Gallons Per Hour Error?

[dcb]

I'm coming back around to a 100k trim pot in parallel with the zener for now, with optional capacitor for pwm based hold circuits. I do want to keep the emphasis on precision however, that is the main advantage the mpguino has over OBD solutions post 96.

I got a peek at an existing peak&hold circuit and not including connectors it had twice as many parts as the mpguino. Basically they combined every option we have discussed thus far, and it still needed trim pot adjustments and resistor swaps

So some guess at the pwm frequency is probably in order. The pwm in this, possibly unrelated to anything automotive, graph has a period of 600uS. We are subtracting a 500uS "mechanical delay" from each pulse. So it is believable.

Unfortunately with a trim pot in the circuit it means the the target reactance of the capacitance also moves around, which means two trim pots for the subset of injectors that are both peak & hold AND where the hold circuit was implemented with PWM. I don't know how to guess at those numbers.

[Coyote X]

Know of a simple DC inductive clamp setup? If we could measure current on the injector wire without cutting it then we would have a perfect 100% setup that always worked.

All injectors pull more than an amp when on and 0 when off so an inductive clamp could read that without cutting the wire. We just have to figure out how they work reliably.

[Coyote X]

27k is a bit much. It misses half the injector pulses. I picked up a 47k potentiometer at Radio Shack today and will put it on there.

One other thought we could do is a voltage comparator. Link. This is a very basic schematic that is generally what I was thinking of. Just the bottom half of the circuit is needed so it outputs when the voltage drop from vcc(battery voltage) is greater than 2.5V. The 10k pot in that schematic could be replaced with a pair of resistors once the value is known to make it trigger on a 2.5V drop or whatever is needed. So that circuit would add an op amp and 2 resistors to the overall design.

That type of circuit should work with just about any injector system and be very tolerant of just about anything that I can think of, as well as give a pretty clean output to the mpguino with just a standard resistor/zener input.

[dcb]

I think in version 2 we can look at using the built in comparator interrupt on the atmega. You can set a rising or falling interrupt relative to an external voltage reference. For pwm we can basically wait for the signal to stabilize and back up in time to when the stabilization occured. But it is going to have to wait a while unless someone runs with it.

[Coyote X]

Been messing with the pot mounted across the zener and have pretty much figured it is ok but I don't think it is going to work out for a 100% solution.

I had the scope and laptop in there running along with the mpguino and I can't get a good position that it reads everything. If I set it so it runs good without missing pulses at off idle speeds it reads high at idle slightly. If I set it to read right at idle it misses some pulses. With the alternator unplugged it reads way off.

So Back to the voltage comparator. I think it is pretty much going to need an op amp to be a 100% accurate solution in all situations. Using the built in comparator won't work since it would have to go to polling to find the injector open time and not use interrupts. With an op amp it can be a 8 pin dip chip and 2 resistors. That would read pretty much any injector signal. I will try and build something this weekend and see what I can figure out.

[Coyote X]

So after thinking about it if the ECM is wanting to maintain a 1 amp current and the resistance of the coil is 3ohms then it will require 3 volts minimum to maintain on. So the voltage drop will always be constant across the injector at 3 volts and that means more or less constant 3V drop across the injector pin. So for a basic proposal for how to measure a constant voltage drop

             12V
               o--------|
                        |
                        |
                        z 3V Zener
                        A               LM741 op amp
                        |
                        |                |\|
                        o----------------|+\
                        |                |  >-  out
             INJ -------)----------------|-/
                        |                |/|
                       .-.
                       | |
                       | | 10k
                       '-'
                        |
                        |
                       ===
                       GND

This could be an add on board in front of the stock MPGuino if the op amp had 12V and gnd hooked up. The biggest problem is that a 3V zener is pretty hard to find so a slight modification to the circuit will let it be tuned from a 0-5V drop...

             12V        .--------------.
               o--------|              |
                        |             .-.     100k pot
                        |             | |<------.
                        z 5.1V Zener  | |       |
                        A             '-'       |    LM741 op amp
                        |              |        |
                        |              |        |     |\|
                        o--------------'        '-----|+\
                        |                             |  >-  out
             INJ -------)-----------------------------|-/
                        |                             |/|
                       .-.
                       | |
                       | | 1k
                       '-'
                        |
                        |
                       ===
                       GND
(created by AACircuit v1.28.6 beta 04/19/05 www.tech-chat.de)

That gives us a way to precisely calibrate the voltage drop.


All the peak and hold injector circuits I could find are 12V > fuse > injector >ecm(gnd) so measuring INJ at the ecm and assuming the 12v going into the mpguino is not significantly different than the 12V side of the injector this circuit should work. I will build one tonight and test it once I can get the van out on the road after the blizzard goes away. If this works it could be added to the standard circuit but it is adding a good bit more mess to it. It should be a 100% solution though

[Coyote X]

Went ahead and tried the op amp circuit. I used a pair of 47k resistors instead of the variable pot. This should give me a 2.5V drop for the injector. From the tests in my driveway this seems to be about right.



I will drive it around a few days and hook the laptop back up to the ALDL to verify that the numbers track evenly. I can brake torque it in my driveway and the numbers are dead stable. I can go up to the point the back wheels start spinning and the gallons per hour seems about right with the amount of throttle I am giving it.

I will try it out in my metro tomorrow to make sure the circuit still works with every type of injector but so far so good

[Coyote X]

Got bored and it was nice outside so I went out and tested it. It works good until the pulse width gets pretty long. The voltage drop on the injector decreases past 2.5V drop towards the end of the pulse. So I put a potentiometer in place of the two resistors so I could vary the voltage drop it is looking for. Turns out it works perfectly 100% with a voltage drop of ~0V to 0.5V. I can turn every accessory in the car on at once and the numbers don't change other than when the computer responds to the extra load on the engine and opens the injectors slightly more. There is no jitter or variance from what the laptop is reporting from the ALDL port.

So it might be possible to do away with the zener and potentiometer and just go straight for 12V on the - input. I think it needs a voltage drop of 0.1V on it though just in case there is a tiny difference in voltage on the MPGuino compared to injector voltage. Maybe just a standard silicon switching diode could give a 0.7V drop and skip a lot of the mess in the circuit. I will pick up another op amp this week and build a breadboard interface circuit for the official MPGuino.

Maybe use a 1N34 diode if I can find one, I think they had a pretty small voltage drop of like 0.3V or something and very little capacitance. A bit more experimenting and I should have the simplest circuit I can get on there figured out. Right now it is 3 components a zener, resistor, and pot. It might also be possible to just connect - straight to 12V and use a potentiometer on the offset null to tune the circuit. Once I get one on a breadboard I can try a few things and see what I can come up with that is reliable and simple now that I know the basic idea works good.

Also the ascii drawn circuit has the + and - reversed if anyone wants to build one

[Coyote X]

I still have no perfect solution to the peak and hold injectors without cutting the injector wire. The opamp circuit works just fine and shows ok looking on the scope. It doesn't give a single pulse but a double pulse. When the voltage spikes from the inductive kick it goes high again. So the only thing I have been able to think of is maybe when coasting down with the injectors on just at 1% duty cycle or so the arduino is missing the interrupt signal and not stopping the injector open timer. I set the delay to 0 to try and take that out of the list of things that could cause problems. Didn't seem to do anything.

If missing interrupts is the case then the only choice is to cut the injector wire and put a current sensing circuit in there to trigger the arduino. The signals are pretty short on the scope. A current sensing system could trigger anytime there is current flowing so it would get rid of the problem of the inductive spike making a double pulse out of the signal.

If it is something else I am out of ideas to try I am probably just going to try building a current sensing circuit and see what that can do for me. It won't be ideal having to cut the injector wire but it at least should reliably read the signals.

For the simple circuits the pot across the zener seems to give good readings as long as the voltage is stable. The opamp gives good readings for various voltages but then goes nuts on very low duty cycle signals. If I can't do any better I think the pot across the zener gives better numbers since the voltage jumps don't seem to be that big of a deal compared to it showing 12-18gph while coasting or slowing down.

[dcb]

Hmm... maybe a clip on hall effect thingie?
Hall effect - Wikipedia, the free encyclopedia

edit: $11 from mouser, not clamp on but looks pretty straight forward:
http://www.tamuracorp.com/clientuplo...L18PXXXD15.pdf