The misfire learn and misfires in general are a common question and problem with the V8V range. It occurs to me that I do not actually know exactly how the detection sensor/software works.

I am hoping Advantage-Dean, the guys from VelocityAP or JdoubleU or others can enlighten me with facts.

From what I understand, the engine misfire is monitored via the flywheel. I presume there is a sensor for the starter ring or similar that notes the rotation speed. Eight times per rotation when the engine management is providing fuel to the cylinders, the flywheel is fractionally accelerated by the combustion in a cylinder. I presume that if the engine detects that it sent fuel and a spark to a cylinder, but there is no or low acceleration, then by definition there was a misfire.

Two questions occur to me. Firstly how does the sensor measure the rate of acceleration change per cylinder firing? Is it as simple as counting the time between starter teeth passing by the sensor or is it more clever than that. Taking smooth derivatives of experimental data is a notorious problem in science and engineering. This may be the route of some of the problems noted over the years with the technology I believe was borrowed from Ford.

Secondly, what is the misfire learn actually doing? Taking the 4.3 where a learn can be invoked at any time rather than the more complex 4.7 where it must be triggered, if the engine is on coast down, does the management unit introduce trial amounts of fuel and a spark to various cylinders and see when an acceleration change is or is not detected so that it has a baseline for real world events. It does this occasionally when the engine is on run down and there are no other demands on the engine (AC, steering input, braking etc.) just to keep the system up to date with an aging engine and transmission.

I am just curious to learn a bit more about this aspect of the engine.

Many thanks

Philip

 

(Deleted accidental double-post)

 

Philip,

Generally engine management will employ a couple simultaneous methods to determine whether misfires occur.

Naturally, there are reams of literature available to get down into the nitty-gritty details, but simply put—the first method is “flame plasma rectification”, which essentially monitors whether electrical current from the spark plug is discharged as expected, whether there is preignitiom, no ignition etc. You can look this up and read all about it.

Second is monitoring angular crank speed—compression cycle will slow the rotating assembly, which will then accelerate at an expected value upon ignition. If acceleration of crank is not as expected, and can’t be adjusted with acceptable mixture and timing modification, the management will record and eventually trigger a misfire fault if it happens too many times.

Both 4.3 and 4.7, along with virtually every car manufactured after 1996 for the USA (OBD 2 compliance) are going to be monitoring for misfires 100% of the time. Not like there is some occasional test sequence every so often, it is real-time monitoring.

Re: misfire “learning” procedure is likely the management system calibrating the baseline metrics under controlled circumstances (coast down procedure) to account for environmental/equipment changes and improve misfire detection precision.

Modern engine management is extremely powerful. You can be pretty sure that a substantial problem is going to trigger a fault. If you have no faults, and the car/engine is operating normally, unlikely you have anything to worry about.

 

ADNH, thanks for the reply. I had not considered measuring the combustion effects on an electrical arc to detect misfires. Could that be why some coil packs have 3 wires and some 2?

I am still wondering exactly how one measures the rate of change of angular velocity of a flywheel thousands of time per second. I know nothing about engine design, but quite a bit about sensing and signal analysis. I imagine that writing robust code that differentiates a position/time signal from a flywheel position sensor is not trivial.

My interest is part theoretical and part practical. The scuttlebutt is that some 2015 V8V models got updated engine management software from the 2014 models. AM then had difficulty getting all engines to learn the misfire procedure off the assembly line and then not throw phantom misfires after a few more miles. The (a?) fault was identified in the new software and an update issued. So far so good. The downside was that some 2015 models left the factory with the fragile software having passed learn and then not thrown phantom codes. So long as a relearn was not ever required, all was well. However, if a relearn was required it became difficult to get the misfire algorithm to learn (which it did successfully) but not then detect phantom misfires later.

Guess what, I have a 2015 that is now throwing phantom misfires after a successful misfire relearn. The cure apparently is a new (old?) software push. I will get that in the spring when I can drive 7 hours each way with no danger of snow or ice on the route.

In the meantime, I am just interested in the technology.

Philip

 

Crank position sensors use Hall effect (magnetic flux) to create a voltage output which is calculated to rotational speed. It is straightforward technology, been around for a long time in the automotive landscape, very reliable and precise. There is no issue in creating and capturing a reliable signal.

I have a PC-based oscilloscope which can chart frequencies on a nano-second time base with a sample rate of 1 gigasample (that’s 1 billion samples per second). Extremely sensitive, extremely precise. I can easily use it to back probe a crank sensor and graph the acceleration and deceleration of the crankshaft. It’s an incredibly powerful piece of equipment, and to put things into to context, only cost about $1000. If you are interested in this sort of thing, you might like to get one (PicoScope) to play around with.

It’s the only way you’ll know if those “phantom misfires” are actually “phantom misfires”.

 

Hello Philip! This is just a quick reply to acknowledge that I've read your post and will reply in some more detail when I get a chance. On the Visteon EEC-VI engine management system, the ECU has a dedicated processor called NNMM (Neural Network Misfire Monitoring), but the Bosch MED17.8.31 the later V12 engines cars doesn't, so there's some differences and overlap between the two. Will explain more a.s.a.p.