PDA

View Full Version : Nick Manns ECU flash learning curve!



Nick Mann
27-02-2013, 05:09 PM
*Disclaimer*
Whilst I understand some of the basics of tuning a car, I am not a car tuner. Nor am I particularly computer literate. Therefore, please don't take anything I say below as gospel - I could well have got the wrong end of the stick! This thread is to help others like me have a bash at DIY tuning, but I am aiming to talk about what I am doing so others can correct/advise/p!sstake so we all learn. I am not aiming to set this out as the best way to tune cars!!

About a year ago, Jason flashed my ECU using ECU flash, my laptop and open port 2.0 cable. Obviously, I bought the cable to have a go at doing it myself, but the first time can be scary! (Oooerr!) Last week I finally forced myself to have a bash at doing it myself, and spent a couple of hours playing to try and get a reasonable result at last weekends dyno day (http://www.clubvr4.com/forum/showthread.php?66786).

One thing is instantly clear - boost control is tricky with a 3 port solenoid! I ended up doing a "dirty" tune, effectively just tinkering with the wastegate duty cycle to increase the boost. Whilst this does work to a certain extent, you get massive boost fluctuations which aren't repeatable, even if evoscan is reporting consistent wastegate duty cycles. So I headed down to Eurospec knowing my tune was dirty, and knowing the boost could hit anything from 0.9 to 1.2 bar.
60358
60359
I was relatively pleased with my results - 243 bhp at the wheels is one of my best ever results at a dyno day. However, the car was running slightly leaner than I would like (my wideband thinks it is in the high 11's, Bens was reading it in the low 12's) and the boost curve was erratic on all three runs, and again, not in a repeatable way.

After a chat with Ben, a chat with Jason and a quick look round on the interweb, it becomes clear that the ECU doesn't control the wastegate solenoid in the same way that a boost controller does. It looks like the ECU has a target load point, which influences the way it controls the wastegate solenoid. Now, I have complicated the issue by installing a 3 port solenoid which doesn't react the same way as the stock one. So now, the ecu is trying to achieve a target load and is at least in part trying to do that by controlling a solenoid that is acting differently to the solenoid it was programmed to control!

So what now? Well, I think that the next step is to change other parameters in the ECU that will change the way the ECU controls the boost solenoid. In effect, I think that means that the ECU is being tuned to match the solenoid. Fortunately for me, I am friends with Jason who has already given me the settings he used with his identical 3 port solenoid. All I need now is some time to play again! As is probably clear from the graphs, if I can control the boost properly, there is potential to get better numbers.

fassi1
27-02-2013, 05:57 PM
Noticed the same thing with AFR, Bens was higher than from stock O2 port.

Kenneth
27-02-2013, 07:25 PM
You might have some joy setting the correction tables to 0 and then setting boost with the wastegate duty cycle.

The problem with remapping the boost is that the load target is checked and then correction factors applied so that it is difficult to work out if your base setting did anything useful to begin with. The closer your base duty cycle is to what you want, the smaller the corrections and more effective it will be at keeping a nice boost curve.

Once you have good base settings, reverting the correction tables should result in a fairly stable boost curve.

adaxo
27-02-2013, 08:52 PM
Eventually some one start thread about flashing for novices /thankyou

What is a bit odd is that you both fassi1 and Nick Mann had spike right on 3000 RPM your both graphs look pretty similar (curve shape wise) but both got spike at the same rpms, its ok for auto owners as torque converter do this (spike) but its a bit odd on manual???

Nick Mann
27-02-2013, 10:17 PM
My spike is my boost control being poor, Adam. Check my boost curve on the first graph.

Ken, that is useful info. I hadn't thought of that. Jason has given me all of his settings which I thought would be a good start, but I might just try that too. I just tried to give you rep, but apparently I have already given you too much recently!!

swinks
27-02-2013, 10:28 PM
Nick, boost, afr and power curve looks almost identical to my one of the first graphs I got in Eurospec years back. Power output was however different, much lower. Over all the issue was running lean under load and ecu pulling boost (via selenoid) and timing (no knocking recorded). IIRC, Ben had to throw massive fuel enrichments in high revs (+15 to +20 on SAFC) to keep it civilized and get down to afr 11.2 to get smooth power curve.

Nick Mann
27-02-2013, 10:49 PM
That's interesting. But how does the ecu know that the car is running lean on open loop?

fassi1
27-02-2013, 11:02 PM
That's interesting. But how does the ecu know that the car is running lean on open loop?
It doesn't know, unless narrow signal is lower than 0.8-0.9V but even knowing that will not react to it by changing fueling.

CANDEE
27-02-2013, 11:33 PM
Nick was Ben using an up the exhaust pipe afr sensor? If so as far as im aware they can have a value difference to an afr sensor in the factory location.

swinks
27-02-2013, 11:47 PM
It doesn't know, unless narrow signal is lower than 0.8-0.9V but even knowing that will not react to it by changing fueling.
Chris it's timing and boost change, not fuelling.
On many occasions seen timing being retard without knocking happened, at least not significantly recorded.

Kenneth
27-02-2013, 11:53 PM
It is fairly normal for all modified VR-4s to do this. The factory tune tries to raise boost quickly to and then regulates it to produce a fairly smooth power curve.

~3500RPM is around where you can expect the engine to be able to drive the turbos for full boost in 3rd gear with a standard exhaust and so up to this point the ECU is running high duty cycle on the solenoid to improve spool.

Once you add a modified exhaust, the turbos spool up faster and tend to get a spike at 3000 (which is around where you can expect full boost with a reasonably modified exhaust), probably because the ECU is programmed to be building boost at that RPM rather than tapering off to prevent a boost spike. At this point the boost error correction comes into the picture and will bring the boost back down. The 1second (1000ms) delay on fuel cut allows this behaviour without any real problems, provided you are not getting detonation.


Eventually some one start thread about flashing for novices /thankyou

What is a bit odd is that you both fassi1 and Nick Mann had spike right on 3000 RPM your both graphs look pretty similar (curve shape wise) but both got spike at the same rpms, its ok for auto owners as torque converter do this (spike) but its a bit odd on manual???

fassi1
28-02-2013, 12:13 AM
It is fairly normal for all modified VR-4s to do this. The factory tune tries to raise boost quickly to and then regulates it to produce a fairly smooth power curve.

~3500RPM is around where you can expect the engine to be able to drive the turbos for full boost in 3rd gear with a standard exhaust and so up to this point the ECU is running high duty cycle on the solenoid to improve spool.

Once you add a modified exhaust, the turbos spool up faster and tend to get a spike at 3000 (which is around where you can expect full boost with a reasonably modified exhaust), probably because the ECU is programmed to be building boost at that RPM rather than tapering off to prevent a boost spike. At this point the boost error correction comes into the picture and will bring the boost back down. The 1second (1000ms) delay on fuel cut allows this behaviour without any real problems, provided you are not getting detonation.

That sounds like explanation to those spikes at low rpm. I do easily get 11.5 psi running stock solenoid and boost but only at low rpm.

Tomek Nick's question was "But how does the ecu know that the car is running lean on open loop? " and that sounds to me like fueling question.

Kenneth
28-02-2013, 12:27 AM
As already said, it doesn't.
However, if you get detonation, then it will take steps to protect the engine, one of which is adjusting the octane rating which can in turn alter the AFR, retard timing and lower the boost.


That's interesting. But how does the ecu know that the car is running lean on open loop?

adaxo
28-02-2013, 12:47 AM
So looks like everything here is pretty straight forward :oops::laugh:


Nick, or anybody who is still in the nappy (as Im) on subject of flashing/mapping, can you put step by step how to in fairly clear, for novice, language ?please


its something like it possible or its just wishful dreaming and cant be explain so easy?

Kenneth
28-02-2013, 01:03 AM
Would you want someone doing surgery on you based on a simplistic explanation of how anatomy works?

Mucking about with the tuning can cause severe problems, potentially total engine failure. Because of this, not many people are willing to put their name to a simplistic explanation which leaves room for such mistakes.

There are plenty of online resources (look for Evo 5/6 tuning guides) which give you information about what there is available and why you use it, then if you are willing to, start working on your car.
If you then have specific questions about something, that is the time to do either some specific searches or, if that fails, start a thread.

Sorry I couldn't be more helpful, but I am not a "tuner" either (more a hacker ;) ) and wouldn't want to put my name to anything in terms of guidelines :P



its something like it possible or its just wishful dreaming and cant be explain so easy?

adaxo
28-02-2013, 01:10 AM
Ha, setup new account as a NoName/lol, honestly I was expecting this is not very easy to explain, I go through merlin guide but its just far too much to remember and understand to me, what i was hoping for its like, plug in cable start this, check that/change this/run/log/ DO NOT touch/change/log this ect.

Im probably just to lazy/busy/old for all this:stars:

Kenneth
28-02-2013, 01:15 AM
In saying that, I can tell you what I did to begin. Not saying this is what you should do, but it got me doing stuff and I was able to move along from there

1) Get wideband installed and logging with your logging software (Evoscan)
2) Do 2 byte RPM and Load mods to MUT table (you of course can download a ROM which already has it)
3) Log AFR, RPM, Load and knock values
4) Play with AFR values to get ~11.5:1 AFR minimum
5) Adjust ignition to retard timing to remove any knock where knock is reported
6) Adjust ignition timing adding advance on other areas of the map where I drive at low loads
7) Smooth out ignition table

8) Increase boost
9) repeat steps 3-7

This is by no means an ideal tune, you might be able to add more fuel to combat detonation instead of retarding timing for more power. You can also look at other ways to combat detonation such as smoothing ignition / fuel transitions around detonation to see what effect it has on the problem cell(s) to see if you can get away with more ignition advance etc.

There is a lot of stuff you can do which comes down to your personal preference, so it pays to get a start, work out what you want to do and then do more research. Then more research.

One day, you will be writing posts on the forums and giving advice, surprised at how everyone thinks you know what you are talking about /haz

Kenneth
28-02-2013, 01:18 AM
Basic stuff like plugging in cable can be found on the internet pretty easily, just have to use the terms such as ecuflash, evoscan and openport interspersed with your question :)



Ha, setup new account as a NoName/lol, honestly I was expecting this is not very easy to explain, I go through merlin guide but its just far too much to remember and understand to me, what i was hoping for its like, plug in cable start this, check that/change this/run/log/ DO NOT touch/change/log this ect.

Im probably just to lazy/busy/old for all this:stars:

adaxo
28-02-2013, 01:28 AM
In saying that, I can tell you what I did to begin. Not saying this is what you should do, but it got me doing stuff and I was able to move along from there

1) Get wideband installed and logging with your logging software (Evoscan)- done/woot/haz
2) Do 2 byte RPM and Load mods to MUT table (you of course can download a ROM which already has it)
3) Log AFR, RPM, Load and knock values
4) Play with AFR values to get ~11.5:1 AFR minimum
5) Adjust ignition to retard timing to remove any knock where knock is reported
6) Adjust ignition timing adding advance on other areas of the map where I drive at low loads
7) Smooth out ignition table

8) Increase boost
9) repeat steps 3-7

This is by no means an ideal tune, you might be able to add more fuel to combat detonation instead of retarding timing for more power. You can also look at other ways to combat detonation such as smoothing ignition / fuel transitions around detonation to see what effect it has on the problem cell(s) to see if you can get away with more ignition advance etc.

There is a lot of stuff you can do which comes down to your personal preference, so it pays to get a start, work out what you want to do and then do more research. Then more research.

One day, you will be writing posts on the forums and giving advice, surprised at how everyone thinks you know what you are talking about /haz


Thanks, this give me some points for a starter, what im after is to tweak my car to run more efficient/power on lpg, lets say I can supply to the engine unlimited (to some extreme point, like three injectors per cylinder) amount of very knock resistant fuel (~106 octans) so in my thinking, things to tweak will be boost AFR and timing, yes? like increase boost as fas as our tubs can handle , then set lpg ecu to supply enough fuel to get afr right and then play with ignition till knock become present??

Kenneth
28-02-2013, 01:35 AM
I would be looking at fuel, timing and then boost. Boost changes the load range you get to, so tuning in the lower ranges will show you some trends and let you extrapolate some at least reasonable values for the higher load range cells, before you actually reach them.

If you change the AFR, then get detonation, you may have to fiddle with the timing before continuing with the AFR.

/edit: And to repeat what I said earlier, I am not a tuner so this is based on what seems to have worked for me, I do not rate my tuning experience to anything near professional, so if you want advice you can bank on, you should talk to a professional (and probably pay them)

Davezj
10-03-2013, 12:21 PM
which rom are people going to start there tuning with?
which would you recommend kenneth? Kenneth

what i mean is are you going to modify the orignal rom in the ECU or go for something like Kenneth's 1.03 rom.
i will be using kenneth's rom 1.03 as the 2 byte load mod is already done amoungst other things and engine management light flashing on detecting Knock which will be very useful when playing in the car.
what i don't know is, if there were any updated need to kenneth's 1.03 rom and def file which were found after the initial release have been put into the downloadable version that is in the first post of the thread. or if any bugs or mods were found when the 2.0 rom release was done for multiple map switching are applicable to the 1.03 rom and def file.

aany info on this would be great kenneth.

Nick Mann
10-03-2013, 01:07 PM
Yes, I should have said that I am using Kens 1.03 rom.

Kenneth
10-03-2013, 11:19 PM
The 1.03 ROM is good for starters. I would probably always at least use this ROM just for the CEL on KnockSum ability.

The 2.0 ROM is a totally different beast, It does away with most of the code in the 1.03 ROM because it became redundant with the ability to switch between configurations on the fly. The actual ECU (in terms of what the processor does) operation is simpler, but the cost is complexity in configuration.

If you don't have multi-fuel needs or change boost profiles etc, 1.03 should do what you want no problems.

fassi1
11-03-2013, 12:35 AM
Just comparing Merlin's and Kenneth's MUT tables for 2-byte logging and Kennth's MUT00=F0CE and MUT01=F0CF
when Merlin's are MUT00=F008 and MUT01=F009
Why are they different and which ones are right? I guess Kenneth knows the answer /Hmmm

foxdie
11-03-2013, 09:44 AM
Just found this thread, quite a potent one :)

I've only skim-read the replies (as I'm in a rush), so if I'm retreading the same ground please forgive me.

As previously mentioned, VR-4 ECUs have no concept of manifold pressure like newer cars, so it has to estimate it with some bastardised calculations based on air volume passing via the MAF vs the RPM of the engine vs other factors (such as various temperatures, knock etc). This is known as "Load" in the tuning world.

Those spikes you're seeing Nick are the ECU trying to match a particular load target, overshooting it, scaling back, undershooting etc. It's a symptom of setting the wastegate duty cycle too high plus a mixture of the turbo boost error correction tables being too aggressive).

The tables I sent you, as you know, were from Raul and every car responds differently (another fact I'm sure you're aware of). To address your issue I personally would try one of two things;

Either lower the duty cycle by about 10-15% in the affected regions - the more aggressive correction tables should help tune the duty cycle better in those regions
Or dial back the correction tables so it's not so aggressive
Personally, I think dialling the wastegate duty cycle back is the safest option, that way the ECU has more flexibility to maintain a stable load target (boost being loosely derived from this).

On Raul, I observed the following load targets making the set amount of boost;


USE THIS WITH CAUTION - ITS A GUIDELINE ONLY!
60665
USE THIS WITH CAUTION - ITS A GUIDELINE ONLY!


That said, as Kens suggested, we could try you with his 2.0 ROM Nick. I've got all this week off if you want me to help (in fact I'm mapping someones car this morning if you wanna pop around?) ;)

Kenneth
11-03-2013, 08:26 PM
Merlin's ones were probably the address for Evo5/6.
Mine should be correct. However if you are unsure, you can easly log both and see which one better matches the single byte load


Just comparing Merlin's and Kenneth's MUT tables for 2-byte logging and Kennth's MUT00=F0CE and MUT01=F0CF
when Merlin's are MUT00=F008 and MUT01=F009
Why are they different and which ones are right? I guess Kenneth knows the answer /Hmmm

Kenneth
11-03-2013, 08:38 PM
Manifold pressure sensing is WAY older than MAF, in fact just about all modern cars have a MAF because they are better under most conditions than a MAP based system. (Disregarding racing/extreme tuning)

There is no bastardisation in calculations, the MAF (when operated correctly) is fairly accurate and can measure how much air the engine is consuming. Knowing this, it isn't that difficult to calculate fuel mass requirements to achieve a specific AFR, which is why our fuel tables show AFR rather than injector duty cycle. It doesn't care about boost pressure because it is mostly irrelevant.
I have recently seen 29PSI of boost on my boost gauge (after fixing the manifold pressure supply and leak testing to 35psi) with no fuel issues or knock. Charge temps would have been somewhat inefficient, but hey.

Newer high performance cars tend to have MAF AND MAP so that they can capitalise on the benefits of both.


Just found this thread, quite a potent one :)

I've only skim-read the replies (as I'm in a rush), so if I'm retreading the same ground please forgive me.

As previously mentioned, VR-4 ECUs have no concept of manifold pressure like newer cars, so it has to estimate it with some bastardised calculations based on air volume passing via the MAF vs the RPM of the engine vs other factors (such as various temperatures, knock etc). This is known as "Load" in the tuning world.

Davezj
11-03-2013, 10:13 PM
just thought i would ask a quick question,
i have been looking at your 1.03 rom and there are 2 sets of fuel and ignition maps one set is marked up as yours and the other i presume are the original are the original maps. and you rom probably only referneces your ignition and fuel maps.
is this the case?

Kenneth
12-03-2013, 02:55 AM
That is correct.

just thought i would ask a quick question,
i have been looking at your 1.03 rom and there are 2 sets of fuel and ignition maps one set is marked up as yours and the other i presume are the original are the original maps. and you rom probably only referneces your ignition and fuel maps.
is this the case?

fassi1
12-03-2013, 08:07 AM
Merlin's ones were probably the address for Evo5/6.
Mine should be correct. However if you are unsure, you can easly log both and see which one better matches the single byte load
Will do that :thumbsup:

Davezj
12-03-2013, 01:16 PM
also just a really nubie question about flashing the 1.03 ron to a 7202 ecu as the rom was designed to flash to a 7203 ecu.
i need to change the def file to state the rom is to be flashed to a 7202 ecu by changing the romid section to of the def file, specifically the memmodel line from H8539FA to H8539F.
<romid>
<xmlid>vr4base</xmlid>
<internalidaddress>2fefa</internalidaddress>
<flashmethod>mitsukernel</flashmethod>
<memmodel>H8539FA</memmodel>
</romid>

do i have to change this in the EMKS11.xml def file as well as i believe the EMKS11 def file is called up by the VR4base.xml

also as mine is an auto do need to change the transmission type to 5AT, and model to LEGNUM as the rom uses these values to make decisions in various parts of the rom, or is that all just a nicity to tell the person reading the rom that it was designed for a particularly car and does not actually matter what is put in this section.

Kenneth
12-03-2013, 08:26 PM
All you need to do is change the memmodel as you have described.

If you have TCL, I have heard that it stops operating in some cars...

Davezj
12-03-2013, 09:31 PM
ok cheers kenneth,
the TCL thing might be an added bonus.

Davezj
21-03-2013, 11:10 PM
Nick Mann
Nick this might be of interest to you for the ecu tuning, i know you have a certain Fox close by, but this was the best set of info i could find that was directly directed to the software and type of tuning that you will be doing.
http://www.clubvr4.com/forum/showthread.php?67698-Basic-Introduction-to-ECUflash-rom-tuning-and-Evoscan-logging