Scatter Plots for Beginners

[rbabos]

I got some data today from a guy that was concerned about fueling issues.  He supplied with a fairly large but mostly tame riding that was in the area he was concerned with.  A VRod is a TPS based VE cal.  We can still tune with what we see in the MAP vs RPM scatter plot. At the same time we can display the MAPxRPM vs Duty Cycle and the RPM vs TPS that controls the fueling. 

http://www.nbs-stl.com/HarleyTuning/VRodTpsBasedFuelingIssues.png

You can see in the center plot that the fueling is fairly well controlled.  I am a little concerned that the light blue areas are lean.  The light blue color is all scaled the same in all three plots.  I would advise the own to increase the fueling by about 10% in those blue areas and retest to see if the bike rides better.

I also advised him to add the O2 voltage to the log so that we can at least determine if the AFR is swinging leaner than 14.7 at this same place in the fueling.

Have fun tuning

Andy

I suck at reading those . What rpm, tps and which cyl is the lean one? Looks like 3k mid 20s kpa?
Ron

[whittlebeast]

I just looked at Ron's tune with o2 included.  It is running fine.  I told him to go ride and have fun.  Wow VRods are way better behaved with the split intake, as I expected.

Andy

[FLTRI]

Wow VRods are way better behaved with the split intake, as I expected.

Andy

Were you surprised or not surprised?
Individual intake runners worked very well with the MM EFI. Too well as a matter of fact. Sounded like a metric bike so they purposely went back to shared intake so the engine got its potato, potato, potato back...at least a little.
Bob

[whittlebeast]

It is just nice to see a guy nail a tune and a motor that is well behaved.  I  assume he did the tune at home, I didn't ask.

Andy

[whittlebeast]

Here are three screen shots of what I am looking at....

http://www.nbs-stl.com/HarleyTuning/VRod%20Scatter%20Plot.png

http://www.nbs-stl.com/HarleyTuning/VRod%20Histogram.png

http://www.nbs-stl.com/HarleyTuning/VRod%20Log%20View.png

It is interesting to watch the injectors as the closed loop jumps in and out on the last plot.  Look at the Closed Loop Integrator trace. 

In the first shot, the color is O2 voltage scaled from 0 to .84 volts

Have fun tuning

Andy

[whittlebeast]

Here is a filter that you can all use to filter out transients on a Harley.

If you create a filter named Transients HD

and then define it as

[Acel Enr] >0 or [Decl Enl]>0 or [Acel Enr-1]>0 or [Decl Enl-1]>0 or [Acel Enr+1]>0 or [Decl Enl+1]>0 or (abs([Field.RPM] - [RPM-4]) > ([Field.RPM] * 0.10) ) or (abs([Field.TP] - [TP-4]) > 5 )  or (abs([Field.MAP] - [MAP-4]) > ([Field.MAP] * 0.15))

Andy

[ToBeFrank]

(abs([Field.RPM] - [RPM-4]) > ([Field.RPM] * 0.10) ) or (abs([Field.TP] - [TP-4]) > 5 )  or (abs([Field.MAP] - [MAP-4]) > ([Field.MAP] * 0.15))

You should be using the derivatives for these. As written they are dependent on the input data.

[whittlebeast]

(abs([Field.RPM] - [RPM-4]) > ([Field.RPM] * 0.10) ) or (abs([Field.TP] - [TP-4]) > 5 )  or (abs([Field.MAP] - [MAP-4]) > ([Field.MAP] * 0.15))

You should be using the derivatives for these. As written they are dependent on the input data.

Frank, could you give an example of how you would write a formula for MAP change?  As it turns out, I am hanging with the MLV programmers all weekend at a Megasquirt meet.

Andy

[whittlebeast]

Here is what Frank is talking about....

If you write the formula  ([MAP]-[MAP-2])/([Time]-[Time-2]) you wind up with the rate of change of MAP per Sec.  In the Megasquirt world, that is known as MAPdot

Andy

[whittlebeast]

I have figured out a way to plot where the HD ECU is trying to get locked into closed loop compared to if it actually went active in closed loop.

In theory, all three screen should look the same.  In this shot, they look really good.

http://www.nbs-stl.com/HarleyTuning/Harley%20CL%20Verification%20Normal.png

Now here is a screen shot of the same motor but running a tune that is so far off on the VEs that the ECU is has having trouble even getting a lock.

http://www.nbs-stl.com/HarleyTuning/Harley%20CL%20Verification%20Trouble.png

I will post the details on how you can do this.

Andy

[hrdtail78]

I have figured out a way to plot where the HD ECU is trying to get locked into closed loop compared to if it actually went active in closed loop.

What do you mean locked compared to active?

[whittlebeast]

I have had several people mention that the Harley flipped upside down compared to my scattor plots.  I just noticed that it is possible to flip the map by tricking the software.  Check out how I set up the limits by flipping the Min and Max settings.

http://www.nbs-stl.com/HarleyTuning/Flipping%20Scatter%20Plots.png

I will run this past the MLVHD programmers and see if this breaks anything.

Have fun tuning.

Andy

[whittlebeast]

Here is how i generated the plots in post #284

I first created a function

ClosedLoopHD

I defined it as

[Set Lambda]>.973?1:0

And set the scale Min Max to

Min = 0   Max = 1.5
______________________

I then created new field called

ClosedLoopActiveFront

I defined it as

[Front CLI]=100?0:.5

And set the scale Min Max to

Min = 0.1  Max = 0.7
______________________

I then created new field called

ClosedLoopActiveRear

I defined it as

[Rear CLI]=100?0:.5

And set the scale Min Max to

Min = 0.1  Max = 0.7
______________________________

You get the the plots I posted.  Yellow is where the motor is in Closed Loop and Blue is where the motor is in Open Loop.  A mix of dots is where the motor is having trouble staying in closed loop.

Have fun tuning

Andy

[whittlebeast]

Can any SEPST dongle log any bike?

Andy

[joe_lyons]

With the proper cable yes, except Magneti Marelli bikes I think

[whittlebeast]

How long will these things log before they get full?

[joe_lyons]

Alot if your streaming into the computer

[whittlebeast]

Apparently the red button holds 15 min according to a youtube video.

[FLTRI]

Apparently the red button holds 15 min according to a youtube video.

It's designed as a diagnostic tool so folks can hopefully capture running issues the shop couldn't repeat.
Bob

[whittlebeast]

I hope to help owners by helping them to verify that their dealer really did tune their bike.

Andy

[whittlebeast]

This is the list of recorded fields from one of the logs I have.

spk_nock_f
spk_nock_r
twist_pct
ve_front
ve_rear
acel_enr
dcel_enl
new_ve_front
new_ve_rear
adv_front
adv_rear
battery_v
eng_speed
eng_temp
front_cli
front_o2
iac_pos
iat
idle_set_spd
inj_pw_front
inj_pw_rear
map
rear_cli
rear_o2
tps_pct
tps_v
veh_speed
af_lambda
warm_fuel_lambda

We should have plenty of data there to help people out.

Andy

[joe_lyons]

There should also be coolant temp and the adaptive fuels right?

[whittlebeast]

Joe, eng_temp is there.

We can calculate AFF based on

new_ve_front = ve_front * front_cli * AFF Front

So in MegaLogViewer HD, you use the formula......

AFF Front = [new_ve_front] / ([ve_front] * [front_cli])

[whittlebeast]

We can do the same sort of thing to generate Duty cycle Front from inj_pw_front and eng_speed

[joe_lyons]

There is a separate line for coolant temp for the twin cooled models.