'm fine with it enriching by 0.4 after I start thrashing it a bit.

Rich

 

Here's an update...

I've finished tuning using wbo2 " at face-value", and it's running beautifully.

If there's some MIVEC shenanigans causing an artificially lean reading, so be it. If it's MIVEC "needing" more fuel, I've got it covered. Maybe I've got it right, or maybe I've created "Rich Spool".

Either way, it's safe, the turbo spools up really fast (helps being a Ralliart with a TD04) and I can sleep well at night.

Here's 3rd gear WOT from 2250-6000...





And 4th, from 2000-5400rpm...





Great stuff...

Rich

 

I think in all of this you really just proved that there is a very slight lag in response from your lc-1 sensor. It does take it a bit of time to respond to the changing exhaust temp and flow rate. You could go really OCD and plot the response curve of the WBO2 sensor for all of us

 

Yeah, you should see the spreadsheets I have - with entries that perform AFR data "offset" shifts by X cells. I reckoned there was a 0.7s delay on my first serial wideband link, a 0.4s delay after changing the USB-serial converter, etc.

But it didn't really turn out to be usable. While it looks like there is a big offset down low, I got spool-up/peak-boost knock if I tuned with an assumed delay of any significant magnitude.

And in the upper RPMs there doesn't seem to be any significant delay at all. Check out the throttle-off AFR response at 7000rpm a few posts back - it's bang-on.


These days, my AFR data is via analog, direct to the OP2 pin 8, with the LC-1 set to "1/12 sec" mode (averages its 300+ internal samples down to 12 a second), and my OP2.0 grabs 13-14 samples a second. So my data-side setup is as good as it'll ever be.

And if the high-rpm response is immediate, then that probably means the LC-1 response overall is always immediate - and we simply have the gas-speed and sensor-position factors to contend with.

Rich

 

The engine is a massive air pump. The faster it moves, the more air it pushes. Because the exhaust pipes are a constant size, the velocity in the pipes increases with volume. Therefore, higher RPM means less wideband lag.

 

An update on this whole saga...


After further testing, I'm now 100% sure that AFR readings from my Innovate LC-1 tend towards the very-lean side during, and immediately after, spool-up. When mixtures are going richer fast, the measurements take quite a while to catch up.

It doesn't matter if you use analogue or serial link, but serial logging suffers an additional delay on my low-powered, USB-only netbook.


Here's what you get if you believe the LC-1 and tune directly from its output...


While the engine was "okay" with that, it was not all that far from factory-rich in spool-up areas... it then leaned out a lot for the higher loads/rpm areas.


After trying this out, I wasn't all that happy with it - the sluggish spool-up made the car feel rather heavy. It also had the tendency to get some knock on spool-up when driven hard.



The LC-1 obviously needed "steady state" tuning - fixed RPM, fixed load, cell-by-cell. So I had a choice of booking in for a pro dyno tune... or solving the problem myself.

I chose the latter. Well, technically my wife chose it, by telling me I wasn't spending $$$ on a pro tune! "You've bought all of that tuning stuff - go do it yourself!"


So... here's GhettoDyno(tm), brake-drag style...

Step 1a. Set Base Wastegate Duty Cycle map to 0%.
Step 1b. Find a straight, empty road, and go WOT in 3rd at 2500rpm.
Step 1c. Left-foot-brake to hold revs at 2500rpm until boost stabilises.
Step 1d. Now modulate left-foot-brake, to go gently up through the revs.
Step 1e. Let brakes cool!

Review the LC-1 logs, and revise fuel map accordingly. Then repeat...

Step 2a. Set Base Wastegate Duty Cycle map to 20%.
Step 2b. Find a straight, empty road, and go WOT at 2500rpm.
etc.


Repeat with BWGDC at 40%, then 60%. Or whatever iterations/WGDC curves needed to get the job done.



By holding revs at a midway boost level, it allows the LC-1 readings to stabilise - and become meaningful.

It's the only way I've found to get reliable measurements from this thing.


I was careful not to overheat anything while doing this - my brakes were perfectly fine afterwards. Used them on-track the very next day - no issues.


Note also, I raised my "starting" rpm along with my fixed WGDC... so I didn't ask my poor engine to hold higher boost fixed at 2500rpm!


Here's the final result...



This felt a lot more spritely on spool-up... as you'd expect!


I hope someone finds this interesting. Or funny. Feel free to laugh it up... I'm just happy to finally put this whole debacle to bed and get a sharp-feeling tune in my car!


Rich

 

I saw something in my MAPCalcs/MAFCalcs logs today, and thought of this old thread.


When I posted most of this stuff up, I was trying to get my head around why "steady state" fuelling at a specific RPM/load cell in the fuel map seems to be richer than that exact same point during spool-up.

And... I think I've just found the answer. It's all in the way the ECU calculates final load.

Check it out... during spool-up, there is a noticable LAG between MAFCalcs load and final load.



If you hold steady at that rpm/load point and take a fuel measurement, the load will be steady. But if you race through there on a fast-spooling turbocharger, the FINAL calculated load will be quite a ways behind the key inputs going into the calc.


edit: By the way, LoatTiming is logged first, near the top of the data reading cycle. MAFCalcs and MAPCalcs are logged towards the end of the data cycle.


Rich

 

subscribed. going to read this thread carefully on the plane today.

 

Also, there is a noticable difference in the way Load rises depending on how hard you accelerate. Mild accelerating of <50% will show MAP=MAF, >50% will show MAP>MAF; shifting and lift-off is where life gets hectic. Right now I have 1 trouble spot, shifting from 1 to 2; if I accelerate really lightly and lift even a little bit I go lean for a fraction of a second (worst ever was 17, typically only 15-16).
Shifting for all other gears will only lean out 0.5 max (progressively less the higher the gear), until I crossed over Open Loop at which time all is well in the land. Logs show it's due to a sudden spike/plummet in MAF Voltage.
SD looks better and better every day
I'm now MORE than halfway on the fence about trying to max out my MAF