PDA

View Full Version : E38 PCM tuning experiences to share....



The Alchemist
July 31st, 2007, 06:40 PM
hi there fellow forum fellows :)
been doing lots of data logging on a Holden SS VE and getting a feel for timing requirements, fuel trims, and knock sensitivity and burst knock sensitivity and have found some interesting stuff that I would like to pass on.
As we all know torque is in the timing, mostly, and the good ol factory tunes in the LS1 were average at best. With the Holden SS VE I found that it felt flat over 3000rpm at full throttle and the initial impression was one of disappointment.

Logging showed 4 to 5 degrees knock retard occuring at 3000rpm on the button leaving the rest of the fun behind. Taking advantage of the multi cylinder knock sensor logging I discovered cylinder B ( hopefully not "B" for BOMB ) was the culprit. All the others logged "0". None the less knock retard retards total timing by 4 to 5 degrees.
Timing was a retarded 12 degrees at full load at 3000rpm with the octane scalar right around at 1 due to all the "knocking" . The car made 194 kw , dyno dynamics, at the wheels on 91 octane gas. Timing at peak torque was 14 degrees (4500rpm) with no knock showing. Timing at 6000rpm was 16 degrees.
Experience and logic says, to me anyway, that if it runs 14 degrees at peak torque with no knock, that it should be able to run at least 14 degrees at 3000rpm if not more. Time to investigate Cylinder B knock sensitivity. Gave it a tweak at the 2800 and 3125rpm points, a couple of test drives and hey presto no knock at 3000rpm. The sensitivity was much more trigger happy than the other cylinder knock maps. Could be a random calibration problem, could be that the cylinder B was more prone to knock during testing > not sure but I doubt it.

Octane scalar learned closer to "0" and the car ran another 2 degrees under full load with NO knock registering anywhere on any cylinder.
The car now pulls hard pass 3000rpm right up to 6000rpm and is much more impressive to drive. During all this I was also tweaking the low & high octane timing map downward by up to 10 degrees in places !!! to stop real knock with all 8 cylinders triggering knock retard. The car was on 91 afterall. The octane scaler self adjusts during these changes and I was keeping an eye on it.
I also tweaked the low MAF calibrations by 2 to 4 % to bring long term trims down from 8% at worst to 0 to 2%. PE was also lowered accordingly as the AFR dropped out to 10.5:1 in the top end on the dyno.

I'm now ready to throw it on the dyno again to see if my seat of the pants dyno reads as well as the dyno dynamics dyno. It certainly feels way better on the road so I'm sure I'll see a good result. We shall see.
Then for some real tuning to find the timing limits on 91 octane with these engines. Will keep you posted.

Just incase this reads wrong to people I don't tune cars by looking at knock sensor readings and adjusting timing for no knock. I use minimum effective timing for peak tractive effort. Its interesting to note that the factory PCM is actually set up to do exactly that > look at knock and adjust timing until there is no knock

Its just that this was a new engine to me and new PCM and its easier making observations while on the road driving in real conditions and making some basic changes and observing response etc etc

Mike

GMPX
July 31st, 2007, 08:15 PM
I'm just glad someone logged the knock per cyl data, it was a headache adding all that in :)

Cheers,
Ross

Maggie
August 1st, 2007, 03:22 AM
I'm just glad someone logged the knock per cyl data, it was a headache adding all that in :)

Cheers,
Ross

Sorry for your headache but the ability to log knock per cyl is much appreciated !!

...:thankyou2:

RonC
August 1st, 2007, 05:28 AM
As we all know torque is in the timing, mostly, and the good ol factory tunes in the LS1 were average at best. With the Holden SS VE I found that it felt flat over 3000rpm at full throttle and the initial impression was one of disappointment.

Logging showed 4 to 5 degrees knock retard occurring at 3000rpm on the button leaving the rest of the fun behind. Taking advantage of the multi cylinder knock sensor logging I discovered cylinder B ( hopefully not "B" for BOMB ) was the culprit. All the others logged "0". None the less knock retard retards total timing by 4 to 5 degrees.
Timing was a retarded 12 degrees at full load at 3000rpm with the octane scalar right around at 1 due to all the "knocking" . The car made 194 kw , dyno dynamics, at the wheels on 91 octane gas. Timing at peak torque was 14 degrees (4500rpm) with no knock showing. Timing at 6000rpm was 16 degrees.
Experience and logic says, to me anyway, that if it runs 14 degrees at peak torque with no knock, that it should be able to run at least 14 degrees at 3000rpm if not more. Time to investigate Cylinder B knock sensitivity. Gave it a tweak at the 2800 and 3125rpm points, a couple of test drives and hey presto no knock at 3000rpm. The sensitivity was much more trigger happy than the other cylinder knock maps. Could be a random calibration problem, could be that the cylinder B was more prone to knock during testing > not sure but I doubt it.

Its just that this was a new engine to me and new PCM and its easier making observations while on the road driving in real conditions and making some basic changes and observing response etc etc

Mike

Like you Mike, I had a lot of KR with the factory tune that limited max timing to 18 degrees. I have done some street tuning with spark timing on my 6.2L and by increasing B1977 and B1978 both up to 150 along with some other changes have allowed me to get it up to 27 degrees on 91 oct with a little KR occasionally. It went from 362 ft lb delivered torque stock up to 401 with the timing changes. That made a major difference in the way it runs. I have not had any noticeable pinging. I didn't realize the potential advantage to log and change each cyl until now and I will go back to fine tune cyl KR sensitivity individually. Looks like a good improvement and we appreciate your efforts Ross.

Ron

The Alchemist
August 1st, 2007, 06:44 PM
I'm just glad someone logged the knock per cyl data, it was a headache adding all that in :)

Cheers,
Ross

Thanks Ross is was well worth it. Good idea having 8 knock sensors but only if they are calibrated correctly. Good work mate.

Mike

ringram
August 2nd, 2007, 12:30 AM
Mike I hope you mean R+M/2 Octane or MON and not NZ 91 RON!

The Alchemist
August 2nd, 2007, 07:33 AM
NZ 91 RON I'm afraid to say..... it was already in the car. I asked the owner and he said he was told it would run on either 91 or 96 when he brought it !!! The timing tables are set up for more like our 98 RON and are hopelessly advanced for 91 RON.
At least when I get this right the damn thing will be to scared to KNOCK :)
Dynoing it today and will post results for you and the other members
Mike

ntae
August 2nd, 2007, 09:19 AM
Been logging a HSV VE of late it's running 98 fuel and has not showed knock to be a problem the tune was very good from stock 12.5 afr & 22/23 deg at wot . air temps are a problem stoped at the lights i see 60/70 and down to around 30/40 driving with out side temp around 15 deg . taking off at the lights i was seeing -7 deg of air temp spark retard . a cold air system looks to be a must

The Alchemist
August 2nd, 2007, 12:17 PM
hey there ntae, you couldn't post up that tune file could you. It would be interesting to see that calibration changes between the base model and the HSV model.

Cheers,

Mike

ntae
August 2nd, 2007, 01:30 PM
The file looks to be to big for the forum i will host it somewere hold on

email your stock file to admin@ntautoelectrics.com.au i will host it as well the NZ file could be the same as the OZ

Geoff

Redline Motorsports
August 2nd, 2007, 02:01 PM
How does the B1977/B1978 get understood? How do Mv correlate to the knock sensor function?

Howard

RonC
August 2nd, 2007, 02:28 PM
How does the B1977/B1978 get understood? How do Mv correlate to the knock sensor function?

Howard

I'm not sure how it does but that was the only thing I could change that would reduce the excessive KR to let the timing come up to 27 degrees. I still don't know how it gets to 27 as my base timing is now set to 18 degrees from .72 grams per/cyl on up . I can't log anything that is "adding" to the base spark, but it works for now. That was before I realized I could log the individual cylinders for the KR, but I will try that later on when I have time to fine tune it some more.

Ron

ntae
August 2nd, 2007, 03:00 PM
hey there ntae, you couldn't post up that tune file could you. It would be interesting to see that calibration changes between the base model and the HSV model.

Cheers,

Mike


http://www.ntautoelectrics.com.au/tunefiles/


if any one has any holden e38 ecm .tun file email them to admin@ntautoelectrics.com.au and i will host them for every one to down load

GMPX
August 2nd, 2007, 04:22 PM
How does the B1977/B1978 get understood? How do Mv correlate to the knock sensor function?

Howard

Basically the knock sensor filtering needs to remove the noise that might be considered knock. The mV relates to the voltage signals from the knock sensors. If you are a musician you might know what a noise gate is?, same thing here.

Cheers,
Ross

The Alchemist
August 2nd, 2007, 06:06 PM
1951

oh good that worked ok :)
Had to post it up then check.....
20% torque increase overall, mixtures leaned out on their own once the timing improved, tweaked them to 12.5:1 after that.
Good old burst knock had to be neutered as it was pulled timing by 3 deg at 5300rpm killing the top end.
I set timing in low octane map to 2 degrees under best tolerated timing so nice and safe then added 6 deg and pasted that into high octane table > should be enough I think. Also tweaked Octane scalar increment rate so it learns upward a bit quicker.
Thats about it for an afternoons work :)

Comments welcome.....

Mike

PS: these things will be wicked once we stick a cam up them

Delco
August 2nd, 2007, 07:46 PM
1951

oh good that worked ok :)
Had to post it up then check.....
20% torque increase overall, mixtures leaned out on their own once the timing improved, tweaked them to 12.5:1 after that.
Good old burst knock had to be neutered as it was pulled timing by 3 deg at 5300rpm killing the top end.
I set timing in low octane map to 2 degrees under best tolerated timing so nice and safe then added 6 deg and pasted that into high octane table > should be enough I think. Also tweaked Octane scalar increment rate so it learns upward a bit quicker.
Thats about it for an afternoons work :)

Comments welcome.....

Mike

PS: these things will be wicked once we stick a cam up them


???? let me get this straight in my head , you set the low octane table to be 2 deg under MBT then set the high octane table to be 4 deg above MBT.

So in effect you are running the engine in knock at all times ?

Normally with the 6L we take timing out of the stock timing tables to get MBT and make power.

These engines are very timing sensitive being higher compression than the LS1 and different combustion chamber and intake port.

They are responding well to cams / displacement increase and superchargers , bit trickier to tune than the older stuff but once you have a handle on it things sort of flow into place.

The Alchemist
August 2nd, 2007, 10:37 PM
hi there Delco,
considering that the car now actually accelerates when you put your foot down I think that the results speak for themselves. I spent most of my time getting out the wrinkles so to speak in the factory tune. :Eyecrazy:

I have set the low octane table below MBT, remember the engine was running 91, with the theory that if it knocks then the octane scalar system will pull it down well past knock threshold. I allowed 4 degrees above MBT for the high octane table because we only have 91 & 96 here and I thought that 4 degrees would be plenty allowing for the odd tank of 98 if you went out of town.
The factory tune runs a spread of 10 degrees (as you would be aware)and my new low octane table is set much lower than the factory low octane table and is of a better match for what the engine wants.

Also with the factory tune the Flashscan was constantly registering "knock" and the octane scalar was ramped around to "1" all the time and still knocking under load > too advanced I believe and to quote you...

"So in effect you are running the engine in knock at all times ?" > was this Holdens intention as well.....?

Now the Flashscan logs little or no knock at all and produced the dyno graph you saw earlier.

If this is not the correct way to do it or yoou have some other advice to share please let me know as this is the first one I have tuned.

"Thought I was doing ok actually and only trying to pass on information"

Thanks,

Mike

Delco
August 2nd, 2007, 10:48 PM
Ideally the engine should never see knock , knock happens when you are past MBT , any knock can be detrimental to the engine which is one of the reasons GM made the knock sensitivity table so sensitive , Holden then retuned and though incorrectly that they could run more timing than they do.

The factory doesnt always get it right , having worked in a engine development enviroment for companys such as this you soon learn that quite often they dont get it right.

You would be suprised at the amount of power that can be picked up on a dyno by removing timng at critical points in the spark table.

The new knock pids are awesome as you can monitor the knock sensitive cylinder ( yes some are more sensitive than others ) and tune accordingly.

I didnt see that you were using average fuel , but really you should tune to the fuel you use 90% of the time .

The Alchemist
August 2nd, 2007, 11:08 PM
Ideally the engine should never see knock , knock happens when you are past MBT , any knock can be detrimental to the engine which is one of the reasons GM made the knock sensitivity table so sensitive , Holden then retuned and though incorrectly that they could run more timing than they do.

The factory doesnt always get it right , having worked in a engine development enviroment for companys such as this you soon learn that quite often they dont get it right.

You would be suprised at the amount of power that can be picked up on a dyno by removing timng at critical points in the spark table.

The new knock pids are awesome as you can monitor the knock sensitive cylinder ( yes some are more sensitive than others ) and tune accordingly.

I didnt see that you were using average fuel , but really you should tune to the fuel you use 90% of the time .

Yeah our fuel here is crap quite honestly and the dealer told him you can run it on 91 or 96 :muahaha: ....
So would Holden or GM have set the individual Knock sensitivities per cylinder tables on an engine dyno in a controlled environment etc.......at all table points g/s rpm etc and why is Cylinder B set so sensitive compared to the rest. It was triggering knock retard at 13 degrees at 3000rpm at full load during a run. I am now running 17 degrees in the same spot with zero knock and 20% more torque. Surely that 1 cylinder can't be that far out from the others > what do you think?

Mike

Delco
August 3rd, 2007, 12:39 AM
Back passenger side cylinder is very sensitive to knock , more so than the others.

It is also the first cylinder to break normally.

4600 -5200 is where the engine has the most tendancy to knoc as well

The Alchemist
August 3rd, 2007, 07:36 AM
good to know thanks.

limited cv8r
August 3rd, 2007, 05:25 PM
anybody have trouble opening the HSV .tun file? I also asked tordne to increase the upload file sizes for the forum a while back, so you should be able to repost the .tun file on here, .tun file size allowed is now 3.91mb

ntae
August 3rd, 2007, 05:33 PM
anybody have trouble opening the HSV .tun file? I also asked tordne to increase the upload file sizes for the forum a while back, so you should be able to repost the .tun file on here, .tun file size allowed is now 3.91mb

it wont open my end either :bawl:


I set the FTP up so any thing could be wrong , if you want a copy email me admin@ntautoelectrics.com.au

Geoff

The Alchemist
August 29th, 2007, 10:52 AM
Had a comparo with that HSV file vs SS-V. Interesting differences in the knock senor sensitivities. The HSV runs almost a flat "3" right through the important bits vs the SS-V which runs varying figures. I wonder, both are different engines, but the HSV appears to have a set figure of "3" > which is it actually set up or is that figure of "3" just been wacked in there?