high voltage injectors (HDi diesel)

[ecoxantia]

Thanks for the reply dcb, having had a look at the code you have my utmost respect!!

Pressure - red line above- this is the pressure of a common rail - a foot long forged steel pipe in other words - some way from the injector itself. I wouldn't have thought its any kind of pwm, certainly not read about anything like that. Basically there's a 3 piston pump that provides the pressure and the pressure is as high as the pump can manage without wasting too many kw! More recent HDi engines apparently have even higher rail pressures.

So I think the blips are a pressure wave from the injector opening and closing. There's one for the pre-injection and one for the main injection - interesting but a "red herring" for my purposes.

20 volts was measuring away from the duino, in the engine bay!! 20V is the limit of my scope...

Thanks for the atmega on/off voltages - shame that the decay is quite slow (shallow slope) and I wonder if that's why at idle the us displayed varies quite a lot.

I agree that the duino does not seem to be measuring what I want. I'll try some signal conditioning and see what happens. It may only be measuring the pre injection, I don't think the scope always shows the rapid drop in signal.

I'm doing analogread on the red line, the rail pressure, by the way. I'll check that out on the bench with a power supply next.

[dcb]

re: longer injector times at idle. It may be that you have to use the other injector lead too. Or that somehow we are looking at the low side of the pulse. But it could be that it is positively switched also.

You might try changing the code for a positively switched injector and see if that helps:
i.e. change
attachInterrupt(0, processInjOpen, FALLING);
attachInterrupt(1, processInjClosed, RISING);

to
attachInterrupt(0, processInjOpen, RISING);
attachInterrupt(1, processInjClosed, FALLING);

[ecoxantia]

Thanks, now I think I need to get a more comprehensive picture of what's going on here; scoping and recording both lines for example. I think you may be right about the low side of the pulse...

[ecoxantia]

I've swapped RISING and FALLING as you suggested and it is at least giving more "uS" for more accelerator! Promising. No time now to set up the scope.

Won't know more until I road test it tomorrow...

[ecoxantia]

...well sort of.

After realising that my injectors are positively switched, dcb's suggestion of swapping RISING and FALLING in the code, at least gets me sensible data for injection timing.

In addition I have reduced the voltage of the injector signal to 3V peak, with a potential divider.

I still have not managed to implement my common rail pressure adjustment, some work is needed on the bench to find out what's wrong there. At the moment it's all commented out.

So at present I have a device that works well at low speeds. At high engine or road speeds, where the pressure makes a significant difference the amount of fuel injected, things are not so good; I get the same mpg at 60mph or 80mph!

I have found that coasting out of gear is better than coasting in top gear, at least according to hdiguido based on v0.70...

[ecoxantia]

Finally I've got it working! Injector pressure and all! Simple coding mistake was the problem; it wasn't reading the right pin.

Real life got in the way for a few months...

Next job is to get a decent enclosure and integrate it with the dashboard- then on with some aero mods.

[dcb]

sweet

[rmccomiskie]

Excellent work! Very rewarding isn't it? You should add your information to the MPGuino Wiki and link to a page about your project.

[ecoxantia]

I intend to write it up soon - and yes, very rewarding. Already seen another 10 to 20% or so mpg increase in the few days since I've had instrumentation; though the proof will be at the next fillup of course.

Naturally there are a few small issues; I think my pressure logging is still not quite right; a cross-check with a voltmeter will shed some light on that. It's fascinating to see when the engine uses high injection pressures and when it backs off.

I have also noticed that the guino reckons coasting in-gear uses a lot more fuel than out of gear. That could be down to the code struggling to pick up on some very short injection pulses, I don't know.

[rmccomiskie]

Originally Posted by ecoxantia

...I have also noticed that the guino reckons coasting in-gear uses a lot more fuel than out of gear. That could be down to the code struggling to pick up on some very short injection pulses, I don't know.

My MPGuino is installed in a 1997 Audi A6. It has a negative injector signal like you. I reversed the external interrupt edge detection so it would read the actual pulse rather than the non-pulse part of the cycle. It works just fine now.

I noticed that the GPH jumps to a very high value when coasting. Looking at the signal, the injectors are being shut off by the ECU. The signal is a flat line. For some reason, the interrupt routines still run though and they pick up the non-pulse part of the cycle. Hence, the jump to high GPH values. As the car coasts slower and slower, the detected pulsewidth increases. You might experiment with coasting down to a slow speed and see if the GPH increases as you go slower.

My workaround involves modifying the processInjClosed() routine to limit the range of pulsewidths that are considered valid. I've limited the range to: lower end=injector settling time, upper=4000uS. Any other pulsewidth is clamped to zero so it doesn't affect the fuel usage calculations. I use 4000uS as the upper limit because, in normal driving, I've never seen a pulsewidth greater than 3000. When the jump to high GPH occurs, the pulsewidth is at least 11000uS. I added a new setting parameter for the upper pulsewidth limit.

unsigned long  parms[]={69ul,13406ul,119500000ul,6ul,420000000ul,19800ul,72ul,4097ul,0ul,2ul,4000ul};//A6 defaults w/ECU signal
char *  parmLabels[]={"Contrast","VSS Pulses/Mile", "MicroSec/Gallon","Pulses/2 revs","Timout(microSec)","Tank Gal * 1000","Injector DelayuS","Weight (lbs)","Scratchpad(odo?)","VSS Delay ms","Pulsewidth Limit"};

#define pulsewidthLimit 10

void processInjClosed(void){      
  unsigned long t = microSeconds();
  unsigned long s = parms[injectorSettleTimeIdx];
  unsigned long x = elapsedMicroseconds(injHiStart, t);
  
  if (x >= s && x < parms[pulsewidthLimit]) {  //if elapsed time > settling time and < upper threshold
    x -= s;                     //then normal pulse, subtract settling time
  }else{
    x = 0;                      //otherwise pulse width assumed to be invalid
  }
  
  tmpTrip.injHius += x;       
  tmpTrip.injPulses++;      

  if (tmpInstInjStart != nil) {
    tmpInstInjTot += x;     
    tmpInstInjCount++;
  } else {
    tmpInstInjStart = t;
  }
  tmpInstInjEnd = t;
}

When coasting out of gear, the engine is at idle and the ECU will feed pulses to the injectors to keep the engine running. While coasting though, to minimize fuel usage, no pulses are sent.

I have no idea if that's what you are seeing but it's worth knowing about.

Good luck.