Honda / Acura K20a K24a Engine Forum banner
1 - 20 of 24 Posts

·
Registered
Joined
·
23 Posts
Discussion Starter · #1 ·
I have Ktuner running on a 2008 TSX ECU (37820-RBB-A71) speed density tune, engine is a K24A8 (economy Vtec) small turbo and 8psi internal wastegate.

when in open loop under heavy load (70%+ TPS ) any time I go over 8 PSI exactly it starts switching rapidly between fuel status 4 (WOT) and fuel status 1 (open loop fault? or is that cat heating?) as shown in log screen shot. if I remain say at 7.9 PSI it's fine and AFR and timing are steady.

when it switches to fuel stat 1 it reduces duty cycle by almost half and increases timing, it runs lean and stutters losing power. also short term fuel trim starts showing a correction in status 1, also noticed that AFR CMD changes slightly but not to the point where it needs to cut duty cycle in half!

I have same values in WOT and WOT cat heating tables

ECT, TPS, MAP and everything else remains steady. only changes happen to fuel status , duty cycle, timing , afr cmd
and short fuel trim.

I thought it was spark blowout so I gaped down to .028" (0.7mm) with no change

any idea what is happening here? what are the triggers for fuel status 1 ?

Rectangle Black Slope Font Parallel

what could be causing this?
 

·
Arouse the DAMPFHAMMER!
Joined
·
9,422 Posts
I have Ktuner running on a 2008 TSX ECU (37820-RBB-A71) speed density tune, engine is a K24A8 (economy Vtec) small turbo and 8psi internal wastegate.
Thanks for the information. I know what you mean, KTuner and FPro states the options MAF or speed/density, but in reality this is a simple injector duration map depending on engine speed and load, which makes a huge difference when you change parts, influencing fuel flow or air flow. In the background the Honda ECU is the still operating like mentioned, one time tune = acceptable, if you change anything = review of the calibration needed.

That and some other methods, coming with the Honda ECU, has to be considered. E.g. the WOT is triggered by an simple value, either MAP or TPS. I assume you use TPS, which is a bad idea using a DBW TB. Do you have an Tuner with access to your calibration?
 

·
Super Moderator
2002 DC5 Type S
Joined
·
2,142 Posts
I assume you use TPS, which is a bad idea using a DBW TB. Do you have an Tuner with access to your calibration?
this is the meat of it ...
 

·
Registered
Joined
·
23 Posts
Discussion Starter · #4 ·
Thanks for the quick reply.
I have the file, I'm working on the tune myself, I have tried both ways with WOT engagement and it did the same thing. I'm always using only MAP for WOT engagement, I had it setup to engage WOT right at 2PSI as it's a small turbo and boost comes on early and fast.
I asked Ktuner on their support forum and they say they don't change anything about fuel status code in the ECU.
 

·
Arouse the DAMPFHAMMER!
Joined
·
9,422 Posts
I'm always using only MAP for WOT engagement, I had it setup to engage WOT right at 2PSI as it's a small turbo and boost comes on early and fast.
That's far to high for WOT, at 75-85 kPa you need to activate WOT, this is around -2.9 psi in your scaling method.

On a boosted system you need to look what time scales you have in your system once you setup control parameters, who affect torque significant. The lower WOT engagement by MAP will stabilize the operation. Another point is your engine speed depended WOT engagement. Please post a picture of this table here.
 

·
Registered
Joined
·
23 Posts
Discussion Starter · #6 ·
I'll take your advice on lowering MAP WOT engagement point and retest, I kept it this high as I was getting in boost (and WOT) easily at light throttle while cursing in higher gears (small turbo) and I wanted to stay as much as possible in closed loop to maintain decent economy.

can you explain what you mean by time scales?
Attached my closed loop settings
Rectangle Product Font Parallel Screenshot
 

·
Registered
Joined
·
23 Posts
Discussion Starter · #7 ·
So I lowered WOT engagement to 85kpa and still have the same issue, the only difference now is that the fuel status doesn't switch back and forth between status 4 and status 1 like before , but AFR CMD still changes along with duty cycle and ignition timing as soon as I get to 8 psi
 

·
Arouse the DAMPFHAMMER!
Joined
·
9,422 Posts
Attached my closed loop settings
Why do you run closed loop so rich, even at settled ECT's (assuming it to be in °C)? Regarding the WOT MAP Based Engagement I would recommend following:
  • Idle till 80 % of redline 85 kPa and from there to redline to 75 kPa, all at lambda 1,00
  • There is no need to flood the chamber, it is still no WOT. Ignition timing need here to be tuned precisely to MBT with respect to knock: minus your safety distance to knock margin (recommended are 2 or 3 degree)
So I lowered WOT engagement to 85kpa and still have the same issue, the only difference now is that the fuel status doesn't switch back and forth between status 4 and status 1 like before
Ok, fine. So next step.

but AFR CMD still changes along with duty cycle and ignition timing as soon as I get to 8 psi
What are your AFR CMD tables look like?

stock map sensor voltage range at 8 psi and triggering an out of range fault?
That's a good point, which should be addressed for any boosted setup.
 

·
Arouse the DAMPFHAMMER!
Joined
·
9,422 Posts
can you explain what you mean by time scales?
Different systems have different time scales where they act to transient conditions. E.g. if you floor the gas, fuel injection reacts accordingly, faster then the air flow and fuel evaporization, therefore we have tip in fueling strategies to compensate those different time scales.

With time scale is meant a process reacts to a stepped demand from 0 to 1 with some time. In control speech systems get described according their delay-behavior. You may read it already systems get characterized with an delay time the t60, t90, tXX, while XX stands for a number from 0-100 %, a time in which 60 % or 90 % of the stationary new value are reached. E.g. spark has a t90 time of around a few ms from trigger till spark. A lambda probe has a t90 time of 150-800 ms, a race engine turbo has t90 time of 500 ms to 1800 ms, while other can have minutes if they huge enough.

Time scales are an important characterization in control technology as something which need time to come up or down can't be run like something which is quick. Not only PID sets are different, also gradients will be different to keep control cycles in a stable condition.

Understood what time scales mean?
 

·
Registered
Joined
·
23 Posts
Discussion Starter · #12 ·
Why do you run closed loop so rich, even at settled ECT's (assuming it to be in °C)? Regarding the WOT MAP Based Engagement I would recommend following:

  • Idle till 80 % of redline 85 kPa and from there to redline to 75 kPa, all at lambda 1,00
  • There is no need to flood the chamber, it is still no WOT. Ignition timing need here to be tuned precisely to MBT with respect to knock: minus your safety distance to knock margin (recommended are 2 or 3 degree)
Ok, fine. So next step.

What are your AFR CMD tables look like?

That's a good point, which should be addressed for any boosted setup.
closed loop is actually running 14.7 AFR, those ECT are in F and not all columns in the table are showing and the last few cells are set to 14.7 at operating temp, these closed loop numbers are practically unchanged from stock.
 

·
Registered
Joined
·
23 Posts
Discussion Starter · #14 ·
Different systems have different time scales where they act to transient conditions. E.g. if you floor the gas, fuel injection reacts accordingly, faster then the air flow and fuel evaporization, therefore we have tip in fueling strategies to compensate those different time scales.

With time scale is meant a process reacts to a stepped demand from 0 to 1 with some time. In control speech systems get described according their delay-behavior. You may read it already systems get characterized with an delay time the t60, t90, tXX, while XX stands for a number from 0-100 %, a time in which 60 % or 90 % of the stationary new value are reached. E.g. spark has a t90 time of around a few ms from trigger till spark. A lambda probe has a t90 time of 150-800 ms, a race engine turbo has t90 time of 500 ms to 1800 ms, while other can have minutes if they huge enough.

Time scales are an important characterization in control technology as something which need time to come up or down can't be run like something which is quick. Not only PID sets are different, also gradients will be different to keep control cycles in a stable condition.

Understood what time scales mean?
I understand transient response time, but this is not like a tip in stumble, it happens mostly at a steady TPS and boost pressure after about 1 second or so of going in WOT and it continues to happen as long as I'm on the throttle for as long as 2- 3 seconds
 

·
Arouse the DAMPFHAMMER!
Joined
·
9,422 Posts
It's the stock Honda MAP sensor which should be good to 10.5 PSI
Theoretically yes. But the precision of the sensor goes down on the higher values. Therefore my recommendation would be to adapt the sensor and have at least 30 % to max left for better measurements.

I am not sure if this connects to the observed issue as the MAP doesn't block or fluctuate. Maybe we have to summarize:
  • lambda fluctuation solved - issue was WOT engagement to high
  • lambda CMD, Duty cycle and ign. timing still run away or fluctuate when exceeding 8 psi
Seems to be a threshold thingy. Safety modules for Lambda, boost, engine speed, gear have to be checked if there is any boost related limitation.

BTW, did I mentioning KTuner is a bug, one can't share simply the calibration for check, you need a dongle to be able too. If you are unexperienced with it as customer and have no tuner aside with same hardware it is a mess. This will extend the time to solution if I can't just revise your calibration. That's way I don't like KTuners price/product access policy.
 

·
Registered
Joined
·
23 Posts
Discussion Starter · #16 ·
Theoretically yes. But the precision of the sensor goes down on the higher values. Therefore my recommendation would be to adapt the sensor and have at least 30 % to max left for better measurements.

I am not sure if this connects to the observed issue as the MAP doesn't block or fluctuate. Maybe we have to summarize:
  • lambda fluctuation solved - issue was WOT engagement to high
  • lambda CMD, Duty cycle and ign. timing still run away or fluctuate when exceeding 8 psi
Seems to be a threshold thingy. Safety modules for Lambda, boost, engine speed, gear have to be checked if there is any boost related limitation.

BTW, did I mentioning KTuner is a bug, one can't share simply the calibration for check, you need a dongle to be able too. If you are unexperienced with it as customer and have no tuner aside with same hardware it is a mess. This will extend the time to solution if I can't just revise your calibration. That's way I don't like KTuners price/product access policy.
Unfortunately Ktuner doesn't support any other MAP sensors on this ECU, I'll try a different OEM sensor just to confirm.

Indeed the only condition I have found to trigger the issue is exceeding 8psi

I also thought it could be a threshold issue, Ktuner says they were working on a boost cut option for this ECU but the feature is not available in the software yet and probably will never be!
 

·
Arouse the DAMPFHAMMER!
Joined
·
9,422 Posts
I'll try a different OEM sensor just to confirm.
That's a good idea. I observed a similar issue on a customer engine (K24, Hondata FPro, NA), Hondata pointed that time to the CAT overheating protection function. It is most likely a so called virtual sensor, a mathematical model, collecting sensor data for calculation of an assumed CAT mean temperature. At least Lambda, ECT, engine speed, MAP and IAT go into that, most likely ignition timing to, everything wich can be used to conclude a possible overheating is going on. I would use lambda heater current too to assume the wideband temperature. One of these sensors or a ECU read pin can be offset'd and cause the issue. I would tend to point to the wideband, maybe 2nd lambda probe is still active, maybe that. Just wild guessing, but as it is rare in my experience, condensation to root cause need much more cool down ;).

I also thought it could be a threshold issue, Ktuner says they were working on a boost cut option for this ECU but the feature is not available in the software yet and probably will never be!
There are other options on the market, thanks to competition: OE based and standalone based systems for almost any price.
 

·
Registered
Joined
·
23 Posts
Discussion Starter · #18 ·
well I replaced the MAP sensor with another OEM one I had with no change, so that is ruled out.
I also disabled the catalyst monitor and retested with no change, that might possibly rule out any CAT overheating protection.
I have a couple more tests that I'm thinking about, will report back when I'm done
 

·
Arouse the DAMPFHAMMER!
Joined
·
9,422 Posts
well I replaced the MAP sensor with another OEM one I had with no change, so that is ruled out.
I also disabled the catalyst monitor and retested with no change, that might possibly rule out any CAT overheating protection.
Sounds good, thanks for your feedback to those items.

Would you please be so kind to update here the actual log graph to see what is left on the other sensors?
 

·
Registered
Joined
·
23 Posts
Discussion Starter · #20 ·
Here are the 2 tests I did to figure out if there are other variables involved other than MAP reading going over 8PSI and triggering this issue.

I used a AEM FIC piggyback to manipulate only the MAP sensor signal

first test I increased MAP sensor signal using an AEM FIC piggyback by 15%, so when you see in the log 8PSI in reality it's only 7PSI ( confirmed by both AEM built in MAP sensor and my boost gauge) the issue starts immediately and the ECU cuts fuel pulse to half , notice the zigzag duty cycle!
Rectangle Slope Font Line Parallel



In the second test I do the exact opposite and clamp the MAP signal so it never exceeds 7.6PSI while in reality I'm running closer to 9PSI but the issue never happens
Rectangle Slope Font Parallel Screenshot
 
1 - 20 of 24 Posts
Top