How a Hall ignition works?

[74Sprint]

It's possible to generate each cylinder's spark at the right time by simply knowing TDC #1 instantaneously (almost), and given that provides accurate crankshaft speed, the appropriate time interval, or number of pulses counted, determines subsequent spark times. Gets a bit hairy.

Actually you don't want to measure TDC of #1 because that would be to late. You would never be able to have advanced ignition timing, it would be like try to predict the future. If you did measure TDC then your ignition would be 1 or 2 revolutions behind, 1 for wasted spark. You have to put your crank sensor at least 5 derees more advanced than your maximum timing advance to give the cpu time to do the calculations. ECM's, ECU's, or elctronic ignition units can only delay the spark firing they can't advance it. Less delay = more ignition advance seen, more delay = less ignition advance seen. On one engine I setup a ignition by sensing when the intake rocker closed the #1 intake valve and then it used the crank sensor to fire the cylinders at the correct advance.

Jack,
You still need more than one pulse per revolution to measure the rpm for the purposes of controlling when to fire each cylinder.

My context was referring to using this technology in a model engine ignition application.While you could do it with a single sensor as you describe, it would require a relative fast microcontroller and some sophisticated software to do the pulse width measurement and comparison.

Tim

We use a single pulse per rev. and use the pulse width to calculate the rpm, it's actually quite accurate. We use a PIC12F1840 at 16mhz, it's not that fast or expensive at $1.50 each for a 8 bit cpu. The code isn't that complicated either.

Ray

 

[jack.39]

Jack,

I understand all that. You still need more than one pulse per revolution to measure the rpm for the purposes of controlling when to fire each cylinder. In industrial vibration analysis we often use a single pulse per revolution to establish rpm, but that is for the primary purpose of establishing an index reference for doing phase related analysis, not for control. When doing things like tortional vibration analysis, we use on the order of 60-100 pulses or more per revolution to get the resolution we need.

My context was referring to using this technology in a model engine ignition application.While you could do it with a single sensor as you describe, it would require a relative fast microcontroller and some sophisticated software to do the pulse width measurement and comparison.

By contrast, with two sensors I could do the majority of the logic in a relatively simple PGA - spark advance, etc. could be done with very little additional circuitry.

Tim

Tim, perhaps it is I, and not my neighbor, who is full of sh-t! jack

 

[Noitoen]

Because of the speed involved, most ecu's don't have the potential to calculate the different values to control the engine so, they measure the different sensors and use a lookup table to have faster response. One engine I removed from a car to try to replace my traditional carburettor/ distributor engine on my old Samurai 1000cc, was the Fiesta's MK2 engine. From what I've read at the time, this engine was updated from the previous model by removing the carburettor and distributor and mounted an ECU to control the single point injection and a pair of double lead coils. The timing was done by a magnetic pickup reading a gear on the main shaft's pulley. The gear was like any other normal gear except for a uncut tooth or two located at, I think TDC. I abandoned the project due to difficulties in the adaptation of my original gearbox and legal issues with our transportation department.

 

[picks27t]

I just got a cdi from S/S there pick up is in a round jacket should this be removed?


Sent from my iPhone using Model Engines

 

[74Sprint]

When you say round jacket do you mean the mounting ring or the Hall Effect pickup? I guess I should have asked, which ignition model did you purchase? We can go from there.

Ray

I just got a cdi from S/S there pick up is in a round jacket should this be removed?


Sent from my iPhone using Model Engines

 

[74Sprint]

Because of the speed involved, most ecu's don't have the potential to calculate the different values to control the engine so, they measure the different sensors and use a lookup table to have faster response.

Actually the speed of the ECU might surprise you. When you compare mechanical speed to electronic speed the mechanical speed is very slow. You don't need a super fast CPU because you don't need to check certain sensors that often. For example the engine water temperature can be checked once every 10 seconds instead of 100 times a second. This frees up the CPU to check more important things more often. And yes we do use lookup tables stored in memory.

The timing was done by a magnetic pickup reading a gear on the main shaft's pulley. The gear was like any other normal gear except for a uncut tooth or two located at, I think TDC. I abandoned the project due to difficulties in the adaptation of my original gearbox and legal issues with our transportation department.

Yup this is done quite often like the 58-2 Ford or 35-1 GM, where the Ford should have 60 teeth it has 2 missing teeth to indiacate TDC but the pickup is not mounted at TDC it's in a advanced position I can't recall right now the degrees advance of the position, the GM is the same. So with the cam sensor and crank sensor you can calculate your ignition and injection timing.

Ray

 

[TimTaylor]

We use a single pulse per rev. and use the pulse width to calculate the rpm, it's actually quite accurate. We use a PIC12F1840 at 16mhz, it's not that fast or expensive at $1.50 each for a 8 bit cpu. The code isn't that complicated either.

Ray

I've done a lot of embedded real time controller stuff over the years, but it has been mostly Motorola 68xx/68HCxx and Intel 87xx processors - have never played with the Microchip PIC processors.

Out of curiosity, is your code interrupt driven?

 

[74Sprint]

I've done a lot of embedded real time controller stuff over the years, but it has been mostly Motorola 68xx/68HCxx and Intel 87xx processors - have never played with the Microchip PIC processors.

Out of curiosity, is your code interrupt driven?

Yup it's interrupt driven.
It's not that bad of a setup but, it is designed for R/C airplane single cylinder 2 strokes and can be used on 1 cylinder 4 strokes, must use wasted spark. So this ignition would be good for most of the engines made on this web site. Excellent for a hit-n-miss because it will work with 4.8v Ni-Cads. It has a rev limiter that ******* the timing which of course drops the power a lot. With the way it is setup and a added cam sensor and the speed of the MCU it would be only good for at best @6500 rpm for a 8 cylinder. Would most likely need a faster MCU for an 8 cylinder.

You say you did Motorola 68xx/68HCxx, have you done any work with MegaSquirt?

Ray

 

[TimTaylor]

You say you did Motorola 68xx/68HCxx, have you done any work with MegaSquirt?
Ray

Nope, I have not. The majority of applications I worked on involved either embedded multi-variable process control or safety interlock systems. Lots of real world interface with things like UV and IR sensors, pressure, temperature and various other analog sensors, etc.

Tim

 

[jack.39]

"Actually you don't want to measure TDC of #1 because that would be to late. You would never be able to have advanced ignition timing, it would be like try to predict the future."

Actually, Ford's EEC-IV, using the old mechanically-driven distributor, did not "measure" crankshaft location rotationally, but was simply adjusted manually to obtain 10 degrees BTDC firing of cylinder #1, with electronic connection of the computer's ability to advance/****** spark manually disconnected. Thus, each time the "identified" tooth was sensed, the PCM knew 10` BTDC was present, and subsequent firing advanced beyond that point. Thus, the spark could NEVER occur at less than 10`, under any conditions.

I think! jack.39

 

[74Sprint]

There are several generations of ignitions out there that we can duplicate here but, I like the KISS system. Having said that I should say what is a simple system to one person is a complicated system to another. I still love the first generation GM HEI and hate points. You disconnected the vacuum advance hose, set the base timing, hook the hose back up and away you go. It also had a mechanical advance that can be tailored, even the vacuum advance could be tailored. Once GM worked out the bugs it was rock solid. On Chrysler's first TBI you would pull the water temp. sensor wire off and set the timing and Ford had their PIP wire.

In drag racing some cars have locked in timing. They set the timing at say 26 degrees and lock that in. No advance of any kind and is a killer on starter motors.

I know on the RC forums that a lot of people have a hard time to understand that an electronic ignition can only delay the output spark and not advance it. Less delay = more ignition timing and more delay = less ignition timing, but the base timing needs to be set first. On this forum, I think the best ignition is a Hall Effect. Problem with after market ones like RCexl that say they have a programmed advance is what are they programmed for. You spend your money and you take your chances is not my idea of proper. This is why I'm helping to make a programmable ignition that I can program.

Ray

Ray

 

[jack.39]

Well stated! jack