Electronic glo-plug ignition?

Home Model Engine Machinist Forum

Help Support Home Model Engine Machinist Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.


Well-Known Member
HMEM Supporting Member
Jul 8, 2009
Reaction score
"I have a clever plan milord", or maybe it's just a brain-fart - let's find out.

A little background set up:
Building a multi-cylinder glo-plug engine might be relatively easy(-ish), you can bolt multiple Cox .049's around a single shaft and get a radial of sorts. I've also seen several multi-cylinder engines which used glo-plug ignition, with more than 2 cylinders. But lighting and keeping those plugs lit is the struggle. At lower RPM's it's best to keep power on the glo-plugs, to keep them hot for reliable engine operation. Another issue would be the glo-plug voltage, the last time I ran a glo-plug engine I was using the big 1-1/2 volt dry cells. A 2 volt plug was a "Hot" plug. Kinda dating myself there aren't I.

This is the way that I see a glo-plug working, correct me if I'm wrong. To start the engine you need to get the element in the glo-plug hot using the starting battery. After all, that hot element is what ignites the fuel-air mixture, isn't it. The fuel-air mixture goes bang and the element heats up from the combustion. Once the engine has been running long enough, the surrounding metal is no longer sucking as much heat from the element, the element remains hot enough to support combustion, and the starting battery can be removed. At lower RPM, the element has a long enough time between power cycles that it can cool off to the point where it can no longer support ignition, and the engine dies. This is where an on-board glo-plug battery comes into play, it can keep the glo-plugs hot at the lower RPM - making the engine more stable. This starting battery can be a problem, each glo-plug is a power hog so this battery needs to have a fairly large capacity. The more cylinders you've got, the bigger that battery needs to be. The voltage needed by the glo-plug is another problem, it's typically 1-1/2 to 2 volts. !-1/2 volts is a typical dry-cell battery, but not many other batteries are in that range. Lead-acid batteries are 2 volts per cell, Nicad and Nimh are both 1.2 volts per cell, while the various flavors of Lithium batteries are about 3.7 volts per cell.

So here's my clever plan/brain-fart:
What if we were to take one of the readily available micro-processors out there and use it to switch an output on and off - very fast. We could switch the glo-plug on somewhere around where it needs to come on, leave it on for a little while, then turn it off. The bog standard Arduino Uno has a clock speed of 8MHz and can be over-clocked to 16. I know an Arduino output won't drive a glo-plug, but it can drive a power transistor or a MOSFET that will. I'm a EE and I've done design work like that, but it was so many years ago that it isn't even funny any more. Will an Arduino handle it? I think so - maybe. At 10K RPM, that's 48000 clock cycles per revolution that the Uno is capable of running - right out of the box. That's a lot of computing that can be done every revolution, and there's a lot of faster micro-processors out there.

The glo-plug battery will still be the hang-up, does it have to be 1.5 volts? To run the glo-plug 100% of the time - youbetcha! But what if we were able to turn the plug on and off REALLY fast? I know that years ago some of the white strobe lights on transmission towers in the US were 120 volt bulbs run on 240 volts. The 240 volts was applied for such a short time that the bulb filaments heated into the white hot range, but it wasn't on long enough for the filaments to melt. I would think that you could do the same thing for a glo-plug. There's probably some way to calculate how long it will take a specific glo-plug to melt at, OH say 3.7 volts or maybe even 7.4 volts. I'm a EE and they didn't teach me how to do that in school, or if they did it didn't stick. Will the glo-plug still be a power hog, yeah. But it will only be on for a small fraction of the revolution instead of the entire revolution so the battery capacity can be much smaller.

Brain-fart bottom line:
We take a relatively cheap - but fast - micro-processor, add some power transistors/power MOSFETs, add probably a 2S Lipo pack of some sort, write some code, then mix well, and we've got a distributor-less ignition system for a multi-cylinder glo-plug engine that SHOULD require a significantly smaller battery than if you kept the plugs lit all the time. It would definitely weigh a lot less, and the GEE-WHIIZ factor would be through the roof.

What you are talking about is a switchmode power supply. You don’t need an Arduino to drive the MOSFET although that would be an elegant solution. A 555 timer in astable mode would be fine to drive the MOSFET. A potentiometer would provide the adjustment to the duty cycle.

However, why not connect the glow plugs in series and run them directly from an appropriate voltage battery? For example, five 1.2V glow plugs in series would require a 6V battery. The main drawback of this method is if one glow plug fails they all go out (like Christmas tree lights).
There's nothing especially original about this proposal-this is what aeromodellers and boat modellers have been using to start their glowplug engines for about the past 50 years...certainly since the 1970s-when electric starters became more common...in the form of the near ubiquitous 'power panel'....which runs off a 12V motorcycle or gel cell rechargeable (or latterly-a 3S Lipo, for those who have them)-providing 12V to the electric starter, 12V or 6V to the electric fuel pump (either integral to the panel, or plugged in as a separate accessory) and a purely nominal '1.5V' to the plug....in fact the plug supply is usually a PWM controlled pulsed 12V supply, since it is current that heats the plug element. Virtually all current panels use this system....and in some you can hear the pulse rate as an audible 'singing' of the panel when in operation. Most employ a manual pot to adjust the pulse rate-coupled with an ammeter, to adjust the average current to that needed by any particular plug to give a good glow...which could be anywhere from 1-1/2 (some of the cheaper beginner 049 engines) up to about 4 (heavy duty or racing plug) amps...there are a few that employ a sensing circuit that automatically adjusts the current to the load-but these tend to be a bit pricier.

As jack620 has posted above-the technology to do this reliably and effectively has been around since the mid 70s-using arduino is a case of 'gilding the lily'...and the early models didn't even use mosfets-just the available power transistors of the day...frequently the old reliable 2N3055.

With the increasing dominance of Lipo batteries in the modelling scene, there are a few newer options that involve voltage regulators rather than a switchmode approach-but these are intended to be used with a single suitable 1S Lipo, dropping the nominal 3.7V (4.2V at full charge!) to a safer 1.5-2V for glowplug lighting....these don't seem to have made big inroads into the glowplug starting market yet....but give them time...

Doubt you need any on board glow for a 2 stroke but anyhow....
Built a pwm unit last week using anATtiny85 and a single mosfet (IRLB3034) for the four glow plugs on the boat engine. ( four stroke)
Runs off a 3 cell Lipo and a cut down usb ATtiny85 module. rc input to switch plug on or off.
Can poke the details in here if you like.
Photo below has yet to have mosfet fitted. 2 *2n7000 mosfets and tactile switch are to operate 2 mini voltmeters....as it uses n type mosfet for the main switcher, not possible to use just the one 2 wire voltmeter.
Forerunner( used 2 only IRLZ44n mosfets) in next photo....


  • ASBuiltOwnPCB.jpg
    117.1 KB · Views: 236
  • AsBuilt2.jpg
    142.9 KB · Views: 222
Last edited:
Attached schematic for a 9 cylinder glowdriver, looks easy to add additional plugs, this circuit has been around a fair while but not back to the 80's. Description of circuit and components welcomed from the electronic buffs.Glowdriver schematic 9 cylinder.jpg
You are overthinking the issue.
1) NiCd voltage is a perfect match for modern glow plugs. I hesitate mentioning a specific voltage because a NiCd cell is charged up to 1.5V and can be drained down to 1.1V. They spend most of their life delivering 1.35V

2) You do not need to switch the driver at the maximum frequency the controller (Arduino) can handle, you will be wasting a substantial power in switching losses. The filament has thermal inertia. If an engine can run fine at 2000 RPM it means the filament thermal time constant is much greater that 0.5 mS therefore switching at 2000Hz is plenty fast to eliminate the filament temperature fluctuations.

Since you are switching you could use a higher voltage battery but there is no advantage.
Use a P channel MOSFET to make the drive simple and save power. They are available with ridiculous low ON resistance and high current capability.

The wire from the switch to the plug will necessarily be kind of long, so you have to watch out for the inductive kick back. A simple Diode located near the switch will do it. In this particular case it does not even need to be a Schottky or a fast recovery type.
Regarding the duty cycle What count is the RMS of the current. For a square wave IRMS = Ipeak x Squareroot (Ton/Period)
As long as the Ton does not allow to reach melting temperature, then it does not matter how long it cools (the period) Tungsten melts at 3422 C but you need to het the filament to around 800-900C so you have some safety band. Now if you use a 4.5V source you would have a peak current 3 time as high as you would with a 1.5V source. That is 9 times the heating power. So if your pulse is 1/10 of the thermal time constant the filament should not exceed the normal operating temperature. Mind, these are not precise calculations, are simple setting boundaries. There are many nonlinearities in the phenomenon associated. Not counting the mechanical stresses due to the continuous pounding in the combustion chamber. As I said there is no advantage in using higher voltage, what weight advantage a different battery chemistry may give is lost in higher losses and added heatsink weight.
However, why not connect the glow plugs in series and run them directly from an appropriate voltage battery?

The shared "ground" via the case means they are all parallel. You can only run two plugs in series when they are connected through the case.
The main reason I was thinking about switching the plugs is the amp hour capacity required of the battery to keep the plugs hot 100% of the time. Another reason for switching each plug individually was reliability. I know that you can easily series the plugs and use whatever voltage you wanted to drive them, but if one plug goes bad you lose the whole string. By switching the power to the plugs individually, you lose one plug, you lose one cylinder. And you can easily determine which cylinder is not firing, it'll be the "cold" one.

What I have in mind is kind of like PWM, but then again it isn't. PWM would also result in a significant reduction in required battery amp hour capacity, just because you aren't leaving the stinking plugs turned on all the time. But why waste the power turning the plug on during the exhaust stroke? What I'm proposing is to use a crankshaft position sensor to individually give each plug a kick so that it is hot when it needs to be hot. If anything, this would be more like an electronic set of points - only for glo-plugs.

Yes, with what I'm talking about you are duplicating your driver circuits, but each circuit only needs to handle a fraction of what would be required to drive all the plugs simultaneously. There are probably existing driver chips that would be able to handle the load of an individual plug, probably even some with multiple drivers available on one chip. 10000 RPM would only be 10KHz, which is pretty slow in the electronics world.

Elektor electronics magazine had a special glow plug regulator that could deal with "flooded" glow plugs. Just a simple circuit with a power transistor and op-amp that monitored the PTC characteristics of the glow plug. It was powered by the 12v battery of the starter box.
As already stated, only two plugs can be possibly set in series for obvious reasons.
Regardless of all the clever and not so clever ideas, there is no way on earth to get more energy out of the battery than what is in it and no way to use less energy that required to supply whatever current one needs to keep the plug sufficiently hot at idle. All the timing, crank encoding and clever ideas are not defeating the fundamental laws of physic governing the issue. The only things to optimize are the switching and distributing losses and the weight if that is important.

The first lesson of engineering I learned was: Before trying to invent something new make sure you understand the history of the invention and the trade off found in previous designs. There is nothing wrong with new clever ideas but the wheel has already been invented and in 99.9% of cases is best to be round.

I'd like to counter your argument by saying I'm not trying to get more energy out of the battery than what is in it, or trying to use less energy than what is required to keep a plug hot at idle. No offense, but if you believe that I've got this bridge in New York I've been trying to sell…

Timing is everything, I'm saying we can time the energy use for each plug to only when it's needed for that plug. Do we really need to heat a plug during the exhaust stroke, combustion just did it for us. Would it make a difference if it's a 2 stroke or a 4 stroke engine? By not heating the plug when the heat isn't needed we aren't wasting that energy. I'm not sure, but I think this will wind up being a numbers game in amp hours. Will a PWM driver at setting X use less amp hours than a full voltage impulse at a shorter duration? Don't know. But I do know that a 1 amp hour battery pack weighs a lot less than a 4 amp hour pack.

Also if you are using a common driver, as the number of cylinders goes up so does the load. I haven't looked yet, but I would imagine that it's a lot easier to find a driver on a chip that will handle a 1 amp load than it is to find one for a 4 amp load.

I'm not trying to invent a rounder wheel, just wondering if maybe it can be lightened up a little bit. Maybe it's worth the effort, and maybe it isn't - but it's an interesting mental exercise either way.

Don I am not going to school you in thermal flow. You can find what you need on line. The filament has essentially the same temperature when the current is ON, OFF or at any time during the cycle. The thermal inertia of the filament will not let it respond so quickly. When you connect a battery to the plug you will notice that it take a humanly appreciable time to reach full temperature like 1/2 a second. All your clever timing is not going to do anything. An incandescent light bulb does not appreciably change the temperature at 120Hz rate. If your PWM does a couple of cycle in each of the engine cycles you are good.
I am not getting into a pissing contest with you. I pointed you in the right direction banking on my professional experience, but you think you know better. Good luck.
By the way I had the same problem feeding my Edward radial 5.
Don, I'm glad you posted this separately A) because there is nothing like a burning idea that wont go away LOL
B) some good collective info on glow plugs, drivers & engineering is being contributed.

I'm not sure that airborne battery weight & energy conservation is a huge factor in the grand scheme of things. Just as a random example - this dirt cheap Lipo weighs only 68g. That doesn't translate into much building material. I tried to determine how much energy would be consumed by a full day of flying with 5 plugs lit full time just a ballpark guess. Someone check my math but if correct, suggests not a lot is actually consumed.

The issue (I think) is the nominal voltages of Lixx cells is not a good match for glow plugs. They have plenty of current capability & great energy density but say 3.5-4.2 nominal volts per cell will fry the plug wire element. So something is needed in between to regulate that voltage down to 1.5 ish volts & that device must be capable of the same current level. Nominal voltage of NiCd or NiMH cells are a good match. One could argue they have lower energy/weight density & C-rating over Lixx chemistry, but my guess is there is probably still a Nixx solution in there. I mean after a hard day of flying we have to re-charge the RX, TX... why not an ignition battery too?

We passed through a similar issue with RC servos & receivers. All designed around 4-6v corresponding to 4-5 Nixx round cells for many years. The new Lixx batteries arrived & they were a voltage mismatch. So for a while we had inline regulators or other solutions. Now the radio gear has adopted 'HV' in servo motors or boards & same with RX/accessories and its largely back to plug & play with say 2S lipo (= 7 - 8v). I don't think re-designing the filament wire for HV is slam dunk easy or they probably would have done it by now.

Anyways, good conversation topic.


  • SNAG-1-22-2021 0000.jpg
    SNAG-1-22-2021 0000.jpg
    81.6 KB · Views: 196
Much of the issue described can be eliminated simply by picking the correct glow plug.
I have a 4 stroke model aircraft engine which idles very slowly with a large prop and will happily keep running with no power to the plug.

Latest posts