Gallons Per Hour Error?

[Coyote X]

It is a minor issue either way. I am thinking a 25k or so resistor might do better but I think a pot that is set default to 100k would be a non issue for 90% of the cars. But those few cars that have problems it could be a quick matter to adjust it with 2 people in the car. One driving the car with enough load to make it read incorrectly and the other adjust the pot so it starts reading correctly. It should take like 30 seconds at most to adjust it for the few that need it. The 22k might be a bit low for me I noticed when it is sitting idling with every accessory on to try and lower the voltage I could get it to spike to 5-8gph here and there. So maybe shooting for 2.3-2.4V when off might be ideal.

But for now I would just put a notice on the order page about peak and hold injectors will not work as-is with the first revision of the board and think about adding a pot or whatever to the future boards.

That last video was measured at the AVR pin so it is exactly what it is seeing before the resistor is added. The scale won't mean much though because the probe I am using is broke so I am not sure about the voltages.

[dcb]

I appreciate it.

Also, here is more analysis, rough guesstimate that the major years of issue are 87 to 93, just to get an idea:

Of the 55116 cars on ebay of 1987+ vintage, 2219 are 93 or older (%4).

Of the 1873 cars in the ecomodder garage of 1987+ vintage, 307 are 93 or older (~16%). As expected, we hold on to our cars longer , which is cool except for the higher probability of hitting a peak and hold system.

If you are not sure, measure an injector before you order I guess.

[Coyote X]

I will try a few resistors this week and see what I can figure out. Adding a second set of holes right beside the zener would not really be an issue for the next round of boards but I still think a 100k trimmer pot would be best. If it was left off most boards like the pull up/down vss resistors. But have the holes available for anyone that needs them to run to radio shack and pick up the pot I linked to earlier. Even the pot could be soldered directly on top of the zener pretty easily, just bend the center pin over to one side, so a redesign might not even be worth doing.

I probably won't get much done tomorrow I have to use the van to haul some people around so I just had to clean my mess up in it today and put the dashboard back together. But this week coming up I will try a few more resistors and see what the scope shows me.

This might be more of an issue for the MPGuino than just 10-15%, since a lot of 96 up cars will have people that want to use an OBD2 type setup. And I would guess that probably every GM and Ford built TBI setup is peak and hold. So that will be a lot of people who have peak and hold in the 80-95 range. Also I checked the Holley Commander 950 in my kit car today and it is also peak and hold injectors

[dcb]

Update, I tried these mods to the circuit, it worked on a saturated but I need to get the resistor values dialed in better. The idea is to make the injector firing detection relative to the supply power.

Tapping into the other side of the diode provides a .7 volt gap from system voltage, so there is some built in-tolerance for the hold driver voltage.

Still sorting it out though, not a final recommendation, need to figure out RX and R2, 2n3906 assumed. Changes in blue.




Here is a retrofit of the circuit using ~50k resistors for rx and r2 (again not final), resistor to zener trace has been cut on the bottom side, blob of hotglue would be a nice touch:


Adds one transistor and a resistor to the design if it works.

Weird thing is with R2 and RX at 50kohms, the CPU pin voltage was limited at about 2.5 volts. Is it possible the current is so low that the high impedance cpu pins are doing the clamping?

[Coyote X]

I know the 10x probe was messing with the signals at the cpu so it is pretty high impedance there. I didn't really look into it much other than notice the 10x probe didn't work and moving on to something else.

I will try building that circuit this weekend if I get time. One thing though is pretty much nothing is going to be completely 100% in all situations except for measuring the injector current somehow >~0.5amp and outputting a voltage for the arduino to see. I don't think anyone wants to cut their injector wire to do that though

I messed with the resistors on mine and with a 22k and a 3.3k on there giving me 25k ohms seems to do the trick. I can make the injector signal jump if I am at idle with the wipers and rear blower on high, then simultaneously turning on the headlights and front blower on high. I can get it to read incorrectly for about one injector pulse till the alternator gets the battery back over 11V. Other than that it seems dead stable at around 25k. 27k ohm resistors are a standard size so I would say just use a 27k if you are lazy and don't want to modify it much.

[dcb]

I don't know if there is a 100% fix either. I don't know what happens to the hold voltage when the alternator goes out. I am assume it is a function of supply voltage, but it is conceivable that they got fancy with the hold voltage drivers, in which case supply voltage could drop below the hold voltage.

resistor experiments:
r2>=2520 ohms (base current @peak =.005a @hold (11 volts) = .00063a)

A 50k seems to work too for r2, so I don't think it is too critical. And a 50k offers much more protection than 2520 so I might just go with that. it means about 0.0000036 amps through the base in "hold" which is about 72 times the base cutoff current, and .0003 amps at peak.

rx=50k ohms

[dcb]

I keep forgetting to mention that this circuit in post 24 inverts the injector signal, so you have to switch the occurrences of RISING and FALLING in the source code.

[dcb]

Ok, 50k ohms it is, for rx and r2. I did some bench testing. With a 16.2v supply the pin went high at about 14.9v. That should translate to any injector signal that is 1.3 volts or more below system voltage registering as "injector open".

With 50k on two of the transistor leads, it will be pretty well isolated and should prove reliable.

It will also allow the strongest signal for the saturated injectors (the majority of vehicles have saturated as far as I can tell), whereas the resistor patch attenuates the signal for saturated injectors.

And it seems to be operating within established parameters.

So I like this solution now

Coyote, I'll ship this guy to you so you can play with it. You are short a guino so this works out pretty nicely

[dcb]

welp, the transistor version failed regression testing in my saturn miserably

I tried 50k, 4.7k, even tried adding another diode in series to the emitter so that the drop was at least 1.8v before considering the injector open. The change in resistors had no effect, the instant gph would go way high as compared to the control guino and the resulting fuel used was also high.

Adding the diode introduced a reboot whenever I hit the gas eventually had to reflash the bootloader, stopped doing that when I shorted out the diode. wierd.

I'm running out of inspiration here. This fix for peak and hold is turning into a whole resistor/diode network. Coyote, I want to hold off on shipping a sec and see if any other simple fix ideas come up so I can get you a more "production" like unit.

[Coyote X]

I need to change the resistor across the zener to 27k. I really think that will fix it. I had to haul a neighbors truck home when he broke down and I watched the mpguino running with the 25k and it didn't jump or do anything erratic at all. I didn't try on purpose to make it jump by turning on a lot of stuff at once so I think 25k is pretty close. 27k should be pretty much right in the safe spot so even if the alternator is disconnected the injector will probably quit firing before the mpguino quits reading.

Once I get it where I am happy with it I will stick it in the Metro and scope it out and see how it reads. I don't see any problem with it working just fine on a saturation injector with the resistor across the zener.

I will swap out and put a 27k in tomorrow. I figure if it works 99% with a resistor there is no point in getting really nuts about that last tiny bit of precision. Also I went from 19mpg empty to 14mpg with an extended cab ford ranger on the back