MPGuino release one workspace

[Coyote X]

I think the opening delay is not really that important, it will average out over the tank and with the calibration value. But if you are going for that absolute precision over say like 1000 miles it might throw the numbers off a percent or two

I don't think it is worth adding it now. But when coding it may be good to keep in mind that it might be added in later. So that could help keep from making changes that would make it really difficult to add this feature in the future.

[dcb]

Another thought for me_with_no_oscilloscope is to add some more self diagnostics (like the free memory and CPU utilization screen) to the guino where we tie a couple of the remaining analog read pins to the VSS and injector signals and do some analog reads and keep track of the peak voltages that way. If we are at 1024 then maybe we need to attenuate the signal a bit.

[Yoshi]

Hello all,
Let me explain something...

Regarding to the VSS parameter, it is always consistent or proportional to driving distance. 1% error is always 1% error regardless on a city or highway driving.

Regarding to the injector parameter, it's another story.
On my Toyota Prius, I see approx 2msec idling pulse width, 4msec cruising and 8msec accelerating.
Is the fuel consumption is proportional to the pulse width? NO!
We have to know the injector's mechanical delay. In my research, they are between 0.3msec and 0.7msec.
My original SuperMID M-1 design, I thought I should have this parameter as variables. However, I set it 0.5msec and took a lot of fuel consumption data on a city and highway driving, then I saw cosistent results within 1% variance.

I don't know the 0.5 msec is exact one or not, but there is a certain one on that. You'll see more than 1% inconsistency without it.
My recommendation is use this 0.5msec delay to see a stable results.

Regards,
Yoshi

PS. BTW, an oscilloscope does not help this.

[ttoyoda]

MilesPerTank,
Thats an idea. I was thinking maybe a little capacitor before the zener, to give the zener time to turn on. And when I do mine I am probably going to wimp out and use opto-isolators, and schmidt trigger comparators behind those. I am going to have to condition my signals anyway due to the hall effect switch I will need to put in the speedo. I have no vss sensors.

dcb,
I would think if the pulse is short enough, it may occur and die away before it can be read by the cpu. Again, I have no idea if this is an issue, but I worry about slow and undetected damage to the cpu chip input over long periods of time. Even the digital scope used to show the 65? volt spike may not have detected the peak, if the peak was short enough. We also don't know how long/how-much-capacitance the scope leads were. That scope screen looks like a pc running windows, which suggests it is not exactly the highest MHz on the block.
I am happy to be proven wrong, though.
Also, I want to say you are a genius for coming up with this thing, and I salute you.

[dcb]

Originally Posted by Yoshi

I don't know the 0.5 msec is exact one or not, but there is a certain one on that. You'll see more than 1% inconsistency without it.

I believe you, I just wanted to understand it. It is hard to argue with empiracle evidence

So we need to make sure that we are not double counting injector pulses before we subtract 500uS per pulse.

Originally Posted by Yoshi

PS. BTW, an oscilloscope does not help this.

The oscilloscope is to make sure we are not cooking the atmega. Though a graduated cylinder might be handy.


ttoyoda, the thought is that we take many many many readings while in the wait part of the main loop and only look at the max value we collect. and if we are at 1024 then we know for sure we are at or beyond the limit at some point during operations. Not that we will detect every peak and its absolute value that reach our $3.00 chip.

[MilesPerTank]

From my measurements a 10Hz signal generated an IC of 20.

I can double check monday.

[ttoyoda]

ttoyoda, the thought is that we take many many many readings while in the wait part of the main loop and only look at the max value we collect. and if we are at 1024 then we know for sure we are at or beyond the limit at some point during operations. Not that we will detect every peak and its absolute value that reach our $3.00 chip.

I see your plan on the loop. It is a good plan.
It's not the $3.00. I plan to put this in my car, integrated into the dash. It will look like it was always there, when I am done. I will put it where the seat belt light is now, maybe. I want it to be as reliable as everything else in the car. I don't ever want to have to take it out. And locate and re-discover the tools and software and cables and procedure to program a new chip.

You have done a great job designing this thing, have spent hours and hours with all the parts, and so to you it is all very easy, and I understand that. It's like me with carburetors. Hangers-on like me might be able to follow what you have done well enough to ape it, but for us it will never be as quick to do as it is for you, we will never have the amount of practice you have.

[diesel_john]

Originally Posted by RuHaSs

DCB,
Here is a nice link on injectors.
http://www.megamanual.com/v22manual/minj.htm

good info

this i found interesting

MegaSquirt® assumes NO fuel is injected during opening (which is close to true, since the injectors remain closed until the coils charge, then they snap open at the end of the opening time).
1.0 milliseconds is considered the 'standard opening time'.

[ttoyoda]

on the flyback issue is that why breaker points have condensers or would that make the signal ring worst.

The condenser (capacitor) is there so the points have the time to get far enough apart before the voltage rises to the point it would have jumped the gap when the points were closer together. If that made any sense. A snubber circuit could be made with a resistor in series with the capacitor, to dampen the ringing. I was thinking of a very small capacitor, like power supply decoupling is done for ICs with a little capacitor next to every power pin.

And to add what I said earlier: I think this invention is getting very close to where you could sell it as a finished product, in an enclosure with velcro on the back. But to get to that final step, the interface to the outside world should be hardened a little.
You don't want to be getting returns and paypal chargebacks because someone zapped the input and it corrupted the software for example. I recall reading somewhere here that a MPGuino was acting funny and had to be fixed, although I don't recall the cause or the fix.

[dcb]

Originally Posted by diesel_john

good info

Thanks for pointing that out. Part of the issue is that everyone talks about opening time but hardly anyone mentions closing time. If you are an ECU, then you care about the opening time (and a few other variables) if you are trying to get the timing of the squirt to coincide with the opening of the intake valve.

But that is only part of the "how much fuel is delivered" equation, the others being how long do you hold it open, and how much will it squirt after I cut the power to the injector.

When I looked for a diagram of opening and closing fuel flow I found this:

where it looked like a complete wash, the area under the closing curve seemed to fill the gap in the area under the opening curve, which was Monroe74s point also.

But when you look at the megasquirt, you see an entirely different area under the curve for closing ( in blue):


This shows much less fuel delivered after power is cut off to the injector (described as less than .1mS) but there is a definate slope to the injector flow rate on opening. So I'm satisfied for now that there is indeed a difference between the the flow at opening and closing based just on this information, and since Yoshi has actually tested it, I'm content to call that difference 500uS. Speak up if you have an issue with this AND are willing to test in a most controlled manner under exacting conditions yourself