Hi everybody, I'm new to the site and to electronics, but I'm trying very hard to learn. I have a classic Alfa Romeo and I'd like to remove the distributor points and replace them with a solid state solution. I'd like to use this opportunity to learn something better about electronics and it's a fun and, I think, fairly simple project. I would really love some help if someone can review what I have and tell me where I've headed for the rocks.

The general idea is to replace the points with a Hall Effect cam position sensor that can read the cam lobes on the existing distributor shaft that drive the points and generate a signal to fire the ignition coil. The coil itself will be controlled by a Bosch ignition module for which I simply must provide the ground at the time it should fire the coil.

So as far as I can tell if I use the Hall Effect Sensor, some components to protect it from the automotive environment, and a good high voltage, high current transistor, I pretty much have my circuit? But I'm not as sure about the details and I'm worrying that I'm missing something crucial that it will result in frying something.

My plans are to use this Hall Effect cam position sensor which looks like it does exactly what I want and it can handle automotive temperatures and stresses:

http://melexis-microcontrollers.com/Asset.aspx?nID=4759

I will want to upgrade this system to use a higher voltage capacitor discharge ignition setup eventually so I picked a fast switching high voltage, high current transistor that looks like it fits the job well:

http://www.fairchildsemi.com/ds/FJ/FJD5304D.pdf

What I can't tell very well from the spec sheets, because I don't know what I'm doing, is if the output from the Hall Effect sensor will be high enough to trigger the transistor. Do I need another transistor with a lower base trigger level to in turn operate the FJD5304?

Info in the Bosch ignition module is about half way down here:

http://megasquirt.sourceforge.net/extra/setup-msns.html

So I was going to implement the recommended automotive circuit from the MLX90217 sheet like in the attachment, with the collector from the transistor going to the Bosch unit pin 6 and the emitter to ground.

Does that make sense? Am I missing something.

Help is very much appreciated.

Cheers,
Karl

 

Oh this is a car, so input voltage is 12V but not very well regulated. The 90217 I think has a regulator included though.

 

What are you going to do about matching the dwell angle of the old points system? I would think you would need a comparator with adjustable threshold between the Hall effect sensor and the transistor.

 

Well, it appears that the Hall-effect sensor puts out a square wave.

However - the documentation on the Hall-effect sensor is really not very complete on various maladies that the distributor may be suffering from, and substituting it for points certainly won't correct the problems.

It DOES mention worn cam lobes, which will certainly cause problems. However, no mention is made of worn distributor shafts and/or shaft bearings, which will be even worse, since they will cause seemingly random changes in timing as the cam lobe wanders around. Worn breaker plate bearings will cause similar difficulties. Corroded/stretched mechanical advance springs and worn bearing points are another can of worms, not to mention the 2nd set of bearings that can get frozen up or worn. Vacuum advance modules have a useful life in casual driving of perhaps 7 or 8 years; maybe more in a 4-banger - but their performance will be degraded as the diapraghm hardens, cracks and begins to leak.

Were I going to attempt something like this, I believe I would utilize a Hall-effect sensor on the flywheel ring gear, with an additional Hall-effect sensor on another (added) point on the flywheel to detect TDC. This would give a good deal of precision in detecting the exact RPM and position of the engines' crankshaft as an input to a microcontroller, which could in turn combine those readings along with manifold pressure, temperature and throttle position to determine when to fire the plugs.

Instead of a transistor for firing the coil, I would use a MOSFET. They have some incredible MOSFETS out nowadays, specifically for automotive use like the IRF3703. The only real limiting factor with that MOSFET is the TO-220 case it lives in; otherwise it could sink over 200 Amperes. (TO-220 is limited to around 75 amps before the leads will melt off.) Other candidates might be in the IRF14xx series. For that matter, even an IRF150 will do.

 

I agree with almost everything in SgtWookie's design, with the following comments/changes:

1) That Hall sensor is a gear tooth sensor. One use is shown in the animation with the magnet and sensor on the same side. As such, it will detect every gear tooth, not just the TDC or any other. Of course, the magnet could be put in the flywheel, but if one does that, then almost any Hall sensor as used on model engines (non-latching, unipolar) should work.

2) I would use only one sensor. TDC is a good place, but actually anywhere will work. If you put it at the max BTDC spot or a little before, you will simplify things as that would be the earliest firing and everything else would have a delay. Two sensors require a program to determine which is which, particularly at start-up. There are several solutions, such as putting them at different radial distances or timing of the space, but why complicate matters.

3) The use of a microcontroller as SgtWookie suggested is almost mandatory. It can be done without it, but getting the advance done is really hard that way. With the MCU, getting advance is easy and there is software already out there. The hardest part is of course timing, as each plug fires only once every other revolution. Some people solve that problem simply by firing the plug every revolution--so what if it fires on the exhaust stroke.

John

 

Wow, thanks for all the input guys. I realize I'm not talking about a very sophisticated system and that there are lots of things I could do to improve it. I want to build a system that uses a microcontroller to manage advance, but I'm learning and I don't know how I could do all that right now. I thought for starters I'd do the very simplest thing that could work and then improve it over time until it did everything I wanted. I'm a programmer and that's my philosophy on software.

What are you going to do about matching the dwell angle of the old points system? I would think you would need a comparator with adjustable threshold between the Hall effect sensor and the transistor.

That was one reason I was going to use that Bosch module since it handles the coil dwell for me. Ideally my circuit would do that but I was trying to keep it simple to start with. Plus since I'd like to move to a capacitor discharge system that will make the dwell time less relevant eventually.

That Hall sensor is a gear tooth sensor.

Oh, I read the spec sheet for the Hall sensor and it says it's a cam position sensor, too. It even shows a drawing of it applied in that way. Maybe I don't understand it properly.

Instead of a transistor for firing the coil, I would use a MOSFET. They have some incredible MOSFETS out nowadays, specifically for automotive use like the IRF3703.

Ah, okay! I guess I don't understand why a MOSFET is better. I would love to be educated about why. So pardon my not understanding, but it seems a MOSFET works like a valve operated by the voltage on the gate? So do I need to use a transistor to activate it fully from the hall sensor? Also, can that IRF3703 only handle 30V? That's what I think it says, which won't be enough for when I add capacitor discharge, where I will need to dump at least 200V into the coil.

Thanks for all the advice and for helping educate me!!

Karl

 

So I think the advantage of a MOSFET is the switching time? It looks like it's in nanoseconds. Also, would this one be a good choice? Seems like it:

STP2NK60Z http://www.st.com/stonline/books/pdf/docs/10066.pdf

Thanks!
Karl

 

Oh, I read the spec sheet for the Hall sensor and it says it's a cam position sensor, too. It even shows a drawing of it applied in that way. Maybe I don't understand it properly.

As it is configured using the magnet and Hall sensor on the same side, it looks at the flux as the gear tooth passes. I suspect it is a pretty distinct change -- the tooth is either there or it isn't.

With a cam lobe, the change is not so abrupt. I suspect given system noise and environmental factors, that the exact detection point on the cam will vary. That is, it will always detect when the lobe is nearest, but there might be an unacceptable amount of variation ("dwell") in exactly when the signal changes.

A small magnet (1/8 " diameter) set into the flywheel will give reproducible timings within a degree. For example, a 10" flywheel has a circumference of 31." The Hall sensor will fire on the edge, say within a 1/16" or less. That is less than 1° degree.

John

 

A small magnet (1/8 " diameter) set into the flywheel will give reproducible timings within a degree. For example, a 10" flywheel has a circumference of 31." The Hall sensor will fire on the edge, say within a 1/16" or less. That is less than 1° degree.

Okay, well to put anything on the flywheel I'd have to pull the motor, and to put it on the front crank pulley makes working on it a huge pain, plus that whole part of the motor gets covered in oil so I'd really need an expensive automotive hall sensor. Do you think my idea just won't work, or that it wouldn't be as accurate? Right now it's running points, so I don't think it's all that perfectly accurate to begin with.

Here's a product that seems to work like my idea:
http://www.cranecams.com/index.php?show=browseParts&lvl=5&prt=1973

But maybe they've done something much more sophisticated to fix the problems you mention. It clearly does something more because it has a rev limiter.

Are there problems with my circuit? Thats what I was really hoping to get help with. As long as the electronics _work_ I can improve the design or location of the hall setup. Do I need something more to connect the hall sensor to the MOSFET or transistor? How do I read the spec sheet to tell what the trigger voltage is for the transistor/MOSFET?

BTW: John where are you in Ohio? I was raised in Cleveland and lived in Columbus for 10 years.

Thanks!
Karl

 

One advantage of a MOSFET is it's extremely low on-resistance (RdsON) which means that very little power is consumed within the device itself; thus it will generate much less heat than a typical transistor. Another is it's extremely high gate impedance; they don't take much current to turn on and off. However, if you're going to be using HV capacitive discharge, you may need to go the IGBT (Insulated Gate Bipolar Transistor) route.

As far as mounting magnets on the flywheel - I don't believe that the way it's normally done; instead, the magnet is on one side of the Hall-effect sensor, and ferrous materials passing by the other side are detected as either field strength or presence/absence (depending upon the individual Hall-effect sensor)

The cam lobes on a 4-cylinder engine's distributor shaft are very wide, and the "ramp" of the cam is relatively gentle. When they started using electronic ignition in autos, they used a "reluctor wheel" which replaced the point cam; very pointed tips/steep "ramps" on the reluctors which passed by a coil fed by an opamp that sensed the changes in the coil's current due to the abrupt change in the magnetic field. However, that scheme kept the timing dependent on the traditional vacuum/mechanical advance units, and still had the variable timing due to worn bearings, etc. ad nauseum.

In a project like this, the distributor would still be useful for routing the spark to the cylinder desired. The spark needs to occur every 180° of crankshaft travel (adjusted for timing, of course) and the distributor would be reduced to it's original function of years ago - simply distributing the spark to the cylinder that needs it.

As far as the precision of the timing - the ring gear of the flywheel is the perfect place to get it, along with extrapolating the instantaneous engine RPM on a sub-degree level of accuracy. However, SOME sort of "starting place" must be defined, so that the microcontroller knows when to re-set the count of gear teeth. I suggest that even a very small steel screw would be enough to trigger a magnet-backed 2nd Hall-effect sensor, connected to a high-priority interrupt line on the MCU, resetting the tooth count.

 

Anyone who has used a strobe to time points knows that the timing varies a little, so you really don't need that 1° precision. What I don't have any feeling for is how accurate the cam lobe timing will be with the gear tooth sensor. Is your cam belt or chain driven? Is putting the magnet on the cam drive a possibility?

I have seen things like you show in the link used on small engines, and they work well. But, the question was raised about whether your distributor is in adequate shape. It is a simple solution, though.

Finally, I live a few miles Southwest of Cleveland. John

 

One advantage of a MOSFET is it's extremely low on-resistance (RdsON) which means that very little power is consumed within the device itself; thus it will generate much less heat than a typical transistor. Another is it's extremely high gate impedance; they don't take much current to turn on and off. However, if you're going to be using HV capacitive discharge, you may need to go the IGBT (Insulated Gate Bipolar Transistor) route.

Ah, OK. Well I think that MOSFET I posted above might match the necessary voltage and amperage. Low heat = good.

As for the ramp angle on the points cam, you are right it is fairly shallow. Maybe no good for this, then. I guess I can build this for a few dollars and if it doesn't work, then I do something else.

Anyone who has used a strobe to time points knows that the timing varies a little, so you really don't need that 1° precision. What I don't have any feeling for is how accurate the cam lobe timing will be with the gear tooth sensor. Is your cam belt or chain driven? Is putting the magnet on the cam drive a possibility?

Well it's chain driven, with the two cams completely enclosed in the valve cover so putting a sensor there would be hard. I guess the front crank pulley is really the best solution other than the distributor. The distributor seemed to be a simple place to put it, but since it's gear driven off the oil pump there is definitely some slop in there.

I have seen things like you show in the link used on small engines, and they work well. But, the question was raised about whether your distributor is in adequate shape. It is a simple solution, though.

Finally, I live a few miles Southwest of Cleveland. John

True, my distributor is probably a bit worn. Not too bad in the shaft play (that I can notice by hand). It's a nice old Magnetti Marelli distributor, though, with an advantageous advance curve for this motor. If I replaced it, the only new parts solution is a Bosch unit which doesn't look half as nice and has a less useful advance curve.

There is no vacuum advance/retard on these old motors so I don't have to worry about it being dead.

Am I gathering that given the potentially less than optimal physical condition of the distributor, and if the cam lobe slope is OK for the sensor to pick up, that you guys think the general circuit will work?

Cheers,
Karl

 

Anyone who has used a strobe to time points knows that the timing varies a little, so you really don't need that 1° precision.

Actually, when the engine was brand new, the timing didn't wander very much at all. However, as the various components wear (camshaft sprockets/chain, distributor drive gear, and the aformentioned multiple maladies aging distributors suffer from) the timing becomes more "sloppy". It would not have been possible for the auto manufacturers to meet increasingly stringent emissions and economy standards using such components. I remember having to check/set points and timing every 6,000 miles up to the mid-70's.

What I don't have any feeling for is how accurate the cam lobe timing will be with the gear tooth sensor. Is your cam belt or chain driven? Is putting the magnet on the cam drive a possibility?

Seems to me that they used a chain drive, but I haven't looked at one in a looong time.

I don't even remember where the starter is mounted. However, wherever the starter is would be a good location to look for a mounting point for a Hall-effect sensor. At least you should be able to see or feel the ring gear teeth with the starter out.

 

Front crank pulley - you need to ensure that it is a solid pulley and not a harmonic damper type. If it is a harmonic damper, the flexing of the damper will cause wide variations in your ignition timing at high engine speeds, depending upon engine loading and harmonics (of course).

 

I remember having to check/set points and timing every 6,000 miles up to the mid-70's.

Never tried timing a new engine. Last one I did was a 65 VW Beetle in about 1972. John

 

Ok I think my circuit will not work yet after consideration. The MLX90217 Hall sensor puts out a square wave. I want to have a momentary trigger to ground for the Bosch unit. I don't think it's good to have it remain grounded during the whole length of the high signal that comes out of the Hall sensor. That will be 45 degrees of crank rotation which is pretty long, so maybe it will get too hot, or prevent it from managing the coil dwell correctly. So I want to shorten the pulse that comes out of the Hall sensor.

I think from what I read that I can do that with a capacitor in series with the Hall sensor and a resistor to +12V on the output of the capacitor, then trigger my transistor from there. But I don't know how to figure out what values to use. I don't know how long (time) the ground needs to be from the Bosch unit, but lets say not more than a few ms is probably ok. Say I wanted the transistor triggered for 2ms. How would I figure that out? Or can I only modify the output signal by a percentage, and not a fixed amount? That would be fine, but I'm lost about how to figure that out. Help!

Karl

Diagram of proposed addition:

                               o +12V
                               |
                               /
                               \
                               /
                               \
                               /
                               \
                               |
              o-------||-------0---------> Transistor/MOSFET
From MLX90217

 

Another method of triggering could be a small blade attached to the points cam,that shuts of the light coming from an led,to a phototransister.
regards bodger

 

jpanhalt suggests in one post that there is existing software for this type of application for use with a MCU. Where might this software be?

also, side note, firing on the exhaust stroke improves emissions and thus is a positive...

thanks

 

jpanhalt suggests in one post that there is existing software for this type of application for use with a MCU. Where might this software be?

Here are the links I was probably thinking about:

http://www.rcuniverse.com/forum/m_4344316/anchors_7554669/mpage_13/key_/anchor/tm.htm#7554669

http://www.rcuniverse.com/forum/m_3422690/mpage_9/key_/tm.htm

Both are titled something to the effect of "open source CDI ignition." Bigboat is the main designer, apparently. They use a PIC. There are probably other links out there in the modeler's world. Twin-cylinder engines (opposed boxer) on ignition are fairly common.

The first link is the "new" thread. The actual design and source code are given in the first page or two.

John