Injection Pulse Sensing on 1973 Citroen DS

[Bristol603]

Yes - that's the schematic. I'll tap between the ECU and the injector and set the injtrig parameter to sense positive edges. Thank you for you help.

[dcb]

is there a resistor between the ecu and the injector? you want to tap between the ecu and the first resistor if there is an obvious place to do that.

[Bristol603]

No - I think that resistor is modelling the resistance of the drive transistor for the injector. Will it confuse the pulse sensing if the steady on-voltage across the injector is only 3V? If yes, then would conditioning the sensed signal with an op-amp to a saturated 12V level solve the problem?

[dcb]

it's possible to condition that signal with an op-amp or possibly just a transistor if you bias it right.

it looks like the resistors might be internal to the ecu so you might be able to tap into them there also if you can identify them (will be about 6 ohms away from two injector leads). the point of interest is in the upper right of this picture, where two 6 ohm resistors come together:

[dcb]

I started a wiki on mpguino interfaces with a draft of what might get it for a ~4volt hold 12v peak positive signal d-jetronic in there, lemme know if it looks like it might work:
Interfaces - EcoModder

[mwebb]

as measured on the signal side
the side that is NOT powered up but the side that gets grounded by the ECM

peak and hold


saturated driver


saturated driver at WOT snap
note in this image in the upper left corner , DFCO decel fuel cut off
if you have the scope on the correct pin , and you snap the throttle to wide open , for a tiny little bit , just enough for the engine rpm to get high
then release the throttle
as the engine slows down , the ECM should shut off injector signal
as shown in upper left , if your scope does not show that and the waveform looks weird ... compared to examples above .
you are connected to the wrong injector pin

[Bristol603]

The interface circuit looks to me like it should work and bring the signal to the guino back to something more like saturated-low-is-open. I take it the diode is to protect the NPN from the negative V pulse as the injector closes. My normal world is <45nm CMOS and so these analog circuits are a bit strange to me. I will build this when my guino arrives and let you know if it works.

Did we get to the point of deciding whether the D-Jetronic is pulse and hold or saturated? It seems to me more like a saturated type of control with low resistance injectors that reduce the saturated on voltage across the injector to something inconvenient.

What is the input impedance of the guino for the injector pulse? I assume this is high enough to have no effect on the interface.

[dcb]

I think it qualifies as peak and hold of sorts, but either way the bulk of that signal looks too marginal to reliably trip an unmodified mpguino.

inductive load is why I threw the diode in there, yup.

I've never measured the impedance, but there is a 50k resistor in series with a 5v zener. Here's the repository, the png is the schematic and the cpp is the latest code plus some other artifacts.

opengauge - Revision 176: /trunk/mpguino

Mike, is that your metro? That looks like the signal coyote was dealing with. I'm really swimming around just counting the time between those nice positive peaks somehow, assuming the preceeding low part is a constant. bristols signal is a little different though.

[mwebb]

that is not my metro
but i do have a G10 metro and could send detailed injector waveforms

thanks to JellyBeanDriver
i have a MPGuino too