Gallons Per Hour Error?

[usergone]

I tried hooking the leads from the shunt to the comparator inputs of the op-amp. Then I supplied power to the op-amp, and took the output and put it towards my oscilloscope.

It seems to have failed. When I ran the engine, I did not get a nice 0-12V signal whenever the injector was injecting. I got... well. Not much. At all.

[Superturnier]

Are you actually measuring the voltage across the shunt?
In other words, the A-wire is in front of the shunt, but where is the ground wire connected? Is it after the shunt or in car ground?
And how about the grounding of the variable resistors?

I think that they all should have the same ground so that the comparator can see the real voltage difference.

Good luck!

[dcb]

the isolation is an issue, and why I like the optoisolator approach, anybody have the patience to try one? I've got the opto somewhere.

http://ecomodder.com/forum/showthrea...tml#post106855

[usergone]

My first long post was mostly speculation, sorry if that confused people. When I was messing around with the shunt yesterday, I was measuring the voltage across the shunt, yes.

When I tried using the rectifier and grounding the negative side, I could no longer get the nice waveform of the current.

So, as dcb said, isolation is the issue. Neither side can be grounded, but the sides have to be compared to one another.


I just went and looked at the post that dcb linked to, and it gives me an idea. It appears as though that schematic you drew uses the ECU side of the injector as a switched ground. It seems as though that would make sense, because at least on my truck, the injector is put onto 12v and the other side goes to the ECU, which grounds the injector to activate it.

So, it may be possible to just put a very low current draw device on the lead to the ECU, so that when the ECU grounds the injector, it activates said low current draw device. I suppose a diode would be in good order so that the 12v (when the injector is not grounded) doesn't mess anything up.

[usergone]

I went and made a little transistor switch to indicate when the injector is grounded.
This is the schematic:


When I started the engine, I looked at the oscilloscope and saw this:



So it works! The dots you see correspond to the "B" part of the current graph:


I turned on some loads for the engine, and, not suprisingly, the dots extended to about double their length. So the dots are an indicator of the pulse width of the injector.

I need to see now if the guino can register that as the injector length, or if it will only look at the first part, and ignore the dots. I think some tweaking of the "injector delay uS" is in order.

[usergone]

I just tested this with the guino. At first, it was reading GPH according to engine speed (like before). I went and messed with the Inj delay us, and with it set to 0, the guino would register more GPH if I had the A/C on than if I had no load on the engine.

However, even with the injector delay set to 0, the reading would still change according to engine speed (a lot). This leads me to believe that the time of the "peak" is constant, regardless of RPM. I think I need something like a capacitor that would "hold" the voltage up. The "peak" would charge it, and the pulses from the "hold" would hold it up, but when the injector stops, a resistor drains the capacitor. Like this:


I just don't know what size capacitor I need.

[usergone]

Well, boys (and girls), I think I've got it.

I put a 4.7 uF capacitor in, like the schematic above, and after tweaking the value of the resistor in parallel with the cap, I got a funky looking signal like this:



More importantly, I was able to use the oscilloscope to measure the time of the "peak" part. 2 ms, or 2000 us. Once I set the injector delay us to that, I got (what I believe to be) a fairly realistic reading. It didn't change too much with RPM, but when it did, I think it was an actual fuel consumption increase.

[alszik]

Dear thecheese429,

could you please share some details about your schematic? I'd really like to try this in my Peugeot.

Thanks,
Tamas

[usergone]

I actually had doubts about the accuracy of this method, so I put this on the shelf for the summer (actually, in the glove box).

Basically, the injector has 2 leads, one going to 12 volts, the other going to the ECU. The ECU shorts that lead to activate the injector. In saturated injection, it just leaves it grounded for the whole length of injection.

With peak and hold, (as best I can tell from my observations) the ECU shorts the lead to ground for a very short period of time, then switches it to ground and back extremely quickly, like a light dimmer to limit the current flowing through the injector.

The way that my circuit worked was that whenever the ECU grounded the injector, it also grounded the "trigger' wire on the transistor. The transistor is a PNP (2N3906), so whenever a little bit of current flows from the emitter to the base, I can measure 12 volts at the collector.

The problem that I had with the circuit that did not have a capacitor is that the transistor was switching fast enough that I could see the actual pulses of the "hold" part of injection (where the ECU pulses ground very fast). I was getting a long pulse of 12 volts, then a bunch of short pulses of 12 volts. This wasn't very useful because I wanted to measure the "length" of the short pulses.

The last schematic you see has the capacitor which "holds" the voltage up enough that (hopefully) the guino would see it as one long pulse.


As far as the actual values of the resistors and stuff, most values are not too important, except for the capacitor and the resistor that was across the capacitor.

The resistor between the ECU and the base was 220 ohms, if I remember correctly, the one between the base and emitter was a very high value (to ensure the transistor turns all the way off), something like 10k to 1M.

The capacitor was 4.7 uF, and the resistor I used was 100 ohms, I think.


Something I noticed was that the ECU grounded the lead of the injector when the car was turned off, so that would put constant voltage through the "load" resistor (the one across the capacitor), and because the resistance was so low, it would get kind of hot with the amount of current that would flow through it. The way I solved that was by powering the circuit on a lead that was only hot with the engine running.

Hope that answered your question

[alszik]

Thank you for your detailed answer! I will definitely try your method when I get an oscilloscope