TD_D mod journey. From stock to the 'bastardmobile'

[IGOTASTi]

Well now since you are a Sponsor you can upload to your albums and then copy the code and put in it the thread and bam! Let me know if you need more help.

 

[Td_d]

Got these latency tables off injector rehab, what strikes me is that there is quite a significant variance (latency far too high on mine). Also, the resolution could be greatly refined, since it's unlikely that battery voltage will drop below 10V, so the extra column could be used to up the resolution. These are high impedence, so 3ohm doesnt matter.

Injector Flow CC/min Flow LB/Hr Ohm 10v 11v 12v 13v 14v 15v


RC Engineering 750 71 12 1.36 1.12 0.92 0.76 0.63 0.5

 

[HolyCrapItsFast]

Ok, mechanic's trying to source some new ID1000's for me (those one already moved). I'll log some IDCs - if they're kosher for the moment, and it takes a little time to get the new 1000's in, would you be willing to help me dial in these 750's (and then the 1000's when they're on)?

Am I correct in assuming, looking at the way the trims shift from negative to positive that the latencies are too high at the moment?

Yes not a problem. And yes, exactly... your Latency needs to come down.

Ok, managed to log some IDC over lunchtime - looking fine - absolute maximum was 83% with at WOT with high engine load - so well within the comfort zone. If I want to push higher boost, will definately need the 1000's to give it some headroom. So - until the 1000's come in, I would be very keen to try and sort the tune on the 750's. ECU pulled a little timing at the top - although I suspect this may be a leftover from the stock injectors, as FLKC takes a while to come down. Zero FBKC, so not a worry in anycase - if it is from the current tune will probably need to dial the timing down 05.-1 degree between 5000-7000 RPMs.

George - I am at your mercy! :lol:

I assume that the first step would be to reduce the latency, and then do another set of A/F runs to see the corrections?

Here's the latency table as it stands. Injector flow constant is 714cc/min. And the fuel pump pressure, BTW is 3 bar.


View attachment 712

ARRRRRGGGGGHHH... how the hell do I post tables in here?

Okay, Just a disclaimer before we get started...

Please understand that since we are doing this over the internet that this whole process is going to take some time to complete. If you can dedicate the time and commit, so can I. I also need you to understand that i do not do things half assed and insist on completing my mission especially because this is someone Else's car. I will be happy to explain every thing I do in detail both for your understanding and confidence.

Agreed?

Okay... I have attached a new version of your ROM. if you could flash it to your car this will get us started. What I did here is apply a multiplier to the entire latency table of .75 for an effective decrease of 25%. What this is going to do is bring the low end of the MAF closer to the top end. The thing you will notice though is the entire MAF scale will most likely be on the lean side. That's okay, we can live with it until we get the injectors scaled. If worse comes to worse we can always step back to the previous tune.

The next thing we will work on is getting that to normalize and the injectors will be scaled. The next thing after that is to scale the MAF.

http://www.drift-ready.com/Files/TD_D/750RC_TD_D_02-01-2011.zip

So to recap we will

Dial in the Latency.
Adjust your fuel tables.
Fine tune your load cells.
Scale the injectors.
Scale the MAF.
Address any knock.

I would like to keep this here in your journal as an example of the steps it takes to achieve a decent tune Is that okay?

The other question I had is I noticed that the Primary Fuel Table is set up way differently than I have seen done in the past. I'm not saying it is wrong but is this working for you? It seems awfully rich in places it shouldn't be and a little lean in others. It is also an abrupt transition from one AFR target to another. I will leave it alone if this is desired. Let me know your thoughts :tup:

Here is an example of mine...

And yours...

 

[Td_d]

Yes not a problem. And yes, exactly... your Latency needs to come down.

Okay, Just a disclaimer before we get started...

Please understand that since we are doing this over the internet that this whole process is going to take some time to complete. If you can dedicate the time and commit, so can I. I also need you to understand that i do not do things half assed and insist on completing my mission especially because this is someone Else's car. I will be happy to explain every thing I do in detail both for your understanding and confidence.

Agreed?

Most definately - considering that you are offering to help a complete stranger half way across the world, the minimum I could do is to actively participate. It will take some time, I'm well aware - but that's the price for a _good_ tune - which I am very grateful to be getting assistance for. Only complication will be the time zones - 8 hours - which may slow thing down a little. I'm very happy to keep this up here as an example of the process - NSFW has done some great work on the openecu forums, but I think the nature of RomRaider and the discussion tends to scare people off.

The other question I had is I noticed that the Primary Fuel Table is set up way differently than I have seen done in the past. I'm not saying it is wrong but is this working for you? It seems awfully rich in places it shouldn't be and a little lean in others. It is also an abrupt transition from one AFR target to another. I will leave it alone if this is desired. Let me know your thoughts :tup:

No particular reason - I was trusting the tuner's judgement - so feel free to go ahead and tweak. And no, it's definately not working - it's so rich in place you can actually smell raw gas now.

I'll go and load the map on now - what should I log thereafter?

 

[Td_d]

There seems to be a problem with the file uploaded - wont let me download it. Can you try and reup? Thanks. Hmm, seems drift-ready is unreachable, let me keep on trying.

 

[HolyCrapItsFast]

Try to stick only with the same thing I previously suggested. To many things and we lose resolution. A/F correction #1, A/F Learned #1, A/F Sensor #1 (AFR), Engine Speed, Engine Load, Maf Voltage, OL/CL Fueling, Throttle Open Angle.

Don't do any WOT. Just normal around town and highway driving. Maybe a couple slightly harder acceleration runs but not WOT. You can put it all in one long log. Reset the ECU prior to logging, allow the car to come up to temperature.

Most definately - considering that you are offering to help a complete stranger half way across the world, the minimum I could do is to actively participate. It will take some time, I'm well aware - but that's the price for a _good_ tune - which I am very grateful to be getting assistance for. Only complication will be the time zones - 8 hours - which may slow thing down a little. I'm very happy to keep this up here as an example of the process - NSFW has done some great work on the openecu forums, but I think the nature of RomRaider and the discussion tends to scare people off.



No particular reason - I was trusting the tuner's judgement - so feel free to go ahead and tweak. And no, it's definately not working - it's so rich in place you can actually smell raw gas now.

I'll go and load the map on now - what should I log thereafter?

 

[HolyCrapItsFast]

There seems to be a problem with the file uploaded - wont let me download it. Can you try and reup? Thanks. Hmm, seems drift-ready is unreachable, let me keep on trying.

Yeah I'm having problems with my server. I'll let you know when I get it up! Well running that is.

 

[Td_d]

Try to stick only with the same thing I previously suggested. To many things and we lose resolution. A/F correction #1, A/F Learned #1, A/F Sensor #1 (AFR), Engine Speed, Engine Load, Maf Voltage, OL/CL Fueling, Throttle Open Angle.

Don't do any WOT. Just normal around town and highway driving. Maybe a couple slightly harder acceleration runs but not WOT. You can put it all in one long log. Reset the ECU prior to logging, allow the car to come up to temperature.

Ok, great - I'll keep the same logcfg file as before, and throw in FBKC and FLKC as well. It's CAN logging, so thankfully it can take quite a number of variables before you start losing resolution. Do I need to keep the following variables, or can I ditch: A/F_Learning_Airflow_Range; C_Loop_Fuel_Targ; Final_Fuel_Base ?

 

[Td_d]

Alrighty - since driving home was a perfect opportunity to start logging, I altered the latency table values to 0.75 of each entry and flashed the ROM, plus reset. Here are the results - town and highway driving. Initial impressions - very lumpy at the bottom end, and idled very poorly (and low at 450-500RPM) - kept on wanting to stall (actually did one time - you'll see there's a break in the log).

http://dl.dropbox.com/u/3099895/TD_D_Log_1_Feb_11_75_Perc_Latency.zip

(I landed up having to zip the two CSV files up, combined into an Excel file it was 10Mb combined, Microsoft FTL)

 

[HolyCrapItsFast]

Unfortunately I have to wait till I get home tonight to see it. I will make adjustments to the tune but in the mean time you can flash Back to the previous map so you can drive it around.

 

[Td_d]

No worries - still struggling to upload it to Dropbox in any case... the time difference might actually work in my favour here - your home time would be middle of the night - so I can probably flash the ROM before work, and get the next set of logs in.

 

[HolyCrapItsFast]

Ok!!! After many hours I have something for you to try...

http://www.drift-ready.com/Files/TD_D/2-02-2011.zip

This is another adjustment to latency. Also to offset the latency adjustments I increased the injector size in hopes of bringing the trims back to something normal.

Beyond that I made adjustments to the primary and failsafe fuel tables. The transitions from one cell to another should be much smoother.

In any case I would like it if you could try this map and log again. If it runs poorly that is okay. Just run it long enough to log some data. After that you can reflash back to a working map. We are not quite there yet but I feel we are making good progress. Also this time and you add IAM to the log? Thank you!

I included all of my working files in the zip file along with your rom.

 

[Td_d]

Thanks George - flashed it this morning, and have just done a drive on the way to work. I would like to get some more logs done hopefully over lunch time, and go through the spreadsheet workings. Initial impressions - there's already a noticeable improvement in driveability on the low end over the first map. Idle is still a little funky, but better - idling higher now - 600-700RPMs, but if you depress the clutch coming to a stop from higher revs (i.e. 3000 and up), the needle will dip close to stalling (it did actually stall once) and then recover. From lower revs, it's actually idling much better. I noticed something else that might be a clue - in 4th gear, going up a hill - i.e. higher engine load, at or close to the CL/OL switch (3000 RPMS) with partial throttle, there's bad bucking.

Thanks!

 

[Td_d]

Interesting - so I see the flow has been increased to 750, the rated flow, and you've dropped the latencies across the board - not significantly at the 'operating range' of 11.5V to 14V, but quite significantly at the margins, and compensated for manifold pressure at 14V. From the spreadsheet, I see that the trim errors at the low end have cleared up quite a bit (and across the range) and that top end now has the bigger errors margins.

So - since I'm learning here ;-) - the Primary Open Loop Fueling table effectively determines the required fuel mix after the air quantity (based on the MAF scaling) has been determined - after the various mechanics of whether the ECU should switch to open loop have already been determined? So in essence this is the critical table (along with MAF scaling) determing the A/F across the range?

 

[Td_d]

I'm not sure if I'm doing it right - but I've applied the filtering rules you mentioned, and on the morning run this graph came out - basically, all trims are now positive - spike at the bottom, but mostly at the top end, so I gather the ECU's riching it up significantly. I have not incorporated AF correction #3 - how exactly does that fit in? I gather it's based on the rear O2 sensor?



Based on these 2 logs - will do more - http://dl.dropbox.com/u/3099895/TD_D_Log_2_Feb_11_revised latencies and fueling.zip

Would my logic be correct that assuming the latencies are approximately right now, in scaling in the MAF, a positive trim effectively means the ECU is adding in fuel (i.e. removing air) - in other words it thinks it is getting less air than what it actually is at the various voltages, and scaling would in essence involve increasing those values in the MAF scaling table? Lots of questions, I know

EFIT: Ok - managed to log some highway time over lunch - http://dl.dropbox.com/u/3099895/TD_D_Log_3_Highway_Feb_11_revised latencies and fueling.zip

Busy crunching the numbers, doesn't look like it's changing the graph shape much - as soon as I have it, will post.

Here's the new graph - the correlation, other than the spike at 1.13V is uncany! Does that mean there's more scope to tweak the latencies?

 

[HolyCrapItsFast]

Thanks George - flashed it this morning, and have just done a drive on the way to work. I would like to get some more logs done hopefully over lunch time, and go through the spreadsheet workings. Initial impressions - there's already a noticeable improvement in driveability on the low end over the first map. Idle is still a little funky, but better - idling higher now - 600-700RPMs, but if you depress the clutch coming to a stop from higher revs (i.e. 3000 and up), the needle will dip close to stalling (it did actually stall once) and then recover. From lower revs, it's actually idling much better. I noticed something else that might be a clue - in 4th gear, going up a hill - i.e. higher engine load, at or close to the CL/OL switch (3000 RPMS) with partial throttle, there's bad bucking.

Thanks!

This is working exactly as I expected. You may notice as time goes by that things will smooth out because the ECU is learning the corrections. You can solve the idle issues if you follow this procedure when reflashing or resetting the ECU...

-Start the car and let it fully warm to operating temperature.
-Reset the ECU.
-Reflash the ECU.
-Start the car and DO NOT touch the gas peddle.
-Let it warm to temp and observe A/F correction #1. Once that starts to register changes let the car idle some more for 2 minutes.

The ECU will learn idle trims pretty quickly.

The stumble at 3000 rpm will get better as we continue on with the tuning

Interesting - so I see the flow has been increased to 750, the rated flow, and you've dropped the latencies across the board - not significantly at the 'operating range' of 11.5V to 14V, but quite significantly at the margins, and compensated for manifold pressure at 14V. From the spreadsheet, I see that the trim errors at the low end have cleared up quite a bit (and across the range) and that top end now has the bigger errors margins.

So - since I'm learning here ;-) - the Primary Open Loop Fueling table effectively determines the required fuel mix after the air quantity (based on the MAF scaling) has been determined - after the various mechanics of whether the ECU should switch to open loop have already been determined? So in essence this is the critical table (along with MAF scaling) determing the A/F across the range?

That compensation is a mistake. Thanks for catching that. They should all be the same. No worries though. This won't effect logging any. I will fix that in the next revision

It looks like your getting it all just fine. You will see the big picture of what we did once you see all of the files together over the course of this session. You will see how the adjustments effect the graph and why those changes are made.

The fuel table is the first thing the ECU looks at to determine AFR since this is the desired target. Then compares that to the mass of air and, if it differs, makes a correction. So yes this is the one table you have to get right since everything else depends on it.

I'm not sure if I'm doing it right - but I've applied the filtering rules you mentioned, and on the morning run this graph came out - basically, all trims are now positive - spike at the bottom, but mostly at the top end, so I gather the ECU's riching it up significantly. I have not incorporated AF correction #3 - how exactly does that fit in? I gather it's based on the rear O2 sensor?

Only concern you self with #1 Nothing else matters here.

View attachment 714

Based on these 2 logs - will do more - http://dl.dropbox.com/u/3099895/TD_D_Log_2_Feb_11_revised%20latencies%20and%20fueling.zip

Would my logic be correct that assuming the latencies are approximately right now, in scaling in the MAF, a positive trim effectively means the ECU is adding in fuel (i.e. removing air) - in other words it thinks it is getting less air than what it actually is at the various voltages, and scaling would in essence involve increasing those values in the MAF scaling table? Lots of questions, I know

EFIT: Ok - managed to log some highway time over lunch - http://dl.dropbox.com/u/3099895/TD_D_Log_3_Highway_Feb_11_revised latencies and fueling.zip

Busy crunching the numbers, doesn't look like it's changing the graph shape much - as soon as I have it, will post.

Here's the new graph - the correlation, other than the spike at 1.13V is uncany! Does that mean there's more scope to tweak the latencies?

View attachment 715

We still need to adjust latency more. You want that slope (the line you draw through the center) to be flat.

I can see that we will need to apply even less latency which will even out the slope, and add more (I mean reduce) injector which will drop the global trims closer to zero.

You seem to be understanding this all quite well! :tup:

Once we iron out latency we can work on perfecting injector scale and and then fuel trims. We're getting there

 

[HolyCrapItsFast]

Oh I also made another mistake. I increased the injectors size when I should have reduced the injector size to effect a longer pulse width. I won't make that mistake again.

Here is the next revision. Please give this a try and some moar logz pleez!

http://www.drift-ready.com/Files/TD_D/750RC_TD_D_02-02-2011B.zip

 

[Td_d]

Oh I also made another mistake. I increased the injectors size when I should have reduced the injector size to effect a longer pulse width. I won't make that mistake again.

Heaven forbid ;-)

I'm having a great deal of fun doing this, and learning at the same time. I thought that the latencies would need to still be altered - roughly, the latencies affect the slop of the correlations, whilst the flow moves it up or down, and MAF scaling will smooth it out? Ah, remembering days of statistics 101, beta and alpha...

Jeesh, ninja editing there! Great, busy downloading it, will flash and get some more logs. I wonder how many people are beginning to follow this thread out of curiousity? I'm having so much fun, it's distracting me from my work!

 

[Td_d]

Interesting - it looks like I managed to clean the data up and plug it into the spreadsheet fine - any idea why I had that huge spike up at the beginning? Did you clean up some of the outliers?

 

[HolyCrapItsFast]

All I did was filter the data as prescribed and plug em in.

Did you combine both the A/F correction and A/F learned?