Graph of injector pulse width when revved.

[CMF_queazocotal]

Revved in neutral.
The sawtoorh is I think the ECU fueling only 3/4 cylinders every second rev - when it's at 6000RPM, to avoid the engine exploding.

Now, all I need is to hook it to GPS, calibrate it, and I've got a real-time fuel consumption meter.

 

[CMF_queazocotal]

Oops - ~1998 - CG10de

 

[CMF_kneival]

How did you get this data?

Consult and datascan or something?

 

[CMF_queazocotal]

Not quite.

Undo the trim panels, and solder a connection to the blue wire from the ECU, which as the haynes manual promised, is an injector wire.

Now, 2 47K resistors, and 1 1K resistor, hooked up to laptop sound port, and postprocessing the wav file.

GPS to calibrate the injector time/quantity curve, and this should give me real-time dyno, RPM, and fuel use.

Of course, you know this, but:
Shorting the injector wire to either 12V or 0V will permenantly screw the injector and ECU, and if the engine is running, the cat too.

 

[CMF_Ed]

"Of course, you know this, but:
Shorting the injector wire to either 12V or 0V will permenantly screw the injector and ECU, and if the engine is running, the cat too."

Shorting the blue wire to earth would cause the injector to stay on, whilst definately not a good thing, it would not screw the injector or ecu, infact the ECU would not even notice, it would be engine damage (i.e. cyl full of fuel) if you were unlucky to leave it for any length of time. Shorting it to 12v however would either fry the injetor driver fet or blow the ECU fuse or both, and in this case the Injector would not come on at all. (both sides of the injector being held at 12v)

Quite a neat idea you have there though, but I dont understand the sawtooth on the red rpm scale. The ECUs doesnt do anything to stop the engine exploding as you put it, unless it hits the rev limiter. Do you have the raw data .wav file of this so others can have a look?

 

[CMF_queazocotal]

I diddn't keep the .wav.
It is on the rev limiter - I think that it's cutting fuel not
only at 6K RPM, but also if it's approaching that very fast.

The .wav showed periods of complete inactivity, so it's not a bug in the processing program.

I'm going to take more samples with GPS as well, and maybe a webcam tomorrow, to get 'real' data over a journey.

 

[CMF_squigbobble]

What's the bottom scale on the graph? I'm totally confused by it :S

 

[CMF_queazocotal]

Left hand axis is RPM.
Bottom is seconds.

Red is RPM (really time between injector firings, which isn't always the same thing), and green is the width of the last injector pulse.

This one is another - similar, but the axis on the right is the injector pulse width in seconds, and a blue graph which is computed fuel use,
(injector pulse width * frequency - this is inaccurate, till I calibrate it) again along the bottom is seconds.

Stock CG10DE in neutral.

The sawtooth is noticable audibly when it hits the rev limiter, the real RPM climbed smoothly, sat on the rev limiter for a moment, then fell as I let go of the accellerator.

I need to do some road testing tomorrow - see if it ever stops using the injectors under engine braking/...

Then, it's to working out a nice meter or something for it.
Perhaps something as simple as an LED indicating that goes green or red if it'd be more efficient in the next/last gear at the current engine load.

 

[CMF_queazocotal]

Hokay, got logs.

http://www.mauve.plus.com/scatter.gif (RPM on bottom, injector width on left) is a scattergraph of every injection over about 45 mins of fairly sedate driving.

The line on the bottom is I think engine braking, I assume that no actual fuel is getting injected at this time.

I assume that the minimum active fuel injection line along the bottom curves up at low RPM, as at very low RPM, the throttle has a limited effect, pressure goes up, and therefore to keep it stochiometric, fuel injected goes up.

I've now got to take the GPS logs, and merge them, to get per-gear info, and then add terrain variations, to get me a dataset for engine load vs injector width.

 

[CMF_kneival]

Interesting stuff mate.

How do you get the graph program to work? I have a lot of logs from my consult port if you are interested?

 

[CMF_queazocotal]

Lots of linux tools mainly.

sox -r 44100 -c 1 -b -t ossdsp /dev/dsp0 -t raw -|\
hexdump -e '1/1 "%1d\n"'|\
awk '!o&&($1>160){o=1;last=start;start=(NR/44100)}o&&($1<100){t=NR/44100;o=0;print t,t-last,t-start}'

Takes the output from the soundcard, and processes it into lines of the form

167.729 0.113469 0.00328798
167.839 0.113628 0.00328798
167.949 0.113333 0.00324263
168.059 0.112993 0.00324263
168.169 0.113424 0.00324263
(seconds, time since last injection, injector width in seconds.)

Then I simply use gnuplot to plot it, with a command like
plot "car.dat" using (120/$2):3 with dots

Which just plots column 2 (well, 120/column 2, which is RPM usually) against column 3.

I have no idea of the format of the files you're talking about.
If you'd like to mail them to [email protected] , I'll have a look at them, and see if they are understandable.

Though probably not of much relevance to a stock CG10DE

Eventual planned upgrades are a CG13DE engine, possibly with a BMW mini S supercharger, not to mention MP3 stuff.

 

[CMF_kneival]

Ok will email them :]

I think they are from when I had a stock 1.0, cant remember 100% tho :]