Sack of ECUs with locked timing on 2jz-gte

[gurov]

image.png (2.62 MiB) Viewed 813 times

image.png (4.77 MiB) Viewed 813 times

the setup is the same, ignition advance locked at 15 degrees.
iphone on a tripod, pointed at the intake cam sprocket.

actual advance values don't matter much, we're looking for how much timing jumps around when engine accelerates/decelerates

from bottom left, clockwise:

MS3PRO PNP:




AEM EMS v2:


rusEFI UAEFI:


Speeduino Official Board:


Speeduino Self-roll :



i haven't really watched these too closely yet, they seemed pretty close, with megasquirt being real solid, then rusefi, then AEM, then speeduinos.

these tests were done basically within the same 10 minutes.

36-2 crank angle sensor, single tooth CAM for phase. both HALL

discuss, fight, whatever..

i'll be back to fill in a few more details later.

[AndreyB]

Damn, how do we build objective data/charts from this?!

[gurov]

can’t measure it - can’t fix it.

i’ll try taking a video with a different camera (d800) and lens to make it a bit clearer, something i can control aperture on.

maybe we can rig up an f407 disco, or a secondary uaefi to be a watchdog, where hall3 input is cyl1 ignition. perhaps with knowing the advance and where it should be with a locked timing, we could parse the same trigger wheel and calculate the difference in ticks of actual vs scheduled.

this may require a bit of brute computing force, so fastest MCU we could shoehorn into this, with basically all facilities shut off, majority of interrupts disabled, no fancy things. just counting crank tooth edges, tracking timer that keeps up with ticks and running analytics on what the actual falling edge of cyl #1 is

with this data, it would be trivial to build a map of RPM x LOAD to ignition hardware latency.

megasquirt and aem have a static field for spark latency already.

[gurov]

also, taking this a bit further, since that just basically becomes and automotive signal sniffer/analyzer. you could use this to force an OEM ECU to give up its secrets.

take in cyl1 for timing (or all of them, why not) - timing map
take in inj1 for timing AND pulse width (of all of them, why not) - think direct injection timing.

take in iat/clt/load/tps (without bias resistors, to just take a look) for warmup/different condition tuning

[gurov]

would be pretty trivial to do for me, another patch through harness between the ecu and engine with sniffer device

maybe we can connect timing light's inductive pickup to VR input to capture what the timing light sees ?

[gurov]

@AndreyB - perhaps the part of the code that tracks the current VVT offset could be used with a magnetic pickup on cyl1 to attempt to capture the drift of a particular ECU at a certain RPM, using an extra patch harness.

[mck1117]

maybe we can connect timing light's inductive pickup to VR input to capture what the timing light sees ?

why bother with a pickup when you can just directly probe the igniter drive signal?

[gurov]

hmmm, yes, falling edge cyl1 into cam signal with VVT tracking, interesting. i'll get the other patch harness ready i guess and test this on the bench