t vago's MPGuino workspace thread

[Ardent]

Hmmm, I wasn't specifically watching RPM, but I would say the issue did show up more frequently at a lower RPM -- somewhere around 2000 or below is my best guess.

[t vago]

Examined the code some more, and I found a bug in the part of the code that calculates the minimum valid RPM countdown value / maximum valid injector on-time value. Not even sure why this wasn't caught with testing on the Karen-mobile. No matter - going to test it tomorrow.

[Ardent]

Sounds promising!

[t vago]

Tested the bug-fix, and it did not resolve this "----" issue. I tried to simulate the conditions that would cause "----" to appear, with one of my spare Arduino Uno boards. It was long overdue for me to create a MPGuino signal simulator, anyway. Only problem was, for all of the conditions for where the "----" would show up in my code, 999999 would show up with the original 0.86 code. I tried various combinations of injector pulse widths and injector pulse frequencies (corresponding to on-time and RPM), along with feeding a VSS signal, and for all intents and purposes, both my code and the original 0.86 code behaved correctly.

The only other thing I can think of is that the Injector Response Delay time (550 us) might be a bit too much for your injectors, Ardent. You might try going down to 450 us, or maybe 400 us.

[Ardent]

Injector pulse width is 3 to 4 ms at idle. How much greater than the dead-time does a pulse need to be in order for it to be recognized?

It seems that the algorithm used requires a greater "proof" pulse width to "wake up" from standby or resume from DFC.

The Fundamental Logic board has a 100k resistor on the VSS and INJ inputs. This is greater than shown on the Arduino Assembly diagram on the wiki page, but the same as that shown for the Iduino. What values are you using, and could that have any impact as far as the amplitude of a pulse?

This morning I started the vehicle and the display remained dark. Much like yesterday morning, a throttle blip woke the unit and it then functioned normally.

I'm still planning to test with the 0.86 code.

[t vago]

Originally Posted by Ardent

Injector pulse width is 3 to 4 ms at idle. How much greater than the dead-time does a pulse need to be in order for it to be recognized?

It's not so much a matter of injector pulse length, as such, but of injector delay response time. When current is allowed to flow through the injector solenoid, there is a delay between when that current flows, and when the magnetic field of the injector solenoid is strong enough to move the injector pintle so that fuel can flow. Typically, it's about 500 microseconds. The MPGuino code (both mine and the original code) took this into account. Even for an idle injector pulse width of 3-4 ms, 500 microseconds should not be an issue.

Hm....

Originally Posted by Ardent

It seems that the algorithm used requires a greater "proof" pulse width to "wake up" from standby or resume from DFC.

The Fundamental Logic board has a 100k resistor on the VSS and INJ inputs. This is greater than shown on the Arduino Assembly diagram on the wiki page, but the same as that shown for the Iduino. What values are you using, and could that have any impact as far as the amplitude of a pulse?

The JellyBeanDriver board I'm using has 47K resistors for VSS and INJ.

Originally Posted by Ardent

This morning I started the vehicle and the display remained dark. Much like yesterday morning, a throttle blip woke the unit and it then functioned normally.

I'm still planning to test with the 0.86 code.

See, that does not sound like any sort of injector delay issue. Rather, it sounds like the injector signal just isn't being picked up. Please do test with the original 0.86 code, and let me know how it goes.

[josemapiro]

It's something we had noticed, there schemes that have resistenciads 100k and others with 50k, not better a system presented here http://ecomodder.com/forum/19866-post2.html
I'm not the best person to talk about this, but it seems that this system has more protection than the one used.
What you think.

Thank you
José Rodrigues

[t vago]

Originally Posted by josemapiro

It's something we had noticed, there schemes that have resistenciads 100k and others with 50k, not better a system presented here http://ecomodder.com/forum/19866-post2.html
I'm not the best person to talk about this, but it seems that this system has more protection than the one used.
What you think.

I can't access it - Chrome gives me this warning:



Is this page referring to the use of optoisolators to shield the microcontroller from the vehicle electronics?

[josemapiro]

Originally Posted by t vago

I can't access it - Chrome gives me this warning:
Is this page referring to the use of optoisolators to shield the microcontroller from the vehicle electronics?

lso shows statement, but I ignore and Veijo the same.
This is the picture that was posted by DCB.


Thank you
José Rodrigues

[t vago]

Okay... Those are simple buffer circuits. They would certainly work, but my opinion is, is that it's over-engineering the circuit.

One would get the same benefit, merely by changing out the 100k resistors for 47k resistors (or even 22k resistors).