Variable reluctance pickup detector design

[adinsen]

I’m hoping to pick some analogue brains of people more intelligent and wise than me !

I’m working on a detector circuit for a crank crank sensor fitted on my 1982 Talbot Matra Murena 2.2. The sensor looks like a modern crank sensor but is just a simple coil inductor without the usual permanent magnet inside. Variable reluctance pickups in cars, electric guitars, microphones, etc have permanent magnets incorporated. I don’t have a spare that I can take apart, but this seems to be just a simple ferrite cored inductor. The flywheel, which it is pointed towards, does not have permanent magnets either, only indents which the sensor are supposed to detect.

The sensors were fitted on Talbot engines in the 1980’s to facilitate diagnostics. At least that’s what’s said in the original workshop manuals from the time. Nobody I’ve talked to in our community has seen a diagnostic tool that can use these sensors, though. I’m planning to change that!

My design is based on the following: To produce a permanent magnetic field I'm biasing the inductor with a DC current through two resistors running from a stabilized 9V supply. The balanced AC signal is fed to a differential amplifier and then to a schmitt-trigger detector. It works in my lab detectng even small indents in iron.

Before proceeding with actual testing on engines, I'm appreciate getting some feeback on my design idea, both the overall design and details like:

• Input protection. People will likely by mistake connect the inputs to all sorts of wrong wires, reverse polarity etc. I’ve tried to protect inputs with zeners and transient diodes.
• Noise immunity – there’s probably not a lot that can be done as the pickup output signal is in the order of 20 mV and it will be very sensitive to magnetic fields. Ignition coil and generator will probably create “fake” signals. I plan to do various kinds of software filtering.
  
• The choice of OPAMP for the differential amplifier stage is probably uncritical. The TLC271 is more or less just a random choice, and I can get them cheap. Input is welcome!
• The comparator in the detector is more critical. I have not been successful with the LM393, it latches up and probably oscillates in transition at this level of sensitivity. The TLC372 is better, but it’s also not stable. I’ve previously used the MAX912 for a similarly designed detector circuit with similar sensititivity, and while its 100 MHz bandwidth is quite over the top for this application, I plan to use it for my next iteration. (I even built a vero board version with it that worked.) MAX912 is designed to be stable in the linear region, this probably helps. I have a source with plenty in stock for a reasonable price, and I’m not going to build many of this circuit anyway, but I'm open to input!

So what do you guys have to say? I’ll be grateful for any kind of input! 

[moffy]

To help with CMRR you could connect a capacitor equal in value to C5 across R9.

[adinsen]

I got my PCB with the MAX912 design a few days ago. Assembling it, it worked more or less out of the box, but had an annoying tendency to oscillate around 1.6 MHz. Also, the oscillation not just included the comparator, but also the differential amplifier.

My choice of a 100nF plate ceramic capacitor for power supply decoupling did not work. The MAX912 is "low power", but it's drawing a lot of power when it's oscillating. A cylindrical ceramic capacitor placed under the DIL package directly between the supply pins cleaned the power supply rail, but didn't elimate the osciallations.

My previous design with the MAX912 had a discrete emitter-follower input buffer. Apparantly the TLC271 CMOS OPAMP output impedance is too high for it to drive the 100 input resistance of my detector stage. Replacing the TLC271 with a TL071 fixed the problem.

Revised schematic is attached.

Next job is testing the circuit in the noisy environment in the car.

Speaking of fast comparators, I'd like to point to Jim Williams well written application note AN72 "Guidance for Putting Civilized Speed to Work" is. I'm no expert in analogue design, I feel that if one wants to work with comparators, it's a must-read of practical wisdom: https://www.analog.com/media/en/technical-documentation/application-notes/an72f.pdf