What if you have an early K-Jet or L-Jet equipped engine and you would like to update to either a LH-Jetronic, Motronic, Haltec, or Electromotive TecGT engine management system?

Where will you put the crank reference sensor? Engines up thru 1984 M28/20 may have no bung to hold the sensor on the back of the block.

Yes, there are some nose-mounted crank sensor options available, and they do work, but that is not an optimal location for the crank signal because of crank twist (torsion).

Thats the very reason Porsche placed our crank reference sensor on the flywheel - to remove the issue of lag from crank twist from the engine management system. This gives greater accuracy to the ignition and injection events, and better throttle response.

We have developed an adapter plate that will locate and hold the crank reference sensor in the proper location on the early blocks, and it can be installed with the engine in the car.

We are manufacturing it from Class 12 Nylon GF (glass filled) and it is lightweight, will never corrode, and will withstand temps up to 350 deg F.
Add a 60 tooth timing ring to your flywheel, (or our aluminum flywheel with timing ring installed) and you are all set!

Here are a few pictures. At the moment, this is not for sale separately, but it is a part of our L-Jet Electromotive Tec GT kits.

Thank you for looking.

 

This will not work for L-jet cars as they have a boss there that has not been completed for this sensor.

And I humbly disagree with the fact the sensor needs to be on the rear of the cank. For if that were the case why is pretty much every vehicle made today equiped with it on the front of the engine? As is the hall sensors.

but for people converting from K-jet this very well could be the easiest way to install.

 

Yes, a lot of production automobiles place the crank reference sensor on the nose of the crank. If it isnt a high performance auto, why not? Its easier to service in that location. As I said in my post, it does work.

But just cause thats where they place it for ease of service and manufacturing on your daily driver, does not mean it is optimal.

Because of crankshaft twist, you get faster, more accurate reference readings with the sensor on the point of load - and that is the flywheel. Look around for videos of crank and camshaft twist - they are out there and they are fastenating how much twist there is.

They both work (nose mounted and flywheel mounted) but when you have a choice, flywheel mounted is more accurate.

In a related area - that is why Porsche went to the added expense and complexity to drive the 32v cams from the center.... thereby reducing camshaft twist as much as is possible. Do common production autos do that? No, its expensive and there is nothing to gain on your grocery-getter type car. But performance cars - YES.

 

Not all of them do.
I have five L-jet blocks, three of them have the bosses you speak of.
One of them has two additional bosses on either side that make adding a sensor to this location even more of a PIA.

It wouldn't surprise me if there was a fourth casting floating around with even more stuff in the way.

Either way, nice work Carl. I'll have to post up images of the adaptor Todd made for my car. It's always fun to see the same end result with two totally different paths taken to get there.

 

Hacker - you may be able to help me... I'm trying to figure out definatively which M28/xx numbers have no prvision for a sensor at all.

I know M28/01 thru M28/04 have no provision (1978,79) but once you get to 1980 to 1984 it gets very hit-or-miss.

Of your 5, which blocks have nothing?

Thank you for your time.

 

This is funny! Smokey Yunick made a similar incorrect statement when talking about camshaft twist, in his book.

If the really beefy forged crankshafts (with really high amounts of rod to main bearing cross section) were to bend (in the ultra low load 928 application) think about it for a whole second, before another "urban myth" gets started.

If the trigger is mounted in the front...the rear cylinders will be off by the increment of twist. If the trigger is mounted in the rear, the front cylinders will be off by the exact same increment of twist. Since the amount of twist would not vary, no matter which end the trigger is placed, putting it either in the front or the rear makes absolutely no difference...it just changes which cylinders are getting the incorrect signal.

Anyone need me to draw it?

Nice job on the mount, however.

 

I disagree. When the clutch is let out, the first end of the crank to see a change in RPM is the end closest to the flywheel, and thats where I want my crank reference sensor.

But whatever... glad you like the mount.

Whether you agree or disagree on the merits of nose-mounted or flywheel mounted sensors, I also felt this was the easiest way to get the highest resolution - the stock crank trigger ring is nice and big and yields greater resolution than the little nose trigger rings do.

Allthough the LH-Jetronic system cannot really use all that resolution, the Electromotive Tec GT system can and does, so the higher the resolution at the sensor, the greater our accuracy of ignition and fuel events.

 

There is a nice chart of crank twist (peak to peak) here:
http://www.atiracing.com/products/da...ting/index.htm

 

You actually think there is so much twist that the rear of the crank detects different rpms than the front of the crank....or it sees it sooner???

Seriously?

Regardless of what BS your electronic guy threw your way, and how far you swallowed the hook, here's the real issue with sensor location:

When you trigger off of the outer diameter of the flywheel and the clutch gets released, the flywheel flexes like crazy. This alters the sensor gap and this significantly changes the timing.

This is why Porsche moved the sensor in to read off of a much smaller circle on their Cup engines. This smaller trigger diameter can also be found easily at the front of the crank (damper diameter trigger).

The outer diameter of the flywheel is actually the absolute worst spot to have the trigger location. Now, how much will this effect the timing on a 928 engine? Not very much.

So although your rationalization for the sensor location completely missed the mark, your mount will work fine.

No More Urban Myths Needed...there are plenty out there already.

 

I hope you'll all indulge me, as this is one of those areas of
engineering that fascinates me, even though I have no intention of ever
running an engine where this would be of any consequence! My background as well
as my current occupation predisposes me to ponder these things on my daily commute!

Here is my take:

A camshaft has a reasonably even distribution of load across it's length
and relatively small torque load.(Relative to engine torque). Center
driving a cam as a way to average out drive twist is a comparatively
straightforward proposition.

A crankshaft has a relatively negligible load at the front and a
significant load at the flywheel. Under power, twist in the crankshaft
results in the rear most journal to be 'behind' the frontmost, since the
resistance to torque is almost completely at the rear. Torque from the
piston rods is attempting to rotate the crank faster, but the load at
the flywheel resists this, so the separation (in degrees of rotation)
between each cylinder actually increases slightly.

Mechanically, the rear cylinder will reach TDC slightly later than in
it's unloaded state, but the front cylinder, which has relatively little
change, due to it's distance from the loaded end, remains practically
unchanged. The rear will be slightly retarded.

Inter-cylinder timing is a relative measurement. This means that if we
measure the timing relative to the front cylinder, the rear will seem
slightly retarded, whereas if we measure relative to the rear, the front
will seem slightly advanced - by the same amount. The rear mounted
sensor is slightly more intuitive, since mechanically, the rear is
actually retarded relative to its unloaded state, but for the purposes
of driving ignition and injection, the phase angle difference is all
that matters, and even then, you would need to actually measure it and
adjust for it. A smart ECU could detect the gradual phase shift at the
rear sensor and actually respond to twist, while a front mounted sensor
has no way to sense this, as it's phase angle does not change. I doubt
they actually do that, but it is possible some of the more sophisticated
one's could.

As far as wheel resolution, you have to remember that the resolution of
a toothed wheel is strictly a function of the number of teeth, which
divides the 360 degrees of rotation into N steps - A 36 tooth wheel has
10 degrees per step, a 100 tooth wheel has 3.6 degrees per step. The
actual diameter does not determine resolution, although it does make
manufacturing tolerances less important as you go up in diameter. Once
the VR conditioner sends the signal to the ECU, the toothed wheel pulses
are simply reference points to tell the CPU where the crank is at a
given moment. The intermediate points are interpolated, so, for example,
the 36 tooth wheel's 10 degrees per step can be divided into 10 equally
spaced increments, giving a virtual 1 degree step, 100 increments gives
.1 degree/step, etc, up to the speed of the ECU's signal processing
capabilities. The 100 tooth wheel simply gives more reference points.

The actual resolution of the ECU timing is based on its processing
speed, with the reference pulses used to continually update its internal
timing as engine speed changes. A 100 tooth wheel could theoretically
detect a change in engine speed in 3.6 degrees, where a 36 tooth would
take 10 degrees, but any good sampling algorithm is going to look at
multiple pulses. At 6000 RPM, and engine turns 100 revolutions per
second, or 36,000 degrees per second, which translates to 27.7
microseconds per degree. A 100 tooth wheel updates every 100
microseconds, while the 36 tooth crawls along at a leisurely 277
microseconds per pulse.

Which leads me to the flywheel flex issue. The VR conditioning circuit
in an ECU is responsible for turning the output from the VR sensor into
a pulsed signal. The pulsed signal is not dependent on the amplitude of
the sensor, meaning that changes in sensor air gap due to flywheel flex
do not change the timing signal unless the teeth move so far that the
sensor cannot sense them at all, which would be pretty dramatic flex,
probably on the order of half the width of the teeth plus 50 or 60
thousandths, and then it would just miss a few reference pulses until
the flywheel came back across, then pick up a few, etc as the flywheel
oscillates. Missing a couple of reference pulses is not going to have a
significant effect on the ECU's timing unless it is very poorly coded.

To summarize,
front vs. rear mounting would only matter if you had an
ECU that could vary the phase angle of injector and/or spark firing of
SOME of the cylinders in response to it, where rear mounted would have
the advantage of possibly being able to measure the change, where front
mounted would have to use an empirically derived look-up table or coarse
calculation.

Diameter of the wheel doesn't determine resolution, number
of teeth does, but that is only the resolution of the input signal.

Flywheel flex could possibly result in a few missed reference pulses, which
a good signal processor would correct for in it's error diffusion algorithms.

I hope I haven't put everyone to sleep. At least I avoided going into
signal processing theory, nyquist rates and signal aliasing, which can really
be a snooze fest!


-don

 

Sounds like a lot of harsh personal opinion Greg, isn't that how myth's start? How does the flywheel flex like crazy and how much is crazy? Speaking from an angular point of view probably a lot less than the crankshaft, unless I just missunderstood what you were saying. The crank does flex, and it would see a slightly different instantaneous rpm at the front than the rear followed by the front slowing back to the rear speed (more or less), with the sensor seeing then a slower rpm especially after one of the front cylinders fires. The crank only twists a lot when a front cylinder fires because the rest of the crank is tied to the drivetrain, when a rear cylinder fires the cranks polar moment and the general resistance from the foreward cylinders are the only forces causing it to flex since it ties to nothing except cams and accessories at the foreward end, so it doesnt flex much. There is a method behind selecting firing order that is based around crank flex to reduce stress which basically tries to keep too many foreward cylinders from firing consecutivly. With computer modeling they have determined even further how to reduce crank angle variance front to rear by altering the resonance frequency of the crank (and the drivetrain) which is based largly on its stiffness, but also cylinder spacing etc so as not to have two waves amplify each other.

The flywheel seems like the most logical point to mount the sensor though since its sort of a mass damper to oscillations.

However, I do see merit to the statment you made in regards to whether or not you mount the sensor at the front or the rear it is going to be off in reference to the cylinders at the other end. Ideally you would have a sensor at both and calculate from that where each cylinder is likely to be.

 

Nice product Carl. I could have used it 6 mo ago instead of stacking washes to get the gap correct.
Keep the stuff coming!

 

+1

If I crapped on his posts liked he craps on mine, his head would explode.

 

Don,

I liked your explanation, and I am not an EE. The way it was explained to me the larger diameter wheel with the same number of teeth as a smaller wheel still has a difference: the size of the teeth and the space between the teeth. It will be larger. Perhaps that is offset to a nul gain by the fact is the teeth are moving at a faster rate of speed?

It was explained to me by Electromotive and I am probably not relaying the information quite right.

 

Cool! I figured no-one would get past the second paragraph!
If you think about it, the diameter won't affect the resolution of the signal, but as I mentioned, it does affect accuracy.
If you were to make 2 wheels of 100 teeth to the same manufacturing tolerance, say - a variation in tooth position of +- .02 inches center to center, the larger diameter wheel would be more accurate. This is because, while they both have 3.6 steps per degree, the ratio of the tolerance to the circumference of the wheel is different.
A wheel has a circumference of Pi * Diameter, so an 11.459155902616 wheel has a circumference of 36 inches, or .1 inches per degree. 20 thousandths is 1/5, or 20%, of .1, so that wheel made to those specs would have +- 20% error in tooth location.
Double the size of the wheel and the circumference is now 72 inches, or .2 inches per degree. That's +- 10%.
The size of the teeth and the spaces between have an affect on how well the sensor 'sees' the teeth, since it's easier to detect a larger mass, even though the surface of the teeth is moving past the sensor faster. Essentially, the pulse amplitude is going to be larger, making it easier on the VR conditioning circuit.

In the interest of full disclosure, let me point out that even though I do have a 36-1 wheel in the garage (thanks Colin!) I have never actually installed one on my 928. Yet. But, I have put a lot of different kinds of sensors on a lot of different machines in my career, from 68,000 pound rockets, to toothpaste tube manufacturing equipment to printing presses!

-don