Squeezing extra MPG out of P&G (EOC WHILE KEEPING THE CAR ON??)
 

Ever since I took my car into the dealership to service, I disconnected my kill switch, so I have just been doing regular ENGINE ON P&G this whole week. I noticed something interesting.

This might not apply to all cars, but on my 2013 Subaru WRX Sti, I notice then when I shift to neutral to ENGINE ON coasting, my ScangaugeII indicates 9999mpg for a half second when rpms reach approx. 900rpm.
Does this mean my injectors are actually turning off for that half second around 900rpm? I busted out my Torque App and drove to my regular “testing grounds” to verify.

Testing grounds – 1 mile stretch of road with a slight incline, late at night, in a business park so no traffic to disturb

I setup the Torque App to read FUEL USAGE to the 1000th decimal point, because I know fuel usage for 1 mile neutral coast would be minimal, but this is simply for testing purposes.

I drove up to 40mph, shifted to neutral to initiate coasting , set the app to start reading fuel usage until a pre designated ending point. I did bi directional runs

Straight engine on coasting
.018
.015
.016
.015
AVERAGE - .016 fuel used

Engine on coasting, blipping the throttle to allow rpms to sweep 1000rpms down to 900rpms
.014
.014
.014
.014
AVERAGE - .014 fuel used

Straight engine on coasting
.015
.015
AVERAGE - .015 fuel used

Total fuel used with STRAIGHT ENGINE ON COASTING IN NEUTRAL - .0156
Total fuel used with ENGINE ONE COASTING IN NEUTRAL WHILE BLIPPING THE THROTTLE – .014

That’s a potential extra 10.26% savings in fuel WHILE USING A FUEL SAVING TECHNIQUE!

I’m thinking of the potential benefits for ecomodders
1) Achieve “simulated” EOC while keeping the engine on means you keep power steering on, you keep full braking power, you keep your ac running.
2) Less wear on clutch on transmission if you no longer need to bump start

On this morning’s commute to work, there is a long downhill I normally coast in neutral, typically scangauge II reads 150-200mpg during the whole coast down, I blipped the throttle more rapidly to stay in the 900rpm sweet spot and my scangauge II indicated 9999mpg on the whole way down.

I’ll do more testing and post up my Torque App readings later!

It could be. But keep in mind that scangage & torque don't actually measure fuel consumption, they estimate it. They probably use mass air flow & O2 sensor.

That doesn't make sense. I think you've found one of the many places where the real world and the technological world aren't quite in the same plane. I suspect - and I'm going to subscribe to this thread because I really want to know more - that when you fill up it will turn out you've actually used more fuel.

If you haven't, well, I got nothin'. That would be me with a big heapin helpin of egg on my face.

I suspect it has something to do with the refresh rate and how the fuel usage is calculated by the scanners. I don't think I would trust anything other than an MPGuino in this testing scenario.

Was the car fully warmed up for the first A sets? I would suggest doing another B set after the second A to try to rule that out (so ABAB). Your sets are all extremely close, with your second A being right in the middle of the first A and your B. Looks too close to call it, in my mind, especially without being sure the instrumentation is inaccurate in this situation.

I've still got my old mpguino , I'll hook it up and see.

I'll post my findings soon!

my subaru does the same thing. I think it is actually using zero fuel for an instant as your foot comes off the gas it goes into open loop. which means it is disregarding the O2 sensor and fueling according to the fuel maps. at some rpm >~1000 and off the throttle it goes into DFCO.

am looking forward to confirmation or correction on this.

good, so im not crazy then!

I'm trying to find the floppy disk drive cable that came with my old mpguino, then i'll hook it up this week.

I'll also grab my volt meter and hook it up to the injectors as well to see if the signal goes to 0 during that ~1000rpm range.

Actually this might make a lot of sense. Perhaps its going into DFC for a bit. It could be more efficient to pulse the engine, blips, instead of just idling.

EXACTLY ...this is what I am experimenting!

The potential is that i can get the same effect as EOC while keeping the engine on...thus less wear on clutch due to no more need to bump starting, you get to keep power steering on, you get to keep full braking power on, you get to keep air conditioning on, no need to hack wires to install a kill switch.

My girlfriend approves this technique, she says
"as long as you dont have to kill switch the car, its ok with me..."

I think this has been tested (though not necessarily for a subaru with it's added AWD drag), leaving it in gear pulsing then coasting down in gear with dfco. from what I remember the coast downs were short enough that coasting in neutral with the engine idling was more efficient.

The wear from a bump start is pretty negligible given the forces involved. With a bump start you are getting the weight and drag of the moving parts of the engine going. Compared to just getting the car moving you are getting the whole mass of the vehicle moving.

Yes, in most cases, coasting in neutral is more efficient than coasting in gear/dfco because neutral coasts typically last longer.

My observation on my subaru (possibly other subaru's), when i shift to neutral and the rpms drop....around approximately 1000rpms, the fuel injectors cut off for a split second.

My theory is that when i am coasting in neutral, if i keep blipping the throttle around the 1000rpm sweet spot, i am able to coast in neutral while simulating Engine OFF coasting vs. coasting in straight neutral.

If it works out that a series of blips costs less in fuel than just idling, a solenoid acting on the throttle cam could automate it.

I think the ultragauge is reporting more injector time off than we are actually having though. (I don't know for sure though) it seems like the injectors are cutting off during the drop and the UG is showing that cut off to be longer than reality.

How does Subaru handle the transition from DFCO (clutch in/neutral and foot off throttle) to idle? does it go right back into closed loop. or remain on an "initial idle" program for a moment.

This is pretty interesting. I've noticed that in my Yaris it only goes into DFCO (also 9999mpg on my ScanGaugeII) during actual engine braking in gear, and I can feel the engine braking. I've never seen DFCO outside of engine braking. I will say that I have seen DFCO come on for a split second when I shift to Neutral, but I think that has to do with a very brief moment of engine braking before the coast.

Looking forward to your results with these tests.

I accellerate and engine RPMs are up (lets say 2500rpm) and then you put car to neutral, it seems logical that you get fuel cut - car ecu sees that you want RPMs to go down to idle speed - so to drop rpms, it cuts fuel. Why it should send fuel to injectors if RPMs must go down? It decellerates engine using fuel cut. When engine speed goes to idle speed, injectors are turned on again - otherwise it would stop.

I have verified this with two cars watching mpguino (Mitsubishi Eclipse and Audi A4 diesel).

ScanGauge indicates DFCO between most every gear shift on my Corolla. Not sure if it is actually happening though?

I did some data logging last night on my torque app. I'll post up graphs today.

I'm noticing the dfco may actually be more dependent on throttle input instead of rpm (my original theory).

bestclimb appears to be correct in that the car goes into dfco for a split second when I step off the gas pedal.

Therefore, if I keeep blipping the throttle, I can extend the dfco event for longer periods of time. However, the efficacy seems to diminish when I reach speeds lower than 20mph

Data Logging...theory %50 confirmed?
 

Here are the graphs of the data logging I did with Torque App last night. It pretty much 50% confirms my theory is valid ( I still want to hook up my MPGuino and Voltmeter to the injectors to be 100% sure).

It appears that the engine goes into DFCO for a split second when I step off the throttle (red circle). Throttle position goes down…fuel flow drops to 0 and instant mpg jumps up.

Then I keep blipping the throttle to extended the DFCO event for longer periods of time. Fuel flow drops down to 0 and instant mpg jump up to 255mpg (the max Torque App displays)
http://i101.photobucket.com/albums/m...tiongraphs.jpg

Thoughts? Comments? Constructive Criticisms

I also did more ABA testing.

I changed up my testing method to eliminate more variables. This time, I drove up to 50mph, shifted to neutral to coast, then when I hit 40mph, I hit RESET on Torque, then when I hit 20mph, I hit RESET AGAIN. This way, I only measure the amount of fuel used while coasting. I eliminate the variable of my foot accelerating too hard.

I also measured using 6 decimal spaces, to allow more accurate detail in fuel used.

A1 Runs – Straight neutral coasting from 40mph down to 20mph
0.003973
0.003607
0.003754
0.003500
0.003657
0.003907
A AVERAGE – 0.003733 gallons/fuel used

B Runs – Blipping throttle while coasting from 40mph down to 20mph to simulate EOC event
0.0015200
0.0007780
0.0001470
0.0009400
0.0007230
0.0006480
B AVERAGE – 0.0007926 gallons/fuel used

A2 RUNS
0.003673
0.003458
0.003368
0.003382
A2 AVERAGE – 0.00347 gallons/fuel used

A averages – 0.00362 gallons used
B averages – 0.0007926 gallons used

78.11% improvement over traditional P&G

Obviously, using straight EOC will yield better results.

I also data logged Air/fuel ratios too if you guys want to see that graph.

This might be one of those times where having an external fuel cell would be helpful. :D

Interesting. It's an appealing thought... Pulse-and-glide idle during your pulse-and-glide drive. Yo...

Tonight while at work (pizza delivery), I experimented a little with this.

First off, with my automatic, if I coast (engine on neutral) and then shift into gear when coming to a stop, the car will not engine brake and go into DFCO unless I blip the throttle. Also, usually no matter what gear I'm in or speed I'm going, if I let off the throttle and it doesn't immediately go into DFCO, I can blip the throttle and it will then go into DFCO.

Anyway, tonight I decided to blip the throttle during a few coasts, and sure enough, my ScanGaugeII read 9999mpg after most blips. The problem seemed to be figuring out how hard to blip (light tap or a bit harder tap) and how often to blip (every 2 sec. maybe?). Whether or not fuel was cut off during coasting when my ScanGaugeII read 9999mpg, I don't know, but from what you've tested so far, it seems like blipping the throttle while coasting works.

If further testing confirms this, I like bestclimb's idea on automating it, though I don't know much about setting something like that up.

Nice work. :thumbup:

Thanks. Here's a contribution: Downhill (If it's steep enough,) Putting the 2000 Dodge Caravan 3-Litre into low gear shuts off the fuel injectors.

Yes, coasting in gear activates Dfco. If you are going down a steep hill, dfco in gear is fine because you will not gain too much speed and the injectors shut off. However, dfco in gear is not beneficial with the pulse and glide technique on flat roads because you will lose too much speed on the glide, thus you have to pulse again too quickly.

Pulse and glide (engine on neutral coast) is the second best option when compared to engine off neutral coast.

I discovered that in my car, I can greatly improve the efficiency of the traditional pulse and glide engine on neutral coast by blipping the throttle during the glide portion. The fuel injectors appear to shut off for a split second every time I step off the throttle, therefore, by continuously blipping the throttle, I can keep the injectors off for a major portion of the glide, thus simulating engine off coasting, while keeping the car on.

i found my voltage meter. Need to research where to tap into the fuel injector wires so i can monitor injector voltage in my car while driving

That's an interesting idea. However, my car uses .14 gallons an hour while coasting in neutral. Wouldn't constant blipping of the throttle use more than that?

It might. That's the fun part of modding and hypermiling: what works for me in my car on my route might not work for me in your car on my route. But it might work better for me in my car on your route or any other combination, and I haven't even mentioned weather.

When i "blip" my throttle, its more of a gentle continuous tap. it's not the kind of "blip" i would use when rev matching the engine/transmission (similar to your learning of bump starting :))

On my car, the scangauge shows that when i step off the gas and shift to neutral, the fuel injectors cut off for a split second. Then every time i tap my throttle, the scanguauge shows 9999mpg. So the ecu is detecting "step off throttle...cut fuel...step off throttle...cut fuel...step off throttle...cut fuel."

http://i101.photobucket.com/albums/m...tiongraphs.jpg

So there is definitely a correlation between speed/throttle position/fuel injectors.

Here is my ABA TESTING:

I changed up my testing method to eliminate more variables. This time, I drove up to 50mph, shifted to neutral to coast, then when I hit 40mph, I hit RESET on Torque, then when I hit 20mph, I hit RESET AGAIN. This way, I only measure the amount of fuel used while coasting. I eliminate the variable of my foot accelerating too hard.

I also measured using 6 decimal spaces, to allow more accurate detail in fuel used.

A1 Runs – Straight neutral coasting from 40mph down to 20mph
0.003973
0.003607
0.003754
0.003500
0.003657
0.003907
A AVERAGE – 0.003733 gallons/fuel used

B Runs – Blipping throttle while coasting from 40mph down to 20mph to simulate EOC event
0.0015200
0.0007780
0.0001470
0.0009400
0.0007230
0.0006480
B AVERAGE – 0.0007926 gallons/fuel used

A2 RUNS
0.003673
0.003458
0.003368
0.003382
A2 AVERAGE – 0.00347 gallons/fuel used

A averages – 0.00362 gallons used
B averages – 0.0007926 gallons used

78.11% improvement over traditional P&G

However, this phenomenon may only be present in cars with "throttle by wire." I'm not sure if this phenomenon occurs with older cars with "throttle by cable/butterfly."

Since you have an Ultra gauge, and if you are curious, i urge you to see if this technique might work on your Civic.

This is amazing. It's like finding a button-press combo for unlimited ammo in your favorite shooter game.

Should I go find some eggs? Lol

Is your data logger recording what scangauge reports as fuel consumption or actual injector operation?

Any news on getting an mpguino (or other direct injector duty cycle monitoring) into the mix?

As I understand most OBD fuel monitoring is based on airflow. With foot off throttle and open loop once up to temp = fuel off. If there is some parameters at low RPM that result in open loop operation with the foot off the throttle it could be tainting the results. (would be interested to have that misconception cleared up if I am incorrect)


This find seems almost too good to be true. I have noted that my ultragauge reports 999999 for a, um, noticeable, period of time after the engine rpm has stopped dropping and is idling (indicating that unaccounted for fuel may be being burned). I still think it's a possibility that indirect calculation of fuel consumption lends itself to the possibility of this being an artifact.

Your data definitely indicates that further investigation is in order. A way to get comparatively close to EO(ff)C performance out of engine idling coasting while keeping brake booster and power steering systems pumping is pretty awesome.


Could a signal to the solenoid that actuates the idle air control system automate this? I am guessing not as it seems to be linked to the throttle position sensor (which may be spoofable)

Ohh idea for another test. Monitor engine RPM, spoof the throttle position sensor and see if the ecu reports injector cut off.

not trying to criticize as I am really hoping this effect is workable.

I tried this with the MB100 fitted with an MPGuino, and it didn't work.

Today I tried it with UFI and SGII and got simular results to those posted. Unfortunately, the technique only works when rolling. Although UFI has stop-start, the engine off means no A/C. It would have been awesome if I could sit stationary and keep the A/C on with a 70% fuel saving.

Either way, it's a very distracting way to drive, paying attention to SG and the revs at the same time while trying to blip the throttle the correct amount. For DBW throttles it could easily be automated though.

I will point out that UFI doesn't have a hot wire AFM (or a throttle plate) so SG doesn't exactly produce very meaningful results. This really needs to be verified via MPGuino, then automated via Arduino ;)

Some pertinent questions (I think):

Is the Sub direct or port injection?

If it is port, the fuel cut may be to avoid a spike in exhaust HCs as the airflow alters faster than the fuel empties from the inlet manifold.

What is happening on the other side i.e. how much additional fuel - if any - is being added as the throttle is opened? (Analogous to the accelerator pump in a carb.)

Nice work, great looking graph
 

That really shows how the DFCO subroutine is triggered when the throttle position sensor hits the idle switch. The closed-loop idle circuit has huge control authority to regulate air and fuel to keep the motor running, and will use whatever fuel it needs to keep the speed within idle range.

By blipping or lifting the throttle off of the idle switch you have taken it out of closed-loop idle mode into POT mode.

On the old Bosch DME system the controller used countdown timers on interrupts to control the length of DFCO and also used them to create an 'accelerator pump' action to ramp up the fuel on WOT.

You could really see some gains if you could change the DFCO timer to a longer value--has anybody reversed that ECU yet?

No worries on criticisms. I need everybody's input to help validate (or invalidate) my findings.

What do you mean by "monitor engine rpm, spoof throttle position sensor?" Can you be a little more specific? I want to make sure we're on the same page :)

Yes I noticed the same thing, it only works while moving.

I found my old MPGuino to test and bought some new wire splices. However, the Speed sensor for my car is up in the front wheel wells, which means taking off my tire....which means more labor :(. But i'll get around to testing very soon!

It's port injected. Only the new BRZ and 2015 WRX are direct injection

How should I test if any additional fuel is being used when the throttle is opened? (I have the TORQUE APP and MPGuino)

It MIGHT be possible to tune the ecu/throttle position. User Evergreen tuned the throttle on his 2009 Impreza to give him more control over his throttle (auto tranny) use and gained 3mpg city driving

http://ecomodder.com/forum/showthrea...ore-27518.html

Throttle position sends a voltage (typicaly 0-5v) to the ecu. voltage from the TPS monitored during throttle blips. Then matching voltage sent on TPS signal wire to the ECU. See if the ECU responds by closing injectors. If the effect is throttle position related (rather than RPM) the rpm should fluctuate with open/closeloop switch. (this is most likely a silly idea as I was mostly just thinking of ideas without filtering them much)

Re mpguino. If you are using it for injector testing purposes there is no need to get VSS signal.

65% confirmed? Voltage meter
 

Hooked up my voltage meter to an injector. Setting a baseline

Idle: 14.25v

Accelerating: voltage drops (11v-13v)

DFCO in gear: 14.36v (steady)

14.36v is my assumption that injectors are not spitting fuel because we all agree that DFCO-in-gear shuts off the fuel injectors

If I do my neutral-throttle-blip while coasting: voltage fluctuates between 13.5v, 14.36v, and 14.25v. Pretty much all voltage values show (as measured above). Therefore, my assumption is that while im on the throttle (13.5v), I step off throttle to shift to neutral (14.36v injectors shut off), rpms drop to idle (14.25v). Then as I continue to throttle blip, voltages fluctuate between 13.5v and 14.25v…then every once in a while 14.36v will flash up.

I’m sure my voltage reader refresh rate is too slow. Therefore, I slowed down my throttle blip. I drove up to 40mph, shifted into neutral, started coasting, then I rose revs up to 2000rpm and held the throttle so that the volts read a steady 13.25v. Then I released the throttle, meter shows 14.36v, then down to 14.25v (idle).

I did it again…initiated the coast….held the throttle at 2000rpm…meter shows 13.25v steady…then I step off throttle, meter shows 14.36v, then down to 14.25v (idle)

Thoughts? Comments? Suggestions?

I’m getting ready to hook up the mpguino soon. Just gotta find a 12v source….or buy extra wire to reach the car batter 

14.36v flow off
13.50v higher fuel flow
14.25 low fuel flow

My assumption is that the injectors are supplied with ~14.4 or system voltage (could confirm by checking and comparing voltage at injector to battery voltage with engine running (so charged by alternator) and off)

You are seeing an average voltage (PWM injector signal)
so calling for lots of fuel injector is open (therefore drawing more current) so voltage drops.

calling no fuel injectors off so you see supply voltage.

So... from 13.5 to 14.35 then 14.25 you are seeing call for lots of fuel, then no fuel then a little fuel.

At least that's how I read it.

Bad news

Trying to install the mpguino to finish up the experiment.

I am able to tap into the 12v wire of the injector because it leads back into the main harness, however, looking back at the mpguino wiki, it says to tap into the wire that has 0v.

The problem is that means I have to locate the actual fuel injector.since the Subaru is boxer engine, the injector is buried under the intake manifold, which means removing it.

http://i101.photobucket.com/albums/m...i/IMAG2851.jpg

Furthermore, all 4 injector grounds merge together and do converge at the main harness, but when I hook up the mpguino, I do not get any data at all.

Does the mpguino need to tap into a single injector wire?

For now, Im confident in saying that this theory is 65% confirmed (until I can get the mpguino properly installed). I'll continue tracking mpg on a tank by tank basis.

Bestclimb, you mentioned I don't need vss signal to track fuel usage on mpguino. Without vss, what screen should I use on mpguino?