AFR: day vs night???

[WS6Fury]

the only Manifold temp PID i found was "Manifold Air Temp" and it's an unsupported PID on my '04 GTO.

Ugh!! Just realized while studying everything, I don't have any "Manifold Air Temp" listed in the scanner. So i guess LS1's won't be able to utilize all this great research

[redhardsupra]

Ugh!! Just realized while studying everything, I don't have any "Manifold Air Temp" listed in the scanner. So i guess LS1's won't be able to utilize all this great research

feel free to make requests to chris@hpt, they're all sick of me making requests, it's time for someone else to annoy them

[WS6Fury]

feel free to make requests to chris@hpt, they're all sick of me making requests, it's time for someone else to annoy them

Ha ha! Awesome, I'll shoot him an email for sure. I'm curious now though, I'm wondering if there is anything/sensor in the LS1's to even measure the manifold temp. did a bunch of searches and didn't see much about it...

[redhardsupra]

no, there isn't anything to measure manifold temps, that's why we have this whole complex model of estimating it from two other sensors

[WS6Fury]

Put the request in, so see what happens. Not sure how pids can be added, unless only with updates... hope not, wanna try and hit this bias table down when I install the new fast 90/90 and new fast toys 85mm lid

Any ideas in meantime to fudge factor the curve without the manifold temp scanned?

[redhardsupra]

if you read the full thing, it would tell you that by itself it doesn't do much

[deanm11]

if you read the full thing, it would tell you that by itself it doesn't do much

Ok, I have a very basic question --- why isn't the scanned MAT by definition the same as that calculated using the current (stock or otherwise) tune's bias and bias filter tables, as applied to current IAT and ECT ? Why would the PCM have another way of coming up with MAT ? So we're saying that for some reason GM delivers these cars with MAT out of sync with what they are using for charge temp? Or maybe that is the issue- is there a difference between MAT and charge temp in the combustion chamber? What is the MAT used for if the PCM goes to the bias tables for fueling ?
And finally, if we retune the bias tables to match MAT, what is really achieved if the MAT pid is very possibly wrong ? Or what makes us think that MAT is the right temp to use for SD fueling and if so, why doesn't the PCM just use it ?

[svede1212]

i'm not exactly sure what your asking but the MAT is a calculated number and the "scanned MAT" is just that calculation based off of IAT, ECT, bias and filter seeing as there is no sensor in the manifold for that. seeing a MAT PID would give you an idea of what that calculation was.

[redhardsupra]

Ok, I have a very basic question --- why isn't the scanned MAT by definition the same as that calculated using the current (stock or otherwise) tune's bias and bias filter tables, as applied to current IAT and ECT ? Why would the PCM have another way of coming up with MAT ? So we're saying that for some reason GM delivers these cars with MAT out of sync with what they are using for charge temp? Or maybe that is the issue- is there a difference between MAT and charge temp in the combustion chamber? What is the MAT used for if the PCM goes to the bias tables for fueling ?
And finally, if we retune the bias tables to match MAT, what is really achieved if the MAT pid is very possibly wrong ? Or what makes us think that MAT is the right temp to use for SD fueling and if so, why doesn't the PCM just use it ?

you got concepts right, but not my paper, i should probably rewrite it :/

calculated MAT is different from the scanned MAT because they come with different resolutions. i should've probably round the calculated values, because right now it looks like there's a discrepancy. well, there is one, but it's small and that's not what we're after.

i got some scans from another car that's got temp accounting issues and find some more dramatic discrepancies, where adjusting the bias/filter values would bring the AFRcommanded vs AFRwb closer together.

damn it guys, you're making me make my findings more coherent i'll post it on my site when i find some better examples.

[deanm11]

i'm not exactly sure what your asking but the MAT is a calculated number and the "scanned MAT" is just that calculation based off of IAT, ECT, bias and filter seeing as there is no sensor in the manifold for that. seeing a MAT PID would give you an idea of what that calculation was.

Thanks- this is exactly what I thought... so what is redhardsupra doing when optimizing the tables ? Shouldn't you be able to just plug in the stock (or current) bias and filter tables and be guaranteed to match the scanned MAT ? It seems not for some reason, so I guess I am asking why ... And if you've 'optimized' them so that it matches, once you plug them into your tune, do we expect the calculated MAT (by us in the spreadsheet) to match the logged MAT ? If it didn't before, why should it then ?

[redhardsupra]

stock values produce 'calculated MAT' values to be within 1K of the 'scanned MAT' which is understandable, as the resolution on scanned MAT is much smaller than the calculated (or the calculated is much bigger, as it's just numbers mangled with math).
what i didn't post is me optimizing the VE values alongside with bias params to achieve a smaller AFR%error across the board. at the end of the paper i talk about how you can achieve a better AFR%error just by 'optimizing' bias tables but how that's not the right thing, because then you're practically tuning the car with temperature estimation only, and that's just silly wrong.
until i can figure out an easy way to optimize both VE and bias params simultaneously, temp estimation is gonna be a purely theoretical endeavor, and this paper is only an explanation how this stuff works. i should've probably not mention any 'optimizing' in it. i only brought it up, because it's a good generic approach to optimizing other things (which hopefully one day will show up in a more holistic approach to tuning).

so what redhardsupra is doing? he moved across the country and started a new job, so he hasn't touched tuning in few weeks. back to the tuning night shift for me i guess... :/

[deanm11]

you got concepts right, but not my paper, i should probably rewrite it :/

calculated MAT is different from the scanned MAT because they come with different resolutions. i should've probably round the calculated values, because right now it looks like there's a discrepancy. well, there is one, but it's small and that's not what we're after.

i got some scans from another car that's got temp accounting issues and find some more dramatic discrepancies, where adjusting the bias/filter values would bring the AFRcommanded vs AFRwb closer together.

damn it guys, you're making me make my findings more coherent i'll post it on my site when i find some better examples.

To be totally clear, I am questioning why, in your spreadsheet, why wouldn't MATscan = MATfilter, if one used the bias and filter table IN THE TUNE in use while the scan was made ? Right now, you are optimizing an error term to try to get them as close as possible. Sorry, what I meant by calculated is what you have as MATfilter in the sheet. Thanks for your patience.

[deanm11]

stock values produce 'calculated MAT' values to be within 1K of the 'scanned MAT' which is understandable, as the resolution on scanned MAT is much smaller than the calculated (or the calculated is much bigger, as it's just numbers mangled with math).
what i didn't post is me optimizing the VE values alongside with bias params to achieve a smaller AFR%error across the board. at the end of the paper i talk about how you can achieve a better AFR%error just by 'optimizing' bias tables but how that's not the right thing, because then you're practically tuning the car with temperature estimation only, and that's just silly wrong.
until i can figure out an easy way to optimize both VE and bias params simultaneously, temp estimation is gonna be a purely theoretical endeavor, and this paper is only an explanation how this stuff works. i should've probably not mention any 'optimizing' in it. i only brought it up, because it's a good generic approach to optimizing other things (which hopefully one day will show up in a more holistic approach to tuning).

so what redhardsupra is doing? he moved across the country and started a new job, so he hasn't touched tuning in few weeks. back to the tuning night shift for me i guess... :/

Thanks, I follow it all. I guess I just didn't understand the purpose of your exercises then. I am also trying to think how to solve this thing.. it is like you need to put a peg in the sand, at least at one point in the bias table and get VE 100% right there and adjust the bias (and filter) tables on either side of that point... similar to setting that VE multiplier table in the efilive custom os and picking a pivot point.

[svede1212]

stock values produce 'calculated MAT' values to be within 1K of the 'scanned MAT' which is understandable, as the resolution on scanned MAT is much smaller than the calculated (or the calculated is much bigger, as it's just numbers mangled with math).
what i didn't post is me optimizing the VE values alongside with bias params to achieve a smaller AFR%error across the board. at the end of the paper i talk about how you can achieve a better AFR%error just by 'optimizing' bias tables but how that's not the right thing, because then you're practically tuning the car with temperature estimation only, and that's just silly wrong.
until i can figure out an easy way to optimize both VE and bias params simultaneously, temp estimation is gonna be a purely theoretical endeavor, and this paper is only an explanation how this stuff works. i should've probably not mention any 'optimizing' in it. i only brought it up, because it's a good generic approach to optimizing other things (which hopefully one day will show up in a more holistic approach to tuning).

so what redhardsupra is doing? he moved across the country and started a new job, so he hasn't touched tuning in few weeks. back to the tuning night shift for me i guess... :/

the paper makes sense my only problem is my car reacts opposite of what the theory says it should do. when it gets cold out my AFR goes insane rich (from 14.7>13.9 going from 88*>60*). i'm not sure if you are supposed to tune the VE table cold and adjust the bias or tune when it's hot out and then adjust. i've been chasing this problem since spring...

ah ha. that's why i posted the above. i saw that the VE was muddying up the whole thing. by blind luck i guess it seems i'm getting close to where i should be as the AFR is within acceptable range now (it doen't get to 90* out anymore tho) but i'd like to get it nailed and repeatable for others.