Stock 2019 Mustang GT, file and log.

[engineermike]

The 92 definitely allowed more spark, but the accel curves for 87 definitely appear stronger. We know the energy content is the same and most of the “87 burns faster” stories are myths. However, I wonder....we know the refineries have had a huge surplus of diesel and jet fuel due to covid, and those fuels are a required byproduct of gasoline production which is still in decent demand. Could it be that they are blending some of these heavier, high energy fuels into 87 and the result yielding more power even at lower spark timing? The octane of those fuels is lower but refineries are constantly blending different components to achieve the desired end octane.

[murfie]

OK ran a loop to get it down to 1 mile left as I pulled into the gas station. filled up with fresh 92. Drove 40 miles then took this heat soaked log.

new 92.hpl

After seeing the results, I got dinner and parked the car with the hood open. after about 30 mins I finished eating, and took this log.

92 with cooldown.hpl

Seems to me these gen 3's are just barely detonation limited on regular when they are cool. On premium they are basically at MBT heat soaked or not. I can add octane booster if any one is curious to see if it can get any closer 5k-6k. And to see if octane booster(boostane) actually works.
Still not sure these MBT values are not slightly high(they are most likely perfect). Almost identical to what I remember the gen2 did except above 5750. I also noticed no COT with out the detonation limit. That's going to be a critical point of tuning here in the near future. No more easy way around it.

[engineermike]

I plotted individual cylinder knock response and my opinion about what's going on is....

Region:
1. Spark started at borderline. Fast mode knock advance is advancing timing quickly towards MBT.
2. Various cylinder start knocking at 4600 rpm, perhaps not coincidentally just as actual timing reaches MBT.
3. Switches from per cyl to global.
4. Knock causes further retard. I say this because knock advance goes from ~7 to ~2 in 300 rpm and this is where we see timing fall below MBT.
5. Normal mode knock advance begins advancing timing.
6. Timing reaches MBT, which is falling, at 7100 rpm. So, knock advance starts ramping down.

CYLINDER KNOCK.JPG

[engineermike]

Something weird to me in the Gen3 calibration is the Borderline numbers at high load. In several of the mapped points like 21 and 25, the borderline timing makes a drastic jump upward at 1.1 load vs .95. This is not a typical characteristic. The Gen2 numbers seem a little more normal I think. On one hand it makes me wonder if running WOT cam timing that allows significantly higher spark timing at high load, like MP21, would yield more power than sacrificing 3-4 deg of spark for a little more airflow. Most testing individuals do involve locking spark timing so you don't see the full effect of the cam timing change. On the other hand, I tend to think that if there were more power available by doing this then Ford would have done it. I would imagine ford has run all the permeations and commutations on the dyno...

Something else weird is that the both of them add a whopping 4.6 deg of borderline at .7 lambda.

[murfie]

The cylinder trim of 1-2* on one cylinder makes the spark look worst than it really is. That would be the first change I make, just to clean up the log. Do you think boostane would change anything? the knock detected pid is stuck at no, and was saying yes on 87. MBT and borderline follow a similar trend above 6k, but I think spark follow the changes of MBT more. BDL is much flatter.

MBT-Spark-Borderline.PNG

[engineermike]

I think with a good octane booster you’re going to see the typical timing sample fall right on the mbt curve once knock advance ramps in. I also think the knock detected pid is only checking snapshots at the sampling time, so it can easily miss knock events of they are infrequent enough. It’s pretty clear to me you are getting knock retard on 92 from 4600 to about 7000 rpm.

[murfie]

So I took another two logs with a couple changes. Interesting findings.

First log is with 0 cylinder trim.

Marks stock 2019 GT 0 cyl trim.hpt

4th gear 0trim.hpl

Second log I did two pulls with SDFL all 0.

Marks stock 2019 GT 0 SDLF.hpt

SDFL 0.hpl

I switch to4th gear, i'm more comfortable with it.

I also found to flash the car I have to press the brake pedal after its in assessor power mode with the button or it will not connect to HPTs server. x7F service not supported in active session message if I don't.

[murfie]

Then heres trim stock and SDLF going to 0.

Marks 2019 GT only SDLF 0.hpt

Trim with 0 SDLF.hpl

I think no cylinder trim, with SDLF all 4's is the way I am going to set it up. Not the most power, not the most efficient. The safest as it seems to be working of the weakest cylinder and the nicest to look at in the logs since all cylinders are using the same advance as the weakest cylinder.

[engineermike]

What’s the difference between all 4’s vs just going to global?

I’ve been tempted to go all 0’s since that’s what the roush phase 2 did and that is the warranty kit.

[murfie]

Global allows 3 knock events before switching advance rates. usually this puts advance above the weakest cylinders borderline, even if it was the one that knocked the first 3 events.

Per cylinder is 1, so the first event the cylinder reports knock it slows advance, the rest continue at the fast rate. Sort of a race to the highest advance, the slowest cylinder @5080rpm wins. This finds and puts you at the borderline of the weakest cylinder. I might leave it 0's below 5080.

I see why they trimmed cyl4, but I don't like the way 0 SDLF and any cylinder trim makes spark advance in the logs look, and I want the cylinders to use the same advance. Im ok with what ever I'm missing from the rest of the cylinders being held back.
I would say HPT should add per cylinder final spark advance, but thats 7 more channels needed in the scanner. There is already a lot. Again I don't know how to quantify how much you gain from per cylinder spark advance. It could be minimal and all for emissions or catalytic temperature.

[Justinjor]

Great thread. Subbing

[dsblk93gt]

Very good and interesting read!

[murfie]

I want to start by saying this Boostane has been sitting in my living room for about a year and a half.

I estimated I had 10-11 gallons in the tank. Adding 4 Oz should have brought the octane rating up to 97-98, I figured I saw 87 and 92, thats a 5 point difference, lets go with another 5.

After adding the booster, I cycled the key on and off a number of times just tring to get the pump to mix it up. I then drove 20 miles to make sure I used all the fuel in the rails and lines.

Same file as in my other post, but for reference of what was flashed to the car.

Marks stock 2019 GT 0 cyl trim.hpt

This log is at the end of that 20 mile drive.

4oz boostane and 20 miles, 0 trim.hpl

Boostane heat.PNG

This was after parking for 30 minutes letting it cool down.

boostane cooldown.hpl

Boostane cool.PNG

Improved, but not as much as I was thinking it was going to.

I might try reducing the slope in the borderline values between 4500 and 7000, It might be too aggressive. Im thinking 20* at the transition between MP 25 and 21, rather than 18, just to flatten between 5K-6K. This wouldn't allow the use of 87 octane. But I could slow the advance rate down a little and maybe help avoid the knock near advanced peak.

[murfie]

Before and after an increase to the Borderline and a rate adjustment. I can get more aggressive with this, this was just starting at ~20*. This is "2 degrees" more timing to me. Low RPM high load. That is where knock is most likely to happen(slow piston speed pressure stays higher longer) and where you should be trying to get timing.

Before after.jpg


Where it is at

Current bdl.PNG

What I am thinking

Proposed bdl.PNG

3000-4000 RPM can take 2 seconds in 4th, longer in higher gears shorter in lower. Seconds per degree would come out to 2/4.5= .44. 4000-5000 the bdl will have a greater slope than MBT, so the advance can be slower. .5-.6. Then above 5000 the bdl will help allow the advance rate to still work by pulling timing out as MBT is dropping.

[murfie]

If anyone is like me and likes to use the gear lever to put it in S and have the car downshift to engine brake and slow you down. Here is my downshift schedule. A log as well to see what it does. If you are just coasting it will keep you right around 3500-4000, If you get on the brake it will get more aggressive.

Marks 2019 GT Sportshift.hpt

sport mode down shift.hpl

[murfie]

The latest in borderline changes. The way the RPM columns are setup different in the high RPMs between Bdl and MBT tables makes it difficult to stay below, but not too far below. Then there's the BDL corrections that when its favorable, it just puts it over MBT when you have them close.


50% added to BDL.jpg

Marks 2019 GT Sportshift 50% bdl to MBT.hpt

50% closer to MBT.hpl


If anyone wanted any easy way to make adjustments to spark, this is what I do. I compare this 50% tune to a stock one. then go to the compare view of each borderline table and multiply it by 0-1 to reduce, and 1-2 to increase the spark. You adjust it to the fuel you are running.

Easy Bdl adjustments.PNG

I had MP 14 wrong, and some sheet cells pointing at the wrong reference. I updated the tune file to have the actual values 50% between borderline and MBT.

[engineermike]

Is cylinder 4 trim still zero'd out? It looks to be the first one to knock in both pulls, which means Ford had it right.

Observation: As long as it's using just MP25, there is no knock retard. As soon as it breaks away from MP25 at 4400 rpm and starts blending 13 mapped points is when it starts to knock and pull timing. Then, at 5200 rpm, it locks onto MP14 and 21 because there is a vertical snap line there. The instant it locks to that snap line, the knock retard stops again. MP14 Borderline is the lowest in the area, so it pulls the average down more than before or after it locked. It stays locked onto that snap line until 5500 rpm. I wonder if the Ford engineers intended for it to lock onto that snap line at WOT, after which it starts blending again.

Something I would really like to see is if the MP14-MP21 snap line weren't there, and another snap line was set up from MP25 (-20,20) to MP21 (0,20) instead, then command OP exhaust cam timing to be 20 from 2000 to 6500 rpm. This would lock it onto MP25 and MP21 up to about 5400 rpm. In looking at the Borderline values in that range, this would result in an increase in Borderline timing because it would no longer take MP22 (-20,0), MP14 (0,0), MP20 (10,30), et.al. into account, which all have lower Borderline values. Either a) it will knock just like it does now, or b) you could gain a degree of timing without knock due to the dynamics involved. If it gains timing, then the next question is; was the increased knock resistance worth any losses that the cam timing might have incurred, if any? Many tests seem to indicate a few degrees either way on the exhaust cam results in basically no power difference if the timing is locked, but what if a cam timing tweak gets you closer to MBT?

[murfie]

No that's with trim on. That cylinder is a PIA not trimming it makes for a nice spark advance PID, but very inconsistent as some times it's happy some times its not and they rest of the cylinders get pulled down. Trimmed it's much more consistent. The spark advance just jumps up and down 2* trying to represent it being different from the others.

[engineermike]

I thought about my post above and realized an error in my thinking. The goal should not be to push the borderline timing as high as you can, rather to get the difference between MBT and Borderline to be as small as possible. In other words, manipulating things such that borderline is 30 deg if MBT is 40 might not work as well as a borderline of 20 and MBT of 25.

So, I plotted the delta (MBT - Borderline) for each Mapped Point and .95 load. The black square is roughly the operating point at that time.

5000 rpm:

5k timing delta.JPG

7250 rpm:

7250 delta.JPG

As you can see, Borderline is about 10 deg below MBT at 5000 rpm, but at 7250, MBT is barely over Borderline at all. Throw in a small lambda correction on both and, voila, you're able to run MBT. Only at 7000+ rpm, though, just as the logs show. The problem is, below 5000 rpm the coyote clearly performs significantly better with the intake cam fully advanced and I don't see a path to get from -20 to +25 without passing through regions where Borderline<<MBT.

Maybe....

Optimum path.JPG

Kidding of course...seriously though.

This 5 or 6 dimensional thinking is going to drive me nuts...

That said, I don't think the tables were fully accurately populated by Ford. In some, there are clearly some breakpoints where they just stopped calibrating them. Unfortunately, Mapped Point 14 (0,0) MBT is that way and it's right in the range where we use it.

[murfie]

MBT minimum timing for best torque is really all that matters. You want it as low as possible.

Borderline is just an unfortunate limitation you have to deal with.