5

I would like to measure and log the ignition advance timing angle of a vintage (70s) car. That is, mechanical distributor points (Kettering system), no electronic ignition, no RPM sensor.

For this, amongst other things, I need to monitor the signal that is sent from the ignition coil via the distributor points to spark plug #1, which determines the start of the ignition event for my purposes.

This signal is coming from the secondary side of the coil, and it's a periodic pulse signal in the order of kV. A single period looks like this:

enter image description here

Essentially, I want to convert the burn line region on the picture to a square signal of TTL level, which I can then read with a microcontroller.

This question and its answer explain a way how to do that for the primary side of the coil. As the secondary side mirrors the primary side, the concept of how to condition the input signal is the same: add low pass filtering, clamp the maximum and minimum voltage with diodes, buffer and convert to square wave with a Schmitt trigger.

However, the main difference here is that while the primary magnitudes are in the order of tenths of volts, the secondary ones are in the order of kilo Volts.

This brings a whole new set of considerations: obviously if not managed safely, it can pose a risk to the signal conditioning electronics, or more importantly, to the person handling them.

My experience is with lower voltage signals, and I'm assuming that there has to be more thought put into it than simply adding more attenuation (or optoisolation) to the signal going into the microcontroller.

This "Testing your ignition with an oscilloscope" document (coincidentally, the same vintage as the car) describes part of what I'm trying to do. On page 5, section Connecting the scope, a homemade pick up capacitor is coupled to the ignition wire. I'd like to think technology has advanced since the 70s and there is a more refined way of doing it. Furthermore, I do not often drive with tin foil wrapped around engine wires :)

How would I best go about:

  • Measuring the signal with a scope for testing/prototyping? Using a capacitive or inductive probe with built-in attenuation?
  • Permanently connecting, i.e. interfacing the signal to a microcontroller for monitoring purposes? I'm not asking for an actual circuit (*), rather a recommendation on the best way to physically connect the signal from the ignition cable to the stage that would condition it before going into the microcontroller, which would sit in a box in the engine compartment. Would that be the same answer as the oscilloscope question? I. e. capacitive/inductive probe? Or a regular resistive attenuator?

Thanks.

(*) although suggestions are welcome.

John M
  • 329
  • 2
  • 10
  • A quick search on line shows many ways to do this, from a special probe designed to hook into your scope, to various hacks that use things like wires wrapped round the spark-plug lead, hall effect sensors, capacitive couplers.. Take your pick. – Trevor_G Nov 27 '17 at 15:15
  • Thanks for the reply. The content of the question should show that I already did some research online before asking it, apologies if that wasn't clear. The reason I posted it was that generally StackExchange sites provide more informed/opinionated answers and discussion than simply doing a Google search. Also, any answers would help others facing the same question along the way! – John M Nov 27 '17 at 15:50
  • a) Can't you measure on the primary side? b) Can you measure the _current_ instead of the voltage? (Should, imho, represent start and end of the actual ignition even better (more 'digitally') than voltage.) c) Inductive coupling seems to be a common way. d) You can also try and 'listen' to the ('high' frequency, higher than 3/60 x alternator's RPM, kHz maybe) noise in the car's 12V system. – JimmyB Nov 27 '17 at 16:26
  • The ACS720 current sensor is spec'd up to 3600Vrms. For a hobby project, this may be enough of a margin. – JimmyB Nov 27 '17 at 16:34
  • @JimmyB, Thanks for the comments. a) The single spark #1 pulses are only available on the secondary side. On the primary side, they are intermixed with the pulses to the 3 remaining spark plugs. Otherwise, I'd happily measure on the primary (which I'm already intending to do to get the RPM signal, but that's a different subject). This is a 70s car, it's all pretty bare-bones :) b) Why would I want to measure the current instead and how would that be easier than attenuating and squaring the input voltage? Would you mind expanding on that and on the use of the ACS720? – John M Nov 27 '17 at 17:02
  • c) Ack d) Wouldn't I have to attenuate the signal in this case as well? If so, then it would it not still be easier to measure the higher amplitude, periodic and longer low frequency pulse from the picture? Listening to high frequencies might also lead to false positives. – John M Nov 27 '17 at 17:02
  • Imho, the main problem is in handling the high voltage of 3kV+. I would try to avoid that as much as possible. In contrast, connecting to 12...20V and stepping it dow to 5 or 3.3V is no problem at all. The ACS720 is a hall-effect current sensor which converts the current through it to an easily handled analog voltage between 0 and Vcc, and it seems to be semi-safe for voltages of about 3kV on the current measuring path. – JimmyB Nov 29 '17 at 11:43
  • Measuring on the primary side does, of course, require some way of synchronizing to the distributor or some other signal source to infer the cylinder which is currently igniting. When measuring the 12V system's voltage, some signal processing (analog high-pass + digital post-processing?) needs to be done but it's absolutely doable; you still have to somehow infer the active spark plug though. – JimmyB Nov 29 '17 at 11:47
  • As to the secondary voltage, you may want to consider an open circuit case (bad contact, disconnected spark plug/cable,...), in which you will likely see *much* higher peak voltages than those 2kV under normal operation. – JimmyB Nov 29 '17 at 11:49
  • Using current as an ignition indication has the advantage of a) being more of a 1/0 signal - current is only flowing when and as long as the actual spark is there, it's 0 at other times and b) for the same reason indicates precisely *when* the actual ignition inside the cylinder begins. - Btw, to measure timing *advance* you will need some reference of the crank shaft angle anyway, don't you? – JimmyB Nov 29 '17 at 11:59
  • @JimmyB: Thanks again for the replies. Just to be clear of why I need to measure on the secondary: I only have two signals available: a) *RPM*: from the primary, 4 pulses per distributor revolution, which correspond to the pulses sent to each one of the spark plugs; b) *Spark #1*: from the secondary, 1 pulse per distributor revolution, corresponding to the single pulse sent to spark #1. Signal *Spark #1* is the only way I've got to synch. There is no other way and I don't really need the *RPM* signal for timing purposes. – John M Nov 29 '17 at 12:17
  • If I had another way available that did not require handling HV, I would definitely use it. But I don't see much difference in measuring voltage vs. current in this particular case: once the signal is conditioned and brought to digital levels, they would both indicate that ignition is happening with a '1'. I'm not interested in measuring burn time, just the start point of the spark for timing purposes. The only advantage I can see is that the ACS720 could potentially be connected directly to the secondary, without the need of a capacitive/inductive probe. – John M Nov 29 '17 at 12:18
  • But then again, as you already mention, working peak voltages can be much higher than the depicted 2 kV (the picture is just for example's sake, not actual voltages from this car), which I don't know whether the ACS720 could actually survive. After a quick search, I haven't found any application note or anyone actually using the chip for that purpose. I'm currently leaning towards using a capacitive clamp and a capacitive divider as the first stage of attenuation to a voltage that can be handled by a Schmitt trigger. – John M Nov 29 '17 at 12:19
  • I'd use a coil around the secondary to measure the ignition current. You can size the coil so that you get a signal that suits your voltage levels. – Lelesquiz Jul 12 '18 at 16:00

1 Answers1

2
  • Measuring: use a capacitive probe with 1000:1 attenuation to measure the signal on the scope. By wrapping around the ignition cable, the clamp of the capacitive probe effectively forms a cylindrical capacitor in the order of a few pF. The attenuation block in the probe most probably completes the circuit with a capacitive divider, similar to what your linked document depicts.
  • Permanently connecting: you can use the same arrangement as the measurement and build yourself a capacitive clamp and attenuation circuit. Another option is to use inductive coupling. An example is the VW 113-919-105 sensor that was used to detect the spark on cylinder #1 for some vintage Volkswagen vehicles (Type 2 and Type 3). It was a ring in which the HT wire was inserted, which makes it probably more resistant to vibration than a bigger clamp plus attenuator block.

enter image description here

David Planella
  • 136
  • 4