7
\$\begingroup\$

I'm trying to do some diagnostics on a classic car (aircooled VW). In particular, I'd like to monitor a signal that flags the TDC (Top Dead Center) on every engine revolution.

The signal is available via a plug with 3 contacts, one of which is grounded (it appears to be the shield for the two other wires). This is the schematic representation of the plug from the Owner's Manual wiring diagrams:

enter image description here

According to this source, the signal is generated by a variable reluctance pickup sensor. Quoting from it:

Beginning in '74, a sensor was added behind the flywheel to detect TDC. As far as I know, this sensor was a variable reluctance type of pickup. The flywheel had a steel pin on its back side which swept past the pickup face to create the pulse.

The combination of the TDC sensor and the #1 plug wire pickup allowed the computer to accurately determine initial timing setting and even to observe timing advance with speed change.

In short, it should create some sort of pulse for every revolution of the engine.

For more context, this other source has got more detailed information around the sensor. There it's called an inductive pickup:

It's an inductive pickup that produces a signal when each of two dowels attached to the back of the flywheel pass the sensor and it's used for measuring the engine timing electronically.

As I've not come across such inductive pickup sensors before, I'm not quite certain what type of output is to be expected from the sensor plug. This article seems to indicate it should be a voltage (1 to 2 V), but without previous experience with these, I cannot be certain.

Update: After some more research, I've found out about these ICs as possible candidates to do the signal conditioning:

  • MAX9926: Variable Reluctance Sensor Interfaces with Differential Input and Adaptive Peak Threshold
  • LM1815: Adaptive Variable Reluctance Sensor Amplifier
  • NCV1124: Dual Variable−Reluctance Sensor Interface IC. With an application note - AND8149/D: Understanding and Using the NCV1124 VR Sensor Interface

Of the three, the MAX9926 seems to be the only one with differential input, which seems to be what my sensor provides. The other two seem to expect a ground-referenced input. Regardless of the functionality, unfortunately, the MAX9926 SMD packages seem too small to do manual soldering on a prototype

These ICs seem to be based on detection of zero crossing of the output signal. Information on such sensors seems to be mostly for modern cars, where the sensor is facing a toothed gear with a missing tooth at the TDC point. The output seems to be an AC waveform for these. On my sensor, there is only a dowel (tooth), so I'm guessing the output would not be AC, but rather a positive pulse-like type of waveform.

If I were to feed it as an input to a microcontroller, what sort of signal conditioning should be applied to it to be able to read it? Or could I use one of the signal conditioning ICs?

\$\endgroup\$
1
  • \$\begingroup\$ I would expect leading edge is the reference point, whatever polarity it is. Since EMF changes with RPM, and hysteresis is necessary for rejecting noise, the thresholds need to adapt with rising V and rising RPM. \$\endgroup\$
    – D.A.S.
    Commented Dec 1, 2017 at 1:18

2 Answers 2

3
\$\begingroup\$

I've worked with these before and I get away with using pretty vanilla op-amps or comparators as the voltage from the sensor is rather large. Here's a sample I just made with a Toyota 4A-GE 20V ("blacktop") distributor:

enter image description here

Top trace is the once-per-cam-rotation signal (G1 or G2), bottom trace is 24-per-cam-rotation signal (NE), rotation was done by hand at a speed i measured roughly to be 100 RPM (camshaft). You can see that the voltage on the faster NE signal is 1V peak, and the slower G1 signal musters at least half a volt. these voltages with scale roughly linearly with engine speed - you may have to build a clamp circuit to prevent overvoltage at your op amp.

Here's a cct I never deployed on an automobile, but it did serve me well on the bench:

VR conditioner

You can add filtering and offset to this as you see fit, either in the analog or digital domains. Note that even if your zero crossing detection is a little delayed you can correct for that in software - engine management is all about guessing the future anyways, adding a degree here or there is just a bi more math.

\$\endgroup\$
1
\$\begingroup\$

try an instrumentation amplifier. But most critical is to play with the impedance the sensor sees - this will affect both signal amplitude and also noise, and you need to maxmimise s/n. The amp is as much to provide impedance matching and conversion as to provide gain, and you might end up amplifying the output a bit more (with a standard op amp) and also squaring it through something like schmitt trigger - basically you have to look at it and figure from what you see with a scope. But start with an instrumentation amp. Looks like teh signal is balanced, which is good.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.