For the technical minded: Crank sensor trace printoff - Confirmation?
Turbo Mini Forums

Take a look at this: (click on it and it'll appear in full-screen)



Link: http://www.davecoxon.co.uk/CRANK_SENSOR.jpg

Have I got this right do you reckon - basically, are the points highlighted in the trace, relevant to the EXACT points on the toothed wheel.

I'm 99.9% sure it is - I'd just like a few others to confirm. This is so I can set my VR sensor in EXACTLY the right spot for TDC and know where this falls on my traces when i take the injection timing events from the MPi.
We can then work backwards to figure out the exact injection angle (relative to TDC) for the injection event - and see what it does WRT RPM and Load...




Before you ask - those are VERY old traces when I was seeing what DTA did on the injection event on the E48EXP - where I provided DTA hard, fast proof that their injection event was linked to ignition advance (ie - it moved on its own) - no wonder it ran REALLY badly!
The data I'm taking this time is all digital :) Oh yes!

Edited by turbodave16v on 10th Aug, 2007.

TurboDave,

You're nearly right but you have the reference points back to front.

Assuming your crank sensor traces are from an oscilloscope or alike then the time axis (horizontal scale) normally goes from left to right as time progresses. This would mean that event A proceeds B which proceeds C etc. For this to happen your crank wheel would have to be rotating counterclockwise and not clockwise as shown. Otherwise you're spot on.

I spent the best part of six months characterising these sensors when I used to design ECU's for a living so I'm reasonably familiar with the waveforms.

Good luck measuring the injection timing for the Mpi. I've developed my own ECU for port injection (N/A engine at the moment) which has the ability to control the injection timing against speed and load. I hope to start fully characterising the injection timing needs of the engine in the next few months and will post my findings as they become available.

looks right to me except to keep ABCDEFG in there order the wheel needs to spin the other way.

most applications (including megasquirt) will use a zero crossing detector to decode this signal, whic will detect when the wave croses 0v on either the rising or falling edge, assuming up is +ve you would want to be triggering on the falling edge 0v crossing



reason i bring this up is the 0v crossing detection is only as stable as the 0v to which it is referenced, and any base interference in the signal from the VR sensor. Basically you cant realistically assume that a given detected point = an exact point on the wheel. easiest way around this would be to get the car upand running, fix the spark angle and strobe the toothed wheel! then add corrections untill its where you want it to be. also bear in midn there is hardware latancy, even if you know exactly what part of the wave corresponds to the wheel it takes a finite time for the cpu to do its stuff, as a result there is a small shift in "timing" w.r.t RPM. ms1-extra has a feature to compensate for this, dont know about msII.. all said and done, simply strobeing the wheel and adding correction factors will completely eliminate the guesswork.

turbo 16v k-series 11.9@118.9 :)

Denis O'Brien.

actually, is your trace the wrong polarity? or am i getting confused? im sure its +ve otu the sensor as the ferrous metal moves towards it, then negative as it moves away.

turbo 16v k-series 11.9@118.9 :)

Denis O'Brien.

Most interface circuits do use -ve edge zero crossing detection to generate the trigger pulse. The peaks on the VR signal are coincident with the centre of the tooth and the troughs are coincident with the centre of the gap. This means that providing the tooth-gap dimesions are physically the same then the -ve edge zero crossing will always be coincident with the tooth edge.

Even during cranking the signal from the VR sensor should be several hundred mV and the zero crossing offset for the LM1815 is less than 25mV so is typically smaller than the signal by an order of magnitude. Since the slope of the signal is at a maximum near the zero crossing this small offset will introduce minimal error.

The latency introduced by the software is dependant on how well the code is written. My own ECU uses a simple 8 bit Microchip PIC and has a latency of 12us, my engine has 10 degree teeth (standard Rover pattern) which at 6000rpm means crank pulses every 277us so the 12us latency represents an error of just 0.4 crank degrees, pretty small I think you'll agree.

Regarding the waveform shown it could be that it was taken from a crank sensor wheel with holes rather than teeth (or pegs) as shown in the drawing. The missing hole used as the reference would give the waveform shown.

This was taken back in 2001 ish with the same 36-1 toothed wheel i have now - which is picking up on typical external teeth like shown in the partial wheel shape.
This was piggy-backing on a stock rover sensor (same as i'll be using for the next test).

I could well have had the polarity wrong - I wasn't too bothered for the test this data came from - I just wanted to see where the injection event fell relative to TDC.

So - do either of you know which of the wires on the stock rover sensor (the one with the blue male socket and white/black wires) is positive and negative? I guess that the black is negative, but I'd appreciate confirmation...

What i don't get - is Evo saying that the falling edge is where 0v occurs. Consider the middle of the gap (D) on the wheel - this MUST = Ov? witness it being halfway between the falling and rising teeth either side of the gap? I mean - if there was 20 missing teeth it'd be sat at 0v for a long time - which clearly is not on the falling-rising tooth?

ps - cheers for confirming the reference points being backwards. I realised this whilst driving home. Doh!!!!

Also - Can we confirm if F and B (on the trace) fall in the middle of the tooth, or are these on the corner between the rising (or falling) edge???

iirc F abd B will be the furthurest point from a tooth.

Alex

AlexF

The 0V on the trace correspond to the middle of the gap or the middle of the tooth. One is going up and the other is going down (which is which depends on sensor polarity). So on your graph A would be the middle of the tooth, B would be the edge of the tooth, F would be the edge of the tooth, G would be the middle of the tooth.

What I don't understand is that you're saying that it was from a 36-1 but there are only 34 peaks (or valleys) which would indicate a 36-2. Also the width of the missing tooth transition corresponds to 2 missing teeth.

Jean

http://www.jbperf.com/

cheers rattlytappet for clearing up a few things, was jstu thinking outloud for any possible sources of error, reason i mentioned latency is i know ms1-extra can cause a shift that could be measured in whole degrees, may not apply to what TD is doing but worth metnioning incase :)

as for polarity, hook avolt meter up to the sensor, touching t he snesor with(or against) a bt of iron/steel then pulling it away fast should show a negative voltage if you have the meter on the correct pins of the sensor.

turbo 16v k-series 11.9@118.9 :)

Denis O'Brien.

I'm sure that the EDIS software uses point D to indicate TDC and it would be logical to assume that the MS2/Extra does the same.

Jean has confirmed that the point D would be the centre of the missing tooth, so he has obviously based the siamesed code on that.

I would not get too hung up on this because from a practical viewpoint, we will need some adjustment factor to take account of errors in sensor alignment and pulley machining accuracy.

I'm going to put a small notch on my pully to indicate 22 deg BTDC, which is my normal idle ignition setting with vac advance. This will allow a simple check with a timing gun to check that the MS2/Extra is reading the VR sensor correctly.

Jean, have you used raw VR data for the injection timing and if so can it be factored to take account of the above errors like the MS2 ignition?

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Just checked my Rover VR sensor and it has a blue connector with a black wire and a black wire with white trace. The ve+ lead is the one with the white trace, black wire is indeed -ve.

Firstly - a GREAT Thanks to EVERYONE who has replied. This proves what a great forum this is.

Jean - You are indeed correct about the missing teeth. I had ccompletely forgotten this was a 36-2. I don't remember why this was, but I popped a tooth back a while back.

Evo - Great tip on sorting the polarity. Never thought of that.

Axel - I'm trying not to gt hung upon it, but this has been a long time in coming; I've finally go hold of some trace recorder equipment that is the wrong side of $25K to do the job (and a Dodge Ram later this weekend - hence how I got it) - basically - this is a one shot attempt to get the data we need, and I don't want any unanswered questions or second guesses at the data points.

Rattly - Cheers on the sensor - I'll confirm using Evo's approach.


Ain't this forum GREAT! :) :) :) :)

Rattly,

Good to have another injection man on board.

Why not tell us about yourself and your projects in the Intro section.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Here is the answer guys...

http://www.turbominis.co.uk/forums/index.php?p=vt&tid=171104

I spend plenty of time writing technical docs for work, aswell as other stuff you do as a qualified engineer (reference another post - Here is your bird) so, all you're getting is a powerpoint LOL!

Apologies for the watermarks - but you know how it is... :)

http://www.jbperf.com/