the great IAT debate...

[magius231]

ok so in everything I have been able to find, there doesn't seem to be a clear answer/solution to the Gen 3 lean hot startup issue. Most have said its IAT heat soak (which it can be for sure) but there seems to be more to it than that.


this is on a 1999 Trans Am with the stock ECU, cammed, turbo, 3 bar OS

here's what I did today to try to test this out:

- Cold start from siting overnight, it's 59* out today. drove around for quite a while gathering VE data...I'm running rich all around, sometimes as much as 20%. IAT's around 59*-62*
- idled stopped for maybe 10 minutes until it heat soaked the IAT and I was seeing 97* IAT temps and it started hitting 17-18 AFR
- shut it down for maybe 5 minutes. hard start (almost didn't actually) and IAT's are at 95*. hitting 17-20 AFR at this point
- drove it around with AFR locked around 11.4 to let it actually drive. within 10 minutes or so I was back down to 64* IAT, turned off AFR lock and seeing 16-17 AFR
- Drove it around some more until IAT went to 61*, AFR at 14-15. Better but still higher than it should be, even with IAT normal
- drove it around ~10 minutes more, no change...AFR still 14-15 and IAT still 59*-61*.

I attached logs of the normal drive and after the IAT had cooled down as well as my tune...

so something else is adding to the issue other than just IAT soak...there has to be some other factor...another sensor heat soaked, some part of the code or table we can't see, something.
I could flash to a 2002 OS but if IAT soak isn't the problem, changing the bias table won't really help right?

Has anyone made any headway in determining what else could be contributing to these issues we are experiencing?

[kingtal0n]

I'm not clear on the issue. Having trouble starting? Having trouble staying running?

I went through some minor difficulty which was easy to overcome in gen3 using these
1. Variable resistor on the IAT signal to dial IAT up or down as needed, about 1000Ohms is more than enough adjustment range, I barely use it though
2. Afterstart enrichment based on IAT 'heatsoak' tables takes care of fuel starting when hot
3. Park position and similar IAT based IACV opening takes care of airflow (density) difficulty starting when hot

You have fuel and air adjustments for starting based on IAT, 'heat soak' tables and others
And a fuel adjustment for putting into gear based on something

If it runs 15's 16's during cruise let it, no harm for < 50KPA most mild compression engines on gasoline with properly tuned timing running lean is beneficial to plug cleanliness and reduce carbon deposits.

[magius231]

it's really the same issue thats been discussed - after hot restarts the engine runs very lean for a period of time. i guess what I'm trying to add to this is that it's not just IAT heat soak, there's something else in play here and it would help people a lot of we could determine what that might be.

[kingtal0n]

Some of them say it runs rich when its hot. Some say it runs lean. Nobody can agree on anything

I think
1. Fuel is hot has reduced fuel density and may even contain pockets of gas after a hot restart with dead-head rails
2. System voltage is lower while alternator is hot causing lazy injector opening
3. Fuel injectors open slower when they are hot, hot fuel + hot injectors = leaner as injectors get larger, because larger injectors have lower opening times
4. Insufficiently programmed heatsoak afterstart enrichment table to compensate for these issues

But then we have one who swears his runs richer when the iat is very high so

[smokeshow]

Some of them say it runs rich when its hot. Some say it runs lean. Nobody can agree on anything

Nobody said their results need to agree... Both scenarios are possible, depending on the complex air temp model calibration and environment. Since I anticipate a wall of text no doubt including your little resistor hack and other unrelated nonsense...I'll hold your hand through this one. And hopefully others can learn as well.

This is a little sim I built a few years ago to ground the calibration to log data where actual charge temp was suspected not to match the modeled temp. It takes the filter and bias tables from the existing calibration, your inputs via the dials (or log data) and generates the first order response of the charge temp. To visualize an error, the switch simulates the response with error from modifications to the air flow path...in this case a metal intake manifold. Being the doctor of engineering I know you most definitely are, you have access to MATLAB/Simulink. So download the "csv" file I have attached, change the file extension to 'zip', open up the model and hit the Run button. Play with the IAT and MAF knobs and observe response in the gauges. Then use the toggle switch to simulate swapping the intake manifold out for a metal unit. Notice potential for both positive and negative differences between modeled and actual MAT...meaning rich or lean offsets. Now, last step: throw your resistor away, calibrate it properly with what you just learned and stop telling people to hack up their vehicles.

Class: dismissed.

complex_charge_temp_model.csv

[kingtal0n]

Calibrate what properly? There is no way to access Gen3 IAT based tuning efforts tables outside of PE.

IF you mean calibrate the IAT sensor properly? I Disabled my IAT years ago and it makes no difference. The A/F Still walks around when the engine heats up in traffic even with a steady IAT unmoving.
As the engine heats up, the air is less dense, it should go rich. So why does it go lean? Something else at work here. I suspect a relationship between fuel density and voltage/performance of the injectors in my app.


About 50% to 70% of The A/F swing people experience is not related to the IAT Input as measured. Even with a 'resistor hack' you cannot fully re-compensate the a/f ratio swing experienced from heat soak situations by dialing the IAT Back to cold values.

[smokeshow]

Calibrate what properly? There is no way to access Gen3 IAT based tuning efforts tables outside of PE.

IF you mean calibrate the IAT sensor properly? I Disabled my IAT years ago and it makes no difference. The A/F Still walks around when the engine heats up in traffic even with a steady IAT unmoving.
As the engine heats up, the air is less dense, it should go rich. So why does it go lean? Something else at work here. I suspect a relationship between fuel density and voltage/performance of the injectors in my app.


About 50% to 70% of The A/F swing people experience is not related to the IAT Input as measured. Even with a 'resistor hack' you cannot fully re-compensate the a/f ratio swing experienced from heat soak situations by dialing the IAT Back to cold values.

Follow. The. Instructions.

I literally provided a tool that shows how this all works in my last post. Now you have to put in the effort. I swear, if you were a horse that had to be led to water, you'd die of dehydration.

[redhardsupra]

nope, there is NO way to tune 3rd gen IAT+ECT+BIAS model. <-sarcasm
http://redhardsupra.blogspot.com/200...ng-part-2.html
http://redhardsupra.blogspot.com/200...important.html

[smokeshow]

nope, there is NO way to tune 3rd gen IAT+ECT+BIAS model. <-sarcasm
http://redhardsupra.blogspot.com/200...ng-part-2.html
http://redhardsupra.blogspot.com/200...important.html

I'm sorry, are you arguing with professor kingtalon? You must be new here.

[kingtal0n]

nope, there is NO way to tune 3rd gen IAT+ECT+BIAS model. <-sarcasm
http://redhardsupra.blogspot.com/200...ng-part-2.html
http://redhardsupra.blogspot.com/200...important.html

I don't use ECT bias.
I remove the IAT and it still has lean issues when heat soaking, so its not the IAT.

Furthermore, My first IAT is out by the intercooler, second one is a good foot away, neither are nowhere near the intake manifold. This whole metal vs plastic intake debate is meaningless to me.

Car is tuned just fine, walking lean when it heat soaks once in a while is something I just ignore. It was never something that bothered me, it just bothers other people and I tried to help.
Good luck trying to fix the issue whoever is still having it.

[Pulse_GTO]

Nobody said their results need to agree... Both scenarios are possible, depending on the complex air temp model calibration and environment. Since I anticipate a wall of text no doubt including your little resistor hack and other unrelated nonsense...I'll hold your hand through this one. And hopefully others can learn as well.

This is a little sim I built a few years ago to ground the calibration to log data where actual charge temp was suspected not to match the modeled temp. It takes the filter and bias tables from the existing calibration, your inputs via the dials (or log data) and generates the first order response of the charge temp. To visualize an error, the switch simulates the response with error from modifications to the air flow path...in this case a metal intake manifold. Being the doctor of engineering I know you most definitely are, you have access to MATLAB/Simulink. So download the "csv" file I have attached, change the file extension to 'zip', open up the model and hit the Run button. Play with the IAT and MAF knobs and observe response in the gauges. Then use the toggle switch to simulate swapping the intake manifold out for a metal unit. Notice potential for both positive and negative differences between modeled and actual MAT...meaning rich or lean offsets. Now, last step: throw your resistor away, calibrate it properly with what you just learned and stop telling people to hack up their vehicles.

Class: dismissed.

Capture.PNG

Thanks for posting that tool. Took me a moment to create a MATLAB account and install the MATLAB drive, but I got it to work.

[SiriusC1024]

99 TA HCIE.

I have similar issues. Never thought it was IAT. At the time I had a narrowband that would go all whacky for around 15 seconds after a hot start on a hot day. It went back to normal after driving a little. The gauge would swing so rapidly that I doubt it was anything fuel trim related. Super rapid LED flashes.

Never put much thought into it since it's a minor inconvenience. Just figured it was something with the O2 sensors maybe temporarily fouling. Couldn't be the ECU since the narrowband read directly off the O2.

Also, Denso O2's seem to work better than the factory Bosch's.

[Kinkstaah]

I don't use ECT bias.
I remove the IAT and it still has lean issues when heat soaking, so its not the IAT.

Furthermore, My first IAT is out by the intercooler, second one is a good foot away, neither are nowhere near the intake manifold. This whole metal vs plastic intake debate is meaningless to me.

Car is tuned just fine, walking lean when it heat soaks once in a while is something I just ignore. It was never something that bothered me, it just bothers other people and I tried to help.
Good luck trying to fix the issue whoever is still having it.

Without seeing (any) of these threads I came to a similar conclusion. Without changing any settings, hot IAT should make the car run rich, not lean - but it doesn't.
The ECU by default logically is trying to remove fuel when the IAT is hot to avoid a too-rich condition. However, it just pulls too much out by default.

I came to this conclusion like we all do, being angry at car running lean when hot, and realising retuning the car every time we go for a drive is a bit shit.
...Probably why the OEM cars have a MAF to begin with, and SD is a backup table when MAF fails. If I had've known all this to begin with I would have simply used 4in MAF and tuned with the MAF. I still might.

Following the above logic, skewing the IAT sensor so it thinks its colder than it actually IS 'should' bandaid this inbuilt overcompensation, and is available in the ECU by changing the resistance to temperature ratio.
I realise you've done this by effectively removing the IAT and setting a consistent, cold temperature across the board.

I recorded/tuned a baseline at 14C / 57F track day where I did 30 minute sessions with a wideband on a road course, using the Omega IAT sensor values with IAT's showing 12-14C, and the weather being measured outside the car at the same. it was probably the most thorough tuning I could possibly do. It doesn't really matter if it was actually 14C, or 15C, or 13.2C, what I had was an absolute beating on the road course, and AFR's where I wanted them to be, with an IAT sensor dialled very close to the real temperature of the air.

I knew other days would be warmer, in traffic, etc etc. But I finally had a solid baseline, and my goal was to have it skew rich when it's warmer than 57F/this baseline, which is most days. In my case this worked.

I do not know why it wouldn't work for you by setting your whole IAT sensor to think it's always cold - This logic worked flawlessly for me. In my case I blended it, from my 'correct' temp and about halfway between what the Omega sensor wants/is configured to and a colder temperature where I knew it was right and the VE table was dialled in at. (14C/57F).
I figured I could guesstimate it into being closer-to-accurate as I knew I was compensating for _something_ pulling too much fuel out when hot, I didn't want it to run rich all the time.

I also found that boxing an airbox in absolutely works. My IAT sensor is in a steel pipe that is about 2in from the throttle body. Boxing the AIRBOX/intake in fully - I duct the air from the front bumper in my R34, dropped the IAT's from about 100F+ on the coldest of cold days to 57F, on a 57F day even though the pipe itself is obviously hot as fuck. A friend did this with his BMW and we can find by opening the lid of the box, the IAT jumps up to 100+F very quickly, replace lid and it drops down the ambient air temp from outside, even though the pipe itself remains hot as fuck.

We concluded heatsoak from the pipes/manifolds aren't actually a thing, and it - it being the OEM GM IAT sensor, or the Omega sensor, or the Ford one used in my friends BMW - reads hot because the actual air is actually hot.

My more detailed adventures in said ducting/airbox is here:
https://www.sau.com.au/forums/topic/...omment=7961575

I know it's not a metal intake. But you will find researching here and on LS1tech that people state that putting an IAT sensor near the TB, in a metal pipe, in an engine bay, leads to too-hot-readings that are inaccurate.
Not in my case.

[Hondaeater]

High IAT = hot air, and some say it's rich others lean... Hmm... As pressure increases, with TEMPERATURE CONSTANT, density increases. Conversely when temperature increases, with PRESSURE CONSTANT, density decreases. Air density will decrease by about 1% for a decrease of 10 hPa in pressure or 3 ?C increase in temperature. Does opening TB change air pressure? 🤔 Does a change in temperature change air density? 🤔 What if TB opens but MAP pressure stays constant kPa? Would the amount of fuel be same with this "lie" of map pressure? 😔 Is air ACTUALLY warmer from heatsoak?
Let's plug in some numbers ...
https://atgtraining.com/atg-volumetr...cy-calculator/

[smokeshow]

Thanks for posting that tool. Took me a moment to create a MATLAB account and install the MATLAB drive, but I got it to work.

How'd it work for ya? At least someone tried it lol. Tried to lead ol professor KT to water with that but the horse didn't budge. Color me shocked...

[kingtal0n]

How'd it work for ya? At least someone tried it lol. Tried to lead ol professor KT to water with that but the horse didn't budge. Color me shocked...

Its like you can't read. Some of us have removed the IAT from the vehicle completely and still suffer from heatsoaking lean A/F ratios in some swaps.

There is clearly something other than IAT at work and no matter what you simulate in matlab using IAT maths it doesn't mean anything for us who have this issue with no IAT sensor.... get it yet?

That 'simulation' does nothing and I don't think it can ever help anybody with this problem. It is not only lacking transfer function for realistic physical sensor properties it lacks empirical testing to find constant multipliers associated with referencing frequency response. All freshman engineers learn matlab but it isn't useful without transfer function maths and physical testing for specific arrangements of driver and frequency response including bode plot, phase, and other details in a realistic real world application such as a combustion engine

[smokeshow]

Its like you can't read. Some of us have removed the IAT from the vehicle completely and still suffer from heatsoaking lean A/F ratios in some swaps.

There is clearly something other than IAT at work and no matter what you simulate in matlab using IAT maths it doesn't mean anything for us who have this issue with no IAT sensor.... get it yet?

That 'simulation' does nothing and I don't think it can ever help anybody with this problem. It is not only lacking transfer function for realistic physical sensor properties it lacks empirical testing to find constant multipliers associated with referencing frequency response. All freshman engineers learn matlab but it isn't useful without transfer function maths and physical testing for specific arrangements of driver and frequency response including bode plot, phase, and other details in a realistic real world application such as a combustion engine

Correct me if I am way off here, but the last term of the ideal gas law is temperature... Removing the sensor may....just might....have some unintended effects. FYI... https://letmegooglethat.com/?q=ideal+gas+law

If you actually opened the model, you'd find the transfer function you missed.

[SiriusC1024]

The IAT is referenced differently in MAF vs SD mode, correct? When I unplug the IAT in my car, tuned in SD, runs like garbage.

Regardless, I don't think it's the IAT itself. If anything it's the other part of that circuitry that's getting heatsoaked. This 2-wire sensor is probably one leg of a wheatstone bridge. What's upstream in the circuit?

[kingtal0n]

Correct me if I am way off here, but the last term of the ideal gas law is temperature... Removing the sensor may....just might....have some unintended effects. FYI... https://letmegooglethat.com/?q=ideal+gas+law

If you actually opened the model, you'd find the transfer function you missed.

For diagnostic purposes it is essential to rule out specific sensor inputs such as IAT and CTS. Therefore many people disabled the CTS Bias early on to note the influence of the density algorithm minus the sensor input. For some reason this does not raise any alarms.... and yet

Then, we do the same thing with the IAT, that is remove the IAT sensor input and configure a non-changing temperature for IAT to help diagnose the issue. This is a temporary measure simply for diagnostic evaluation.


Now, both the IAT and CTS have been removed from calculations and their respective sensor inputs are no longer contributing to the leaning out condition.
If the condition persist, then it must not have anything to do with density calculations based on IAT and CTS inputs. Engine gets hot, IAT stays the same, Engine leans out.
Useless. Pointless to look at IAT any further.

Something else is causing these events.

--------------------------------------------------------------------

Ideal gas law now thats funny. This isn't general chemistry and we are not children, I do not know why you bring this up at all. It's called ideal for a reason, it doesn't exist. True gas behaviors for a model sophisticated as combustion engine ECU probably do not utilize a ideal gas law type of model for something like IAT. Ideal gas law has 3 basic variables with linearity in suite. An accurate IAT based model I would expect to have far more, perhaps 10 or 20 variables with squares and radicals and exponents and logs all mish mashed based on modelling and testing with empirically derived constants for correction and it would only closely work with and apply to the OEM engine configuration for that particular ECU/tune/engine.
I would know it if I saw it. This is because temperature and pressure are scalars which cannot assume any velocity vector or kinetic energy component as the realistic engine is running and those are places to store energy related to molecular motion which can become temp and pressure at some point during a collision with some container and that is only the beginning of the issue, we also have input from residual combustion as valve temperatures localized to the chamber where airflow must pass closely, a boundary layer of molecules along every corridor which will be heated further as it moves less, resulting gradient P and T functions, and radiated energy infrared from nearby components and insulating factors as manifold and head materials transverse all of which influence the air behavior and ultimate density and even sound or acoustic pressure associated with molecular motion near valves. These add complexity to the model which is why we disabled the CTS and IAT for diagnostics to rule out any influence on behalf of the OEM ECU trying to compensate in an aftermarket installation using internal clock or poor tuning habits to make poor decisions in terms of fueling based on outdated information.

*Its much easier to disable an input and rule it out than determine whether an input is 'correctly adjusted in terms of output' or not. *
What is a very simple diagnostic solution you turned into a pile of code and software and vague vain attempt grasping for straws that mean nothing and just wasted your time.


As for air density relationship,
There is no reason to re-invent the wheel, plenty of online calculators including one recently posted in this thread can be used to model air density and VE for a cylinder, we don't need matlab to do that for us. Another waste of effort.

[SiriusC1024]

For diagnostic purposes it is essential to rule out specific sensor inputs such as IAT and CTS. Therefore many people disabled the CTS Bias early on to note the influence of the density algorithm minus the sensor input. For some reason this does not raise any alarms.... and yet

Then, we do the same thing with the IAT, that is remove the IAT sensor input and configure a non-changing temperature for IAT to help diagnose the issue. This is a temporary measure simply for diagnostic evaluation.


Now, both the IAT and CTS have been removed from calculations and their respective sensor inputs are no longer contributing to the leaning out condition.
If the condition persist, then it must not have anything to do with density calculations based on IAT and CTS inputs. Engine gets hot, IAT stays the same, Engine leans out.
Useless. Pointless to look at IAT any further.

Something else is causing these events.

--------------------------------------------------------------------

Ideal gas law now thats funny. This isn't general chemistry and we are not children, I do not know why you bring this up at all. It's called ideal for a reason, it doesn't exist. True gas behaviors for a model sophisticated as combustion engine ECU probably do not utilize a ideal gas law type of model for something like IAT. Ideal gas law has 3 basic variables with linearity in suite. An accurate IAT based model I would expect to have far more, perhaps 10 or 20 variables with squares and radicals and exponents and logs all mish mashed based on modelling and testing with empirically derived constants for correction and it would only closely work with and apply to the OEM engine configuration for that particular ECU/tune/engine.
I would know it if I saw it. This is because temperature and pressure are scalars which cannot assume any velocity vector or kinetic energy component as the realistic engine is running and those are places to store energy related to molecular motion which can become temp and pressure at some point during a collision with some container and that is only the beginning of the issue, we also have input from residual combustion as valve temperatures localized to the chamber where airflow must pass closely, a boundary layer of molecules along every corridor which will be heated further as it moves less, resulting gradient P and T functions, and radiated energy infrared from nearby components and insulating factors as manifold and head materials transverse all of which influence the air behavior and ultimate density and even sound or acoustic pressure associated with molecular motion near valves. These add complexity to the model which is why we disabled the CTS and IAT for diagnostics to rule out any influence on behalf of the OEM ECU trying to compensate in an aftermarket installation using internal clock or poor tuning habits to make poor decisions in terms of fueling based on outdated information.

*Its much easier to disable an input and rule it out than determine whether an input is 'correctly adjusted in terms of output' or not. *
What is a very simple diagnostic solution you turned into a pile of code and software and vague vain attempt grasping for straws that mean nothing and just wasted your time.


As for air density relationship,
There is no reason to re-invent the wheel, plenty of online calculators including one recently posted in this thread can be used to model air density and VE for a cylinder, we don't need matlab to do that for us. Another waste of effort.

You gotta chill with the dick measuring. You're not talking over anyone's head, and you know damn well that you don't need to solve boundary value problems in order to simulate AFR. These ECU controls are based on the ideal gas law and PID loops.

P - MAP
V - VE tables
=
n - solved
R - Constant
T - IAT