Model Engine Builder Magazine

Yes, they do have a high standard! Did you destroy the mis-print? I'll keep mine and in 100 years, it will be a rare collector's item. Like stamps. :-))

BTW MEB: My V-twin is almost ready, but I'm *still* working on the µC-controlled ignition I decided to build. Will have much more features than that engine needs, but it will be my last design for all engines with spark-plugs coming.

• 24 cylinders (or more, depending on timing of crankshaft)

• up to 60000RPM (keeping the timing to within +/- 0.5 degrees crank)

• ignition advance table (depending on load and RPM)

• adjustment of coil's charge time depending on supply voltage

• ...

Nick

I have a hard time throwing anything away and this won't be one of them.

Is that the v twin if I remember correctly that was in the magazine earlier this year or another one?

How about a link to pictures and sound when ready.

Is that ignition using coilpacks? (two plugs one coil)

Wes

The one from MEB (forgot the name; looks like a Harley, open crank case)

Sure. When completely finished. I'll make a video too.

Haven't investigated in coilpacks yet. I'm using self made coils.

0.5ms charge time @ 12V (fast!!!) 23A, 50mJ. Size is about 25mm diam, 55mm long. And it looks like a scale ignition coil. :-)

Nick

I remember reading this a while back. HTH some.

Thanks. But this type of coils are made for breaker contacts. They are current limiting by their inner resistance. Mine has a very low R (0.05 Ohm) and it needs to be driven by electronics, or it will simply melt if charged for too long. The short-circuit current would be 200A. But this is the only way to make a fast coil with constant energy up to the maximum revs.

Nick

Well that explains how Steve's link didn't fit with what you are constructing. I rather enjoyed the link though.

Did you build the coils too? Am I going to see a Nick Mueller article some day in MEB?

I believe I'm a day late wishing you a Merry Christmas since everyone doesn't observe it on the 25th. Considering tomorrow the 26th is the day Americans shop for bargains with our typical 'gusto' maybe we could call it boxing day ;)

Merry Christmas,

Wes

Yes. I didn't get what I wanted and the coils available for model engines had no technical data at all. So I wound my own. And built a coil winder. And a counter. And vacuum impregnated them. And built a vacuum box. And repaired a vacuum pump. And ...

In the next issue (it is already on the way). :-) But not about coils and ignition systems.

What!? We do have Christmas? ;-) Same to you (and all those reading).

Nick

Well, I look forward to reading it. I just renewed the subscription a few minutes ago.

Wes

That is really beautiful. Any chance we could know turns, wire sizes and core material?

Nice!!!

Primary is 93 windings of 1.12mm, secondary is 1600 of 0.1mm. Core is iron powder from Micrometals (43mm long IIRC). But again, the coil needs electronics! The charge time has to be limited so you don't exceed the 23A. That makes 0.5 ms at 12V. If you use it on a breaker, the coil will go up in smoke. It would draw 200A from a 12V battery.

I tried triple-isolated TEX-E wire on the secondary, but it makes you poor! The wire is just too expensive and it is too thick. So secondary is double isolated wire with inter-layer isolation. Takes some time to acquire the skill of winding. :-) I'm getting 8 kV on the secondary, but that requires high voltage through induction on the primary. I do have about 500V on the primary (@ supply-voltage 12V) with protective switch-on of the coil to reduce the primary voltage (happens when the secondary has no spark plug connected).

Did I describe the electronics before? It's µC driven. 8 processors working parallel, each processor having 20 MIPS

3 (maybe I'll change that to 4) processors do handle the position sensor of the crank, calculate the timing, switch on and off the coils. Other processors do handle reading the supply voltage, read the RS232, etc. I do have a theory, that I can calculate (no throttle sensor required) the load of the engine if there are less than 4 or such cylinders. I still have to prove that (or fail to do so). The core of the software is working, I'm finishing the setup-stuff today and plan to have a working beta tomorrow. Technical data:

• max RPM 60000, at 40000 RPM timing is accurate to +/- 0.5° crank degrees. That are about +/- 0.75µs! I could get more RPM out of it, but you'd have to show me that engine before. :-)

• Minimum RPM is something around 20 RPM

• mapped timing. Parameters are load (external sensor) and RPM.

• mapping table is bi-linear interpolated to about 0.5 million points

• configurable via RS232. Laptop (any terminal does it) can be used as online display for several (configurable) parameters like RPM, V_supp, load etc.

• up to 24 cylinders (and even more under certain circumstances)

• Multi-spark: Spark sequences for lower RPM or starting. Number of sparks configurable (I'm using 3; tried it with seven). RPM where it switches to single-spark configurable. I do have a software-issue with multi-sparks at high RPM with many cylinders (> 16) and many sparks (overlapping). Shouldn't be noticeable under normal conditions and I'll fix that later.

• Charge time of coil adjusted in realtime

• Rev-span (where the +/- 0.5° are kept) is 1:20 (idle to max RPM) Normal engines do have a span of about 1:10

• Rev-limiter

• Max current for coils is 40A constant, 120A peak. That should make possible a coil with 0.2ms charge time that is good for >> 50000 RPM. I know, that's a bit on the high end.

• other things that I forgot now :-)

I *hope* to fire the V-twin first time at new year's eve, when others do waste their time drinking Champagne and firing fireflies. :-)

Nick

This gives enough spark under compression?

Until now, yes. :-) There are several factors for a spark to initially jump:

• compression

• condition of spark plug electrodes

• ratio of the mixture (rich, lean; a lean one needs higher voltage)

• dU/dT of the voltage

• and of course the voltage (having 8 kV with this coil, limited by the protection on the primary)

I have tested this one with an engine that has quite a low CR, but a spark plug that is oiled beyond belief.

Don't forget that I do have 500 V on the primary side and a dU/dT at the primary that is by magnitudes higher than required (forgot the numbers I once calculated), and also higher than a conventional system. It more resembles a CDI (but it ain't).

The energy stored is 50 mJ (cars do have something at 120 mJ to 180 mJ). But literature says that 0.5 ... 10 mJ are enough to ignite the mixture. I'll try that as soon as everything is up and running.

Spark duration was measured to be about 4µs with a clean spark plug. A oily one will dramatically reduce the duration.

Nick

Working is what counts!

Some friendly observations from an old ignitor designer:

dV/dt isn't relevant except where parasitic capacitance is an issue. d (flux)/dt is what counts, which boils down to dI/dt. Your secondary open-circuit voltage will be N * v1 * k where N is the turns ration, v1 is the primary voltage and k is the coupling coefficient between primary and secondary -- typically between 0.4 and 0.8 in a design like this though it appears that you're getting rather good coupling. v1 will be L1 * dI/dT where I is the primary current. I designed high-voltage ignitors for oil and gas burners, not engines. I don't pretend to know much about automotive ignition. I'd guess (pure conjecture) that automotive ignitions may have far more energy and OCV than are necessary in order to get acceptable ignition for very cold starts and at higher speeds as efficiency drops due to shorter dwells. You won't have sub-zero (< -18C) cold start issues, and your elex will maintain efficiency at high speeds.

(and ignition in the higher compression environment of an engine)

My picture is: dV/dt counts on the secondary to start the spark. If there is stray capacity it also has some R. A fast rise will prevent energy to be charged into that capacity. As soon as the gas is ionized, the spark's burning voltage drops by magnitudes. At least this is the argument for CDI. They do have a higher dV/dt. That's what I have read, but also I never found a very good explanation. I *think* the reason that the classical transistor-ignition being not as good as the CDI of those days was the lack of suitable transistors. Things changed some years ago with the IGBT. There are also dedicated IGBTs available for ignition systems that do have an integrated anti-parallel Zener. I'm using an IGBT (U_gs 600V) and so it resembles more a CDI that works with about 400V. Classical transistor ignitions work with about

100 .. 150V.

That's right, my fault.

Yes, it looks like. I have measured the voltage on the secondary (with a self-built HV-probe; capacitive divider).

That's absolutely right! The safety-margin is by magnitudes. But also -with environmental regulations- they are running on the lean side and thus need more energy. Ignition coils of modestly modern cars are still partially self limiting (by current) but not as much as the coils we know from out first VW-Beetle. :-) The most advanced coils that are integrated into spark-plugs should resemble my design. But I didn't get my hands on one until now.

Nick

Would a knock sensor possibly help?

The sensor is a ceramic microphone tuned or filtered to the fundamental resonance frequency of the engine block. Knock causes a large jump in the output signal level. In a closed-loop engine control the uC slowly advances timing until knock begins, then quickly retards until it stops. You could initially leave that out and use the sensor to light a warning light, or show the level on a meter.

Right conclusion, anyway. Parasitic capacitance and secondary self-inductance combine to form a resonant frequency. The self-resonant frequency limits risetime to given peak voltage.

Yes.

With older flyback (Kettering-type)ignition there was usually a capacitor across the primary to preserve the points. This formed a resonant circuit with the primary inductance to limit peak primary voltage, which was still much higher than battery voltage.

More modern solid-state ignitions used a transistor to interupt primary current, and a zener to limit primary voltage to around 400 volts. One such transistor was MJ10012.

In this case the secondary dV/dt was limited by secondary self-resonance.

CDI drives the primary to high voltage by dumping a charged capacitor into the relatively low primary inductance. This produces much higher primary dV/dt, but more importantly also higher dI/dt.

Partially. The early CDI's used SCR's on the primary side. 600 volt SCR's have been available for decades. Bipolar transistors capable of

800 volts and more have been readily available since the 70's. Transistors, MOSFETs and IGBT's can be used in a flyback ignition because they are capable of being turned off -- SCR's can't be turned off once fired. SCR's also have limited dV/dt capability.

I don't know what a "classical" transistor ignition is, but the MJ10012 (400 to 600 volts) was specifically intended for automotive ignition in conjunction with a 400-volt zener. They were sometimes used in conjuction with a chip that maintained essentially constant dwell time over varying speeds. This wasn't a microcontroller, just a simple bipolar ignition control IC, MC????.

Perhaps you did this, but just in case: secondary parasitic capacitance can be significantly reduced by winding the secondary in "pi" windings: several tall narrow windings in series rather than several full-length layers. Imagine a stack of washers perhaps 0.75 to

1.5 mm thick, each separated from the next by perhaps 0.75 to 1.5mm. I made bobbins by cutting narrow grooves (0.75 to 1.5mm wide, sometimes as much as 20mm deep)in delrin rod. The tricky part is making the vias from the top of one layer to the bottom of the next. Commercial H.V. coils using this method use molded bobbins. Tooling for such bobbins is rather expensive to make, but the bobbins themselves are then quite cheap. I was able to make 40 KV coils with self-resonant frequencies on the order of 50 KHz, or rise time to peak voltage of 5 uS.

Thanks. I'm did think about integrating one in a next version. Requires some spectrum-analysis and tuning to the engine. So not a simple task!

Nick