T(h)or's GT86 hammer (Europe) - essential track setup

[nikitopo]

As a side question, how is dyno tuning get around ECU learning?

You make the correct questions. Usually, they don't handle the ECU learning thing. They just set IAM to 1.0, instead of 0.7 which is the factory setting, and dyno the car directly after the flash. This means that you don't know if the gains are still there a 100% after the ECU adapts to new parameters. I had a discussion some months ago with a racing driver in Nürburgring and he was mentioning that if you want the best results, then you need to install a new ECU to an open one. By changing the ECU, you have the advantage that you can change the parameters live during the dyno (you change the parameters while the engine is running) and you have more control how the ECU will behave afterwards! The factory ECU in certain areas is a black box. Unfortunately, the cost is very high for my tastes. Approximately 2,000 EUR for the new ECU + new harness wiring and an additional 1,250 for the ECU mapping by a professional.

[Wayno]

Originally Posted by nikitopo

You make the correct questions. Usually, they don't handle the ECU learning thing. They just set IAM to 1.0, instead of 0.7 which is the factory setting, and dyno the car directly after the flash. This means that you don't know if the gains are still there a 100% after the ECU adapts to new parameters. I had a discussion some months ago with a racing driver in Nürburgring and he was mentioning that if you want the best results, then you need to install a new ECU to an open one. By changing the ECU, you have the advantage that you can change the parameters live during the dyno (you change the parameters while the engine is running) and you have more control how the ECU will behave afterwards! The factory ECU in certain areas is a black box. Unfortunately, the cost is very high for my tastes. Approximately 2,000 EUR for the new ECU + new harness wiring and an additional 1,250 for the ECU mapping by a professional.

Sounds like he's a vendor for ecus. Factory ecu is very capable and hasn't been a black box since maybe a month or 2 after release. Ign timing and iam is the very very least of your worries when tuning.

[Wayno]

Ok, so that 1.02 log is what I used to create the current MAF scale you have now, so it should settle right on top of commanded AFR upwards of 3800.

My car adds about -3 FLKC in areas with petrol on track compared to on the street, without any IAM drop. You would need to pull about 2 degrees out of base timing to make it "knock free" on track, which is de-tuning the car tremendously. That's why I say don't bother. Note that only applies over 5k, if you get -3 FLKC below 5k it will drop the IAM like a rock!

If you run the regular 98 map it may still pull some timing at track temps with your fuel, but will not drop IAM. At the end of the day you can run more and let the ECU pull, or run less and have less power on the street. Either way you'll end up with around the same lap times. Again, ANY timing you add will get pulled at the track at high temps.

The only thing you might want to do is add more timing below 4000 for the street. But if you don't drive it on the street like you say, then that might be a waste of time too.

Petrol on track:

http://datazap.me/u/wayne/stg2-ace-9...om=12687-13952

Compared to only pulling about -1 FLKC for E85 due to fuel knock resistance:

http://datazap.me/u/wayne/stg2-ace-e...om=12956-14001

[Tor]

Originally Posted by Wayno

Ok, so that 1.02 log is what I used to create the current MAF scale you have now, so it should settle right on top of commanded AFR upwards of 3800.

I don't know what I am doing wrong.

I drove the car 75 km yesterday, 100 km to the track today. Did two 15 min sessions. Drove to get something to eat. Drove 100 km back home. Besides on the track all very light footed and in lots of different rpm ranges. Slow pulls, steady pace, varied pace etc. How exactly am I supposed to drive to get it to learn?

After all this almost 300 km driving and doing a log 5 minutes before arriving at home, I get this:

http://datazap.me/u/tor/tor-106-pull...7&zoom=223-331
+4 LTFT and AFR 11.27.

Is my ECU having a learning disability? Or am I a bad teacher?

Or could this be due to temperature swings? 30-35 deg at day (where most learning was done) and 20 deg in the evening (where I did the log)?

Or could other factors be at play, from free fantasy like injectors being poorly calibrated?

My car adds about -3 FLKC in areas with petrol on track compared to on the street, without any IAM drop. You would need to pull about 2 degrees out of base timing to make it "knock free" on track, which is de-tuning the car tremendously.

That's why I say don't bother. Note that only applies over 5k, if you get -3 FLKC below 5k it will drop the IAM like a rock!

In a previous post (about why I shouldn't do pulls from 2k rpm), you mentioned the Stg 1 tune was for catless headers in that area. That's why I decided to keep the 98 timing table below 3600 rpm (where I never drive anyway, except for cruising) and make a hybrid table.

If you run the regular 98 map it may still pull some timing at track temps with your fuel, but will not drop IAM.

The hybrid map seems to work pretty well? And I only run slightly less timing than the 102 table (-0.35 except for a few cells with -0.70) up in the high rpm range.

If the full 102 table would result in more FLKC, wouldn't that be subtracted from total timing anyway? Say I re-add the 0.35 deg but it results in -0.65 FLKC, wouldn't my total timing be less?

If I have to flash again because of the MAF, I will consider just flashing the 102 timing table straight up based on your explanation and the example logs you posted. Maybe I worry too much about the FLKC...

Thanks again for your help.

[Tor]

As mentioned above (keeping the engine stuff separate), I was at the track again today and drove two 15 min sessions.

Before going, I went to my alignment shop and had camber increased from -2.5 (2˚30") to -2.75 (2˚45"). Also, I had my slight front toe-in changed to slight toe-out.

Optimally, I would have preferred to change only one parameter at the time. But with time and cost constraints one has to compromise. Now I can only speculate which change led to what, and that annoys me. But trying my best not to get hung up on details (which we already have established is an issue for me ), the good news is that it had a major positive effect on both handling and tire temperature.

I was a bit wary driving off. Because frankly toe out feels weird at first! When going on a normal straight road, it feels unpredictable. And I started to have second thoughts.

Arriving 1 1/2 hours early I decided to go to the Nordschleife to have a look and send @Iceman a photo of Döttinger Höhe (the main straight) in beautiful sunshine.



Strangely there weren't that many people there and I was tempted to do a lap. But I chickened out (again) due to the alignment changes.

Over to the Grand Prix circuit, I went... Man this is a luxury to have such a choice.

The first session was an absolute disaster, with a mix of bikes and cars. Also, I was expecting them to send the bikes out separately so I wasn't prepared as we were sent off. So no video. Dodging bikes and other confused drivers don't make for the most interesting video anyway. I had to drive strange lines and had several ABS stops which I think fucked up my temperature data.

The second session was a lot better, as they had separated the bikes into their own session. The unpredictability of the toe-out going straight, transforms as the wheels are turned. Gone is the pushing over my left front wheel. Once the car is turned in, the toe out acts completely opposite. Now it adds a surefooted feeling like the outside tire is stuck in an imaginary groove that pulls the car around - predictably.

There is still understeer, but it has greatly diminished. Like if throwing the car into the turn and still if going too early on the throttle. That's fine too, as the car doesn't feel scary that way. At low speed, turning in, the rear feels more lively. This is just what I wanted!!

Tire temperature looks good too:

August 23rd:
-2.8 Camber, 0.5 mm toe-out

Session 1:
35-39				Cold approx tire temps

Front:
2.15		2.15		Cold pressure
2.54		2.46		Hot pressure
64 67 76	72 66 62	Hot temp O M I I M O (left to right)

Rear
2.15		2.15
2.48		2.45
61 62 65	64 61 57


Session 2:
30-33				Cold approx tire temps

Front:
2.12		2.04		Cold pressure
2.45		2.35		Hot pressure
62 60 66	61 57 56	Hot temp O M I I M O (left to right)

Rear:
2.17		2.10
2.46		2.40
54 58 61	59 56 52

First session, the front insides got a bit too hot. My guess is that it's caused by the ABS braking?

Second session, it looked pretty good but some fine tuning needed. Also I am getting better at using the Alfano Tyre Control, so I think these temps are the most reliable.

Now I have a basic setting I can use, instead of setting the pressures blindly. Next time I will start out with 2.2 Bar cold in the front and 2.15 in the rear. I am still suprised at the fairly high pressures.

A thermometer with a needle probe is, in my opinion, a MUST-have for anyone who is tracking and wants to know what their suspension and tires are doing. The Alfano makes it very easy.

Ohh, and the lap times (with traffic) of second session:
2:40.45
2:40.86
2:41.85
2:40.75


I am very happy with the consistency, and that making changes doesn't cost me time. Also, I don't expect doing a few changes will magically make me 5 seconds faster. In fact, I don't even think one is necessarily better than the other... But it feels better.

[nikitopo]

Originally Posted by Wayno

Sounds like he's a vendor for ecus. Factory ecu is very capable and hasn't been a black box since maybe a month or 2 after release. Ign timing and iam is the very very least of your worries when tuning.

Nope he isn't the vendor. He is a racing in Nürburgring (not during tourist days) and he is one of the best in his class. In fact we would need to go to Belgium for the ECU change. The vendor was located there. Factory ECU is designed for normal driving, not for racing. The ECU code is not open source and we don't know all the small details. Reverse engineering can provide an insight, but up to a point. If you want to be serious with these things, unfortunately you have to pay a lot of money. It is an expensive sport. If you want to have some fun or if you running in a class that the car should be stock, then it is fine.

[Tor]

Something different for a change. I guess it's becoming pretty boring watching me drive around the GP circuit.

And I felt more in the mood to do a video edit from today's car washing (though I will upload the video from yesterday's second session as well).

[ame="https://www.youtube.com/watch?v=UWTtm3TWTKs"]Caring for my baby... - YouTube[/ame]

[Wayno]

(too much) Toe out on the front can kill the inside of your tyres in a couple laps

[Wayno]

Originally Posted by Tor

After all this almost 300 km driving and doing a log 5 minutes before arriving at home, I get this:

http://datazap.me/u/tor/tor-106-pull...7&zoom=223-331
+4 LTFT and AFR 11.27.

I don't know why it's compensating for the ports or directs so much, could be any number of things. Most likely is sitting in one specific cell somewhere for too long that skews the learning. Or that the PI is scaling horribly at high loads around 2000 rpm.

It's clear from this and an earlier log with enough miles on it though the difference is about 6.4%-6.6%.

http://datazap.me/u/tor/wayno-stg-1-...zoom=4040-5265

Instead of worrying so much about it, here's another solution. I'd run this as is but you can add whatever timing you want to it. The AFR will be around 11.8 and LTFT around 0.4 up to 7000 rpm where your power drops off on stock header anyway. The lack of PI at 2000 rpm might take away the ECU's ability to learn those high LTFT values above 5k/7k, but if it doesn't this will let you run correct AFR, regardless of the LTFT issue is resolved or not.

[Tor]

Originally Posted by Wayno

I don't know why it's compensating for the ports or directs so much, could be any number of things. Most likely is sitting in one specific cell somewhere for too long that skews the learning. Or that the PI is scaling horribly at high loads around 2000 rpm.

Thanks a lot. I am working late shift for the next days so I'll drive the car and see give it a chance to see if it changes. With the stock A00G it did it too:
http://datazap.me/u/tor/stock-a00g-m...&zoom=450-1488

If it's scaling poorly at 2000 rpm, is that just a natural variation in the production quality of the injectors?

I understand that I that I am potentially losing some power at high rpm. But how rich does it have to be before it becomes a problem in terms of carbon deposits or fuel getting into the oil? I assume the values I have are far far away from causing such problems?

I should add the car runs great, from feeling.

Instead of worrying so much about it, here's another solution. I'd run this as is but you can add whatever timing you want to it. The AFR will be around 11.8 and LTFT around 0.4 up to 7000 rpm where your power drops off on stock header anyway. The lack of PI at 2000 rpm might take away the ECU's ability to learn those high LTFT values above 5k/7k, but if it doesn't this will let you run correct AFR, regardless of the LTFT issue is resolved or not.

I see that you made the following change in v98:

v98:
- Restored 20% Port Injection > 5000 rpm in petrol tunes to prevent carbon build up on intake valves.

The rpm scaling is different from e.g. OTS and the 20% is introduced at 7200 rpm instead of 5200. Could that influence it as well?

[Tor]

Originally Posted by Tor

I am working late shift for the next days so I'll drive the car and see give it a chance to see if it changes.

Problem solved:
http://datazap.me/u/tor/tor-106?log=...zoom=1172-1329

I just drove it normally two days to and from work, 100 km in total. Could it be that I babied it too much during learning and never reach the volts/loads required? If that's the case it renders many of my previous logs invalid.


As promised the video from last Tuesday:
[ame="https://www.youtube.com/watch?v=jBuEeUIBCdI"]"The only negative in your life should be camber and toe" - YouTube[/ame]

[Wayno]

Yeah, i don't know for sure the logic it uses to average the top bracket fuel trims, only know the area it does it at. Best to do another log every 500km or so and see where it moves around.

[Tor]

Originally Posted by Wayno

Yeah, i don't know for sure the logic it uses to average the top bracket fuel trims, only know the area it does it at. Best to do another log every 500km or so and see where it moves around.

I will do. Thanks for all your help.

And thank you very much for suggesting me to get into this in the first place. It is really lots of fun and a great learning experience.

[Wayno]

So, what the hell, 3% difference over a week?

+1.2
http://datazap.me/u/tor/tor-106?log=...zoom=1072-1394
- 2.1
http://datazap.me/u/tor/tor-106-0?lo...4&zoom=493-835

I still reckon the AU.Rich scale will give more consistent trims, being that it's way less lumpy compared to the openflash scale.