A new ignition circuit

[jgedde]

EDIT: This circuit has been revised and improved. See post #11.

For my Farm Boy build, I wanted to wind my own ignition coil and build my own ignition circuit.

I searched around for suitable circuits, but found virtually all unacceptable. They usually had one or more of the following faults:

1) Excessive power consumption.
2) Just plain poor design.
3) Would overheat the driver transistor and/or the coil if the engine was not running with the crank in a position where the ignition was triggered.
4) Consumed power to keep the coil turned off.

So, as an Electronics Engineer by trade, I designed my own. I haven't yet built this, but I have run a PSPICE simulation and it works.

My design goals were:
1) No heatsink required for the driver transistor to keep the unit small.
2) Very low current draw with the motor not running and when the coil is off.
3) A timing indicator (LED).
4) Robust design.
5) Turn off coil current if the engine is stopped with the hall effect sensor triggered.
6) Prevention of internal arcing in the coil if no spark plug is connected.
7) Usable with breaker points instead of Hall effect sensors.
8) 4.5 to 14V operation. Nominal is 6V (4 AA cells)
9) Very small.

Q1 is a 400V Darlington power transistor with internal flyback diode. It serves as the coil driver.

D1 and D2 have a few functions. The obvious one is to clamp the voltage across the transistor to less than 400V. The second function, believe it or not, is preventing the coil output voltage from getting too high (how high it will get is dependant on your coil) if the spark plug is disconnected. Allowing the coil voltage to get too high can damage the itsy bitsy coils we use. Being connected between the collector and base rather than collector to ground protects the transistor's collector to base junction as well. If the two zener diodes conduct, it will actually turn on the transistor by feedback enough to keep voltages within limits.

Q2 is a small Darlington used to provide enough base current to Q1.

D3 is the timing indicator. JP1 is a jumper that can be removed to save power if the LED isn't being used.

R3, R4, D4, and R2 are the input circuit. The coil charges when the Hall sensor output goes low (or breaker points close). The coil fires when the sensor output goes high (or the points open). C2 prevents the coil from being turned on too long. It'll turn the coil off after about 0.3 seconds if the hall is stuck low (or shorted to ground), or the points don't open. This keeps Q1, the coil, and batteries from being killed.

Again, I haven't built the circuit yet. I just ordered the parts from Digikey. I'll follow up with more info when it's wrung out.

IN the meantime, I hereby solicit comments from the gang...

John

 

[dalem9]

Well John it is wonderful for you to share this with us . Just goes to show were their a will their is a way . Dale

 

[Niceonetidy]

John,

Thanks for sharing, I'm very interested in this and look forward to seeing the results.

Kind regards,

Colin

 

[canadianhorsepower]

Nice circuit but R1 at 470 the led wont work are barely at 6 volts
would a 1k trim pot be better it would cover all the voltage your mentionning
4.5 to 14v

 

[cheepo45]

Thanks for your efforts. I will be following this closely, as I am building a 1/4 scale Gade motor and I would like to find an ignition that fits in the battery box. Let us know what kind of coil you will be using.
cheepo45

 

[Walsheng]

I will be following as well. I have a Little Brother casting kit for one of my next projects.

John

 

[123RWO]

I think you need to get rid of the flyback diode. If you don't it seems to me that it will clamp the coil primary pulse preventing any seconday output. I suggest this transistor instead:
http://www.irf.com/product-info/datasheets/data/irgb14c40lpbf.pdf Digikey carries them.

RWO

 

[jgedde]

I think you need to get rid of the flyback diode. If you don't it seems to me that it will clamp the coil primary pulse preventing any seconday output. I suggest this transistor instead:
http://www.irf.com/product-info/datasheets/data/irgb14c40lpbf.pdf Digikey carries them.

RWO

The primary pulse is positive, so the diode won't conduct. The flyback diode is built into the part. If the diode were in parallel with the coil however, you'd be right.

I like your choice of the output transistor. I discovered those IGBTs designed for logic level ignition coil drive after I posted the circuit. I've started designing a new and better circuit that uses an IGBT from Fairchild.

An IGBT has some notable advantages over a Darlington (a Darlington cannot saturate. But this is a good thing in some cases - no issues with recovery time in trying to turn it off)
1) An IGBT can saturate yielding lower collector to emitter voltages.
2) MUCH less base current is required - hence it'll use less battery power.

Thanks,
John

 

[jgedde]

Nice circuit but R1 at 470 the led wont work are barely at 6 volts
would a 1k trim pot be better it would cover all the voltage your mentionning
4.5 to 14v

The LED I chose will give me about 5 mA at the low limit of 4.5 volts. It'll be lit, albeit nowhere near as bright. That's another reason I chose a green LED. The human eye, especially mine (being red colorblind) is much more sensitive to green. I chose power consumption over brightness.

At the high limit of 14V, the LED would be overdriven. I honestly put no thought into the LED when I considered the circuit at 14V.

My new circuit will have a choice of LED drop resistors for the voltage for which the circuit will be used. I designed it for use (in my case) of 4.5 to 6V - what I'd get from 4 standard alkaline AA cells.

John

 

[CallMeAL]

This looks very interesting for an IC engine I hope to build soon. I wish I understood electronic schematics better and knew how to build a circuit!

I'll follow this and see if I can figure it out!

Al

 

[jgedde]

OK. Here is a version I'm happy with. It uses an IGBT instead of a Darlington output transistor. This reduces power consumption of the driver circuit, reduces part count, and gives about 20-30% more coil voltage at 6V.

I'm relatively sure the hand drawn schematic with the Darlington will work if you want to try it. However, I no longer have plans to build that one. I will be building this one instead.

I've also addressed the LED resistor issue.

Operation of the circuit is more or less the same on the input side. As far as the output part, Q1 is the coil driver (an IGBT), Q2 drives the IGBT gate. Q3 is used to quickly turn off Q1 which is what is needed for a good hot spark. I could have eliminated a transistor by driving the gate of the IGBT to ground when it needed to be turned off. But, this would have consumed power just to keep the coil off...

If anyone is curious, I used Altium for the CAD schematic. Also attached is a PDF version which is easier to read.

John

[Edit: 4/1/13 The circuit has been built and tested. See final version in post #47. J. Gedde]

 

[canadianhorsepower]

Curious to know what kind of coil your planning to use at 6 volts

do you use PCB artist for your PCB board?

cheers

 

[jgedde]

For PCB layout and schematic CAD I use Altium Designer. I have a license for it through my job.

I don't plan on making a PCB, although if I see any interest, I may lay one out and put together a kit for sale.

My version will just be put together on a small solder type protoboard. The coil and driver circuit will live inside a small model buzzcoil box on the engine (although it isn't a buzzbox). See the picture...

The CAD model for the engine was done using Alibre.

For the coil, I plan on making my own. The book on model ignition coils by Bob Shores gives lots of info on how to do it. We have coil winders and laminations at work.

To test the circuit I have a small 6V coil from J.E. Howell.

John

 

[Maxine]

Thanks for posting that schematic.

 

[stevehuckss396]

Will the circuit fire a coil with a primary resistance of 0.8 ohms

 

[Lakc]

Will the circuit fire a coil with a primary resistance of 0.8 ohms

I am not so sure it can. Resistance is only one measure of a coil. In your case, what is probably different then a standard coil is its reluctance, "L" measured in henries.

 

[jgedde]

Will the circuit fire a coil with a primary resistance of 0.8 ohms

Yes and no. The 0.8 ohms isn't necessarily a problem, but the inductance of the coil is really what matters. Do you have any further info on your coil?

If we consider the coil as a plain inductor (there's a bit more to consider): The inductance divided by the resistance represents a time constant. After 5 time constants, the current through the coil will be set by the resistance.

The idea is that your dwell angle only turns on the coil for 4 or 5 time constants after which it will obtain no more energy and just heat up. Of course this doesn't happen in practice since spark timing is an issue and the time the coil be be on will be reduced at higher RPMs since dwell angle is fixed.

All that said, this circuit can and will drive automotive ignition coils. Inn fact, the output transistor was designed for just that.

I've done a PSPICE simulation using a MSD HEI coil with a primary resistance of 0.7 ohms and a primary inductance of 8 mH. It does work, but the transistor would need a heatsink. This was without using a ballast resistor.

For what it's worth, my coil has a primary resistance of 1.05 ohms and a primary inductance of 4.4mH.

John

 

[dsage]

OK. Q3 is used to quickly turn off Q1 which is what is needed for a good hot spark. I could have eliminated a transistor by driving the gate of the IGBT to ground when it needed to be turned off. But, this would have consumed power just to keep the coil off...

John

John:

Yes your circuit looks remarkably like mine. Great minds thing alike I guess.

I had mentioned in my original post:

http://www.homemodelenginemachinist.com/f26/ignition-circuit-help-19673/index2.html

that I was lazy and used the symbol for a regular transistor just to make the circuit more familiar looking.
Something else you might not have noticed about my circuit (and now yours too since you quoted the same IGBT) is that I'm using the "L" variant of the IGBT with has a Logic input (can't speak for the alternate part number you spec'd).
You can drive this transistor directly from a microprocessor or other logic level device. I have used this driver end after a simple 555 timer oscillator to create a BUZZ COIL like output. You set the 555 for a frequency around about the equivalent of 1000 rpm and the hall sensor or whatever enables/disables the 555 to produce a burst of sparks instead of just one. Works really well. My friend also uses the same IGBT driven by a PIC micro on his EEVIC engines.
http://www.evicengines.com/

A couple of comments on your variation on the circuit. I haven't checked them..

Is there enough leakage from the base of the IGBT to actually make Q3 work? Q3 would need some voltage on it's emitter and enough current to bias Q3 off. I guess you could tell by measuring the voltage emitter to collector on Q3 when sitting idle or with a scope. It should be pretty low (like 0.2v) if Q3 is turned on.

You have different resistor values around the coupling capacitor I and and I haven't checked but, is the coupling capacitor value high enough to pass REALLY slow rpm's as you get when cranking the engine. If not you will only get a very short dwell time passing through it. I selected mine to pass everything up to several tenths of a second ( I've forgotten exact figures). Basically anything BUT a continuous input. It's just there for safety.
I had to use a scope to see that the capacitor was NOT charging up enough to have the circuit time out for very low RPM's

The last point is that like any darlington bipolar transistor the IGBT has a Collector Emitter voltage of around 2volts when ON. If you're going to run it on 6v your coil would need to be rated for 3 or 4 volts or so, else it will be starving a bit. (6bat - 2Vce = 4) Having said that, there is usually enough output from most coils for a model plug regardless of what voltage you have.
You only need about 2Kv on a small plug under compression.

Good work.

Sage

 

[jgedde]

John:

Yes your circuit looks remarkably like mine. Great minds thing alike I guess.

I had mentioned in my original post:

http://www.homemodelenginemachinist.com/f26/ignition-circuit-help-19673/index2.html

that I was lazy and used the symbol for a regular transistor just to make the circuit more familiar looking.
Something else you might not have noticed about my circuit (and now yours too since you quoted the same IGBT) is that I'm using the "L" variant of the IGBT with has a Logic input. You can drive this transistor directly from a microprocessor or other logic level device. I have used this driver end after a simple 555 timer oscillator to create a BUZZ COIL like output. You set the 555 for a frequency around about the equivalent of 1000 rpm and the hall sensor or whatever enables/disables the 555 to produce a burst of sparks instead of just one. Works really well. My friend also uses the same IGBT driven by a PIC micro on his EEVIC engines.
http://www.evicengines.com/

A couple of comments on your variation on the circuit. I haven't checked them..

Is there enough leakage from the base of the IGBT to actually make Q3 work? Q3 would need some voltage on it's emitter and enough current to bias Q3 off. I guess you could tell by measuring the voltage emitter to collector on Q3 when sitting idle or with a scope. It should be pretty low (like 0.2v) if Q3 is turned on.

You have different resistor values around the coupling capacitor I and and I haven't checked but, is the coupling capacitor value high enough to pass REALLY slow rpm's as you get when cranking the engine. If not you will only get a very short dwell time passing through it. I selected mine to pass everything up to several tenths of a second ( I've forgotten exact figures). Basically anything BUT a continuous input. It's just there for safety.
I had to use a scope to see that the capacitor was NOT charging up enough to have the circuit time out for very low RPM's

The last point is that like any darlington bipolar transistor the IGBT has a Collector Emitter voltage of around 2volts when ON. If you're going to run it on 6v your coil would need to be rated for 3 or 4 volts or so, else it will be starving a bit. (6bat - 2Vce = 4) Having said that, there is usually enough output from most coils for a model plug regardless of what voltage you have.
You only need about 2Kv on a small plug under compression.

Good work.

Sage

Thanks Sage,

Yes, I have checked the values in the input circuit. They'll give up to about 350mS of on time.

Both the Fairchild and the IR IGBTs have very good Vce(sat) figures, even at 4.5V of drive. Much less than 2V or so like a Darlington... The datasheets show about 1.4 to 1.6V being the max value, with more like 1.2V-1.3 being typical at max coil current. Of course, the coil doesn't draw this current as it charges as the inductance limits the current. That said, during charge-up, Vce is very low... Maybe about 0.2 to 0.3 volts - it's hard to tell from the graphs in the datasheets. Of course, if Vce rises as the coil charges, the coil wil lose some energy.

0.2-0.3V is way better than a Darlington, simply because a Darlington cannot saturate while an IGBT can. I'd have considered using a MOSFET with a low Rds on, but then I'd have to somehow get 8-12V of gate drive and add the clamp zeners back in. Also, a low Rds on, high voltage MOSFET has a huge amount of gate charge which will cause issues in getting the device to turn on or off fast. There are ways around all of these issues that I employ at work (I design motor controllers for space vehicles), but they add unwanted complexity and would be hard to make work at low RPMs (bootstrapping a switched capacitor circuit).

In any case, my current coil design gives about 12kV at 4V. I'm using ultra high mu, high saturation, laminations so I get a high energy product. The usual trick of tearing apart a power transformer is not what I'm doing...

As far as Q3 is concerned, an IGBT has a comparably large amount of gate charge (17nC for the Fairchild, 27nC for the IR) which has to be dumped in order to get it to turn off fast. That's where the current for Q3 comes from. When Q3 turns on, it dumps the gate down to about 400mV. This is because Q3 doesn't saturate. The internal resistor in the IGBT from gate to emitter pulls it down the rest of the way.

There is a vintage appnote from TI that I believe details the turn off circuit I'm using. If you Google MOSFET gate drive, you'll find it.

I think I might play around with a 555 timer to generate multiple sparks at some point. But, wouldn't multiple sparks wear out the spark plug faster? Certainly it will use more battery power. If the first spark effectively ignites the mixture, then all is well. My CR is only about 4.5:1 so I don't think I'll have too much trouble unless my carburetion is really bad.

Thanks,
John

 

[jgedde]

John:

My friend also uses the same IGBT driven by a PIC micro on his EEVIC engines.
http://www.evicengines.com/

Sage

Wow! Those EEVIC engines are amazing! Electronically controlled and actuated valves, are in my opinion, the Holy Grail of engine design.

I wasn't aware that anyone had really pulled it off at high RPMs...

I dream about a small block Chevy with electrically controlled valves! Turn a knob for a mean choppy idle! Seriously though, in these days of high fuel prices and environmental concerns, that technology would be the cat's meow in a motor vehicle...

John