Stock 2019 Mustang GT, file and log.

[Grim5.0]

bullitt cams.png

[murfie]

Heres my log on 87 octane after using 1/4 tank.

87 oct.hpl

I did a few low load partial throttle test at different RPMs so you can see where it is detonation limited. I use a similar strategy to test unknown or questionable fuels. start at 3500 try to hold .4-.5 load let it add up MBT or borderline. Then repeat at the same RPM trying to hold .6-.7 load. Then drop a 1000 RPM and repeat until I find where the borderline truly is. Cylinder #3 seems to be the most common trigger.
Ford has used the same strategy with these gen 3 coyotes as they have been since adding knock sensors to their cars. knock advance until just after peak load(peak torque), then run MBT or actual borderline of knock out to the higher RPMs as the airmass in the cylinder falls, knock/ more intense knock is not likely going to occur. Theres a flat spot between 3000-4500 in both MBT and borderline where they let the knock advance determine where the timing is going to be at. I think if I started this pull at 4500+, it would have had better results higher in the RPMs. I agree with the bump at 5300 and dip in MBT at 5750, but don't really think it should fall as the load does with rising RPMs. Then these values, in the mid range, are nearly the same as gen 2 coyotes with 11:1 CR, I think they are on the high side. Need a dyno and high octane to confirm. With everything being so close, I'm just going to trust Ford with this.

COT kicked in after 6500. Ill run this tank through and move back to premium, but certainly don't be afraid to run regular, just keep your foot out of it in the low RPMs(<2K). The choppy spark line is cylinder trim on firing cylinder 4(#8). I'm not really sure why they have per cylinder knock and still use cylinder trim. But I think cylinder 3 would need some trim as well(# \4). At least on this car. The two back cylinders.

MBT-BDL diff.PNG

Curiosity got the better of me and I decided to go back out and test the high RPM start to trick it into more timing. Nope you cant trick it, you get worst results. Its best to let the knock advance find the borderline.

87oct. 4500 start.hpl

seems like there's a RPM switch at 5000 that below knock is per cyl., above its global.

I think my first log the station gave me ethanol free 89(popular with fisher men and boat motors around here), not 92 E10 like I wanted. Fuel trims on 87 are better, knock advance isn't. If the gas pumps around you are single hose, be kind and let your friends fill up with premium first. If I get true 92, i'm sure it will get to MBT. I think cars with calibrations with adaptive knock turned on don't run at MBT at any point, like eco boost.

[Bill88stang]

Incase anyone is curious the Mach 1 tune is fairly identical to the bullet tune including the op cam angles.

[engineermike]

Fascinating stuff!

I plotted the 92 log and the 87 log on the same graph, along with Borderline and MBT. 92 definitely allowed more spark but only individual cylinders hit MBT between 4 and 5000 rpm, then got pretty close again at ~7100+. There is a lot of scatter between 4 and 5000 due to individual cylinders knocking and getting retard. Once past 5500 global takes over and it starts ramping up again with knock advance but appears to knock again at 6900 rpm. I wonder why your gen3 is getting knock at lower timing values than your gen2 did.

SnipImage.JPG

Also interesting that the 87 timing curve matched Borderline very close at 5500 and up, but not at lower rpm. It's almost as if it didn't even try adding past 5500.

Another weird thing I noticed when plotting was that the 60-120 time on 87 was 10.5 seconds vs 10.9 on 92. Is the VCMScanner offset not an accurate timer or was something else going on?

Regarding "seems like there's a RPM switch at 5000 that below knock is per cyl., above its global"...I've wondered about this myself. Look under Spark->Knock Sensors->General->SDLF Average Mode (ECM 45323). This appears to toggle from 0 (Middle Values) to 4 (Whole Engine Value) at 5500 rpm. This can't be coincidence. However, SDLF apparently means "Spark Delta Limiting Function". 1 is Minimum Values, 2 is Maximum Values, and 3 is Specific Cylinder Tracking. What I can't figure out is if this is changing the way the Scanner displays the knock retard data or if it changes the engine from per cyl to global. What I can verify is that both the GT500 and Roush SC calibrations both have these set to 0 all the way through the rpm range.

[Ranger Coyote]

I love this thread... All the best guys on HPTuners picking apart a "perfect" ford tune... Who knew it would be so fun to look over logs from an unmodified car?

[engineermike]

...I'm not really sure why they have per cylinder knock and still use cylinder trim...

Ford seems to do their best to predict where knock will occur and start there before knock advance happens. The cylinder trim is just another example like the other borderline corrections, tip-in knock retard, and so on.

[engineermike]

I still think OEMs design all their engines to see MBT on the lowest octane available, it is the only way they can offer a warranty...

.... I think cars with calibrations with adaptive knock turned on don't run at MBT at any point, like eco boost.

I don't understand how both of these can be true.

[Ranger Coyote]

When I bought my 727HP Roush kit they voided parts of my cars warranty but the phase 1 kit only difference was no boost a pump and a bigger pulley and you could keep the warranty. If they let you keep the engine warranty with a 670hp Roush slapped on it, I don't think Ford gimped the MBT tables for warranty reasons...

[engineermike]

Ford may perform the warranty work on the roush cars but roush foots the bill for it.

[murfie]

I don't understand how both of these can be true.

Me learning and changing my hypothesis. Hence "I think...."

[engineermike]

Anyone have an idea of what that SDLF average mode is?

[Ranger Coyote]

Spark Delta Limiting Function

[ECM] 45323 - Knock Sensor SDLF Average Mode: Determines the mode for spark delta limiting function reference average. 0=Middle Values, 1=Minimum Values, 2=Maximum Values, 3=Specific Cylinder Tracking, 4=Whole Engine Value, 5=Variable Displacement Engine.

[engineermike]

Ranger coyote, I got that much from Hptuners and posted it in #24 above. I was hoping for a better explanation.

[murfie]

Look up patent US6247448B1
The 4th characteristic of an engine operating at borderline.
"If the absolute spark advance is the same for all cylinders, usually only one or two cylinders are contributing to knock, the rest are below their knock limit, hence below their torque/ efficency"

There is another patent on knock sensor reference levels I cant find at the moment, i'm not sure if its Fords.

I agree that's probably the switch to global I was seeing.
I think it has to do with how it is determining the intensity of the detonation that is the "safe borderline", what the reference "no knock" noise level per cylinder should be when comparing each of the 4 different sensors.

0- use an average over time
1- use the lowest reference over time
2- use the highest reference over time
3- use the reference for each cylinder, determined per each cylinder individually
4- use a global reference(one cylinder for all)
5- use the reference from only the active cylinders

I wouldn't want the difference between cylinders to get to great, to me that would either cause a problem or would indicate a problem.

[engineermike]

Thanks, that’s helpful.

So the gt500 and roush sc calibrations both leave it at 0 all the way up. Seems like slightly riskier but more powerful approach than switching to 4 global. 3 seems like a smart way of doing it, but perhaps computationally intensive. Would switching to 1 at high rpm keep it in individual cylinder mode but apply a low noise threshold such that it applies knock retard more readily?

[murfie]

It would switch it to the knock sensor that had the lowest noise threshold in the last x amount of time. The most sensitive to false knock.

[engineermike]

So which SDLF mode do you think would fall somewhere between the stock setting of 4 and the roush setting of 0 at 5000+ rpm in power and safety?

[murfie]

Im not sure.
I would want all the knock sensors to have some input into determining borderline, so I like 0's. The switch to a global eases my mind.

I don't really know how to quantify the benefits of cylinder trim and per cylinder knock advance. Seems excessivly complicated and like splitting hairs for a little benefit . To me spark should be balanced and even and each cylinder should have a very similar borderline, because they should have similar airmass, fuel mass, compression, heat, ect. If a cylinder is a few degrees different than the rest that's an unbalance and a different advance isn't going to fix what's causing the imbalance.

[murfie]

Another log on 87. This time I wanted to pull over let the car cool down for a while like you would at a dyno or drag strip. not a ton but 10-15 minutes. While doing so I got bored and also decided to remove the stock air filter.
MAF signal gets really bad with out the filter in place. Lid was latched back down to the bottom.

87 cooldown no filter.hpl

[murfie]

Another weird thing I noticed when plotting was that the 60-120 time on 87 was 10.5 seconds vs 10.9 on 92. Is the VCMScanner offset not an accurate timer or was something else going on?

Something definitely happened on that 92 log. reset KAM, reset trans adaptive, drove a bit more, a few 2-5th shift logs, and added 87,then drove a bit more is all that happened between them.

Accel G's.jpg

Shows in the RPM slope as well, so what the wheels did the engine did, so not slip. I thought both pulls were on relatively flat straights, but maybe not, I don't really remember the exact spot I did the first log. I have like 25 miles of straight, empty, country road between me and the closest gas stations.
Cars faster on 87 /s LOL.