270 Offy

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.
You're close. Here's the one I used as a model:

It was simple enough to throw together a test fixture for the completed ignition, but like most things in my shop that start out simple, it grew into a project. I wanted to use the spindle of my Tormach to drive the magneto since it can be easily and accurately spun up to 5 krpm - a perfect range for testing. However, I wasn't sure if the EMI radiated by the plug wires might create problems for the Tormach's electronics or the computer controlling it. The fast rise times of a CDI's EMI-rich secondary waveform is one of the reasons CDI's fell out of favor with the automotive industry. Even though they're capable of generating impressively long arcs, stray capacitive coupling of their steep leading edges created difficult to control cross-fire problems in full-size engine applications.

I didn't want a rigid mechanical connection between the mill's spindle and my fixture that might damage the magneto should EMI cause one of the axes to take off on its on. So, I used a breakaway Delrin coupler between the two. This required the magneto and its trigger disk to become a standalone assembly with its own input shaft. What should have been a one hour side project turned into a full day of throw-away work.

The scope photos were collected using a high impedance probe coupled to the output of the CDI through only the stray capacitance between the two. Although they're accurate snapshots of the actual waveforms, they provide no real information about voltage levels. The extremely fast rise times in the CDI's ringing outputs are an order of magnitude faster than anything created by an equivalent transistorized ignition. Each lobe on each waveform can potentially create a unique discharge, and after-market suppliers didn't miss their opportunity to promote CDI's as 'multi-spark' systems. The short durations of these sparks, however, limit the energy carried by them even while jumping impressively wide gaps in display cases on part suppliers' counters. In a model engine where less than a mIllijoule is probably needed to reliably fire a plug, they're hard to beat when the importance of the ignition's small physical size outweighs its cost.

The ring or resonant frequency of the output of a points or transistorized ignition is limited by the large inductance and interwinding capacitance of their low Q energy storage coils. The external capacitor nearly always used with mechanical points will dominate any stray capacitance and keep its resonant frequency in the tens of kHz. The small inductance and interwinding capacitance of the CDI's output transformer pushes its ring frequency into the tens of MHz and leaving it heavily dependent upon stray wiring capacitance. This effect is evident in my measurements where a minor change in stray capacitance between my initial bench setup and the later one on the mill dropped the resonance from 67 MHz to 25 MHz.

Although I don't know how much actual energy the S/S Magnum CDI is really capable of, I saw no significant drop in its output in my four cylinder test over an effective crankshaft range of nearly 5k rpm. After ten minutes or so at full speed everything seemed to be still working, and so it looks like it's time to begin work on the cylinder head. - Terry

Last edited:
Very nice setup for testing.
Out of curiosity have you ever tried building a pressure chamber for a spark plug? I never got around to it (yet).
As you know ignition system performance changes significantly with pressure and fuel / oil etc. I wonder how the S/S unit performs under "real" conditions? I have several of them and I've always wondered if they are the source of poor running. Although my V8 seems to be ok at <5k. The maximum I feel it's mechanically safe to run it.
Would be an interesting test since you have such a nice setup made there. (variable air pressure via regulator to the chamber to simulate and record compression).
It has always surprised me that my scope has never been affected by the EMI considering they have processors in them these days. I see yours was pretty close by.
My Fluke meter just sitting on the bench (un-connected) near a CDI freaks out. But then DVM' generally have little shielding.
Nice job.
BTW. You might be able to estimate the energy output of the CDI unit by the energy input.
I'm sure you are more capable in the math for that one.
I once saw a voltage divider-like setup using a series stack of well spaced zeners to drop most of the HV output and a final resistor to ground where you could measure voltage across the resistor to make a real measurement.
I never bothered making that (yet) either.
Hi Dave your FPGA driver for a CoilOverPlug Coil drives my
blown V8 with compression ratio ~>10 at over 7 krpm no problem.
Also Every Day Practical Electronics had an issue devoted to
spark energy. A whole! bunch of zeners and a pic chip to measure the
Yes my driver works well but as Terry pointed out (and I agree with) it's a lot nicer to be able to hide the ignition system - not easy with a full sized coil. I like to hide the ignition in the small wooden plate / box below the engine. That's also why I'm using the CDI units (for the most part). Terry's solution was really nice.
Yes I think that article is what I referred to concerning a real way to measure the ignition output. I'd like to build one of those. I need to find a cheap source for high power zeners (a couple of dozen as I remember it).
Not sure how to do the calculations on such a spike of power from a CDI though.

I once tried making the measurements you're asking about in order to determine just how much energy is needed to fire the plugs in a model engine. I first tried using a variable spark gap and the Paschen curve to account for the pressure inside the cylinder. However, this didn't include the effects of a wet fuel which I suspect are also important. My very shaky guesstimate from those tests came out to be somewhere between 3 and 4 mJ.

I later used a TIM-6 and a scope connected to its output through a voltage divider while running my Howell V-4 and graphically computed the energy in the output voltage waveform with help of its current waveform. As I reduced the coil voltage, I found the plugs continuing to fire below 2mJ and not showing signs of misfiring until somewhat below 1 mJ. Using the ignition's average current draw and making some efficiency assumptions gave me a similar result.

I didn't know how to handle the complex multi-spark waveform in Roy's (at that time, his 'old' style) CDI. I was able to trace out its schematic, however, and make some assumptions about the capacitor and it charging voltage. In those days, Roy also offered an 'off-menu' high rpm option that used a half-size capacitor. (I eventually used a pair of these CDI's in my 18 cylinder radial.) Making ratio'd comparative tests between the two versions of his CDI with their two different size capacitors, I eventually reached the conclusion that the minimum number was probably around 1 mJ which has been my assumption ever since. During all my testing, however, I forgot to include cold-start-up conditions. Of course, actual compression as well as the placement of the plug in the cylinder are also important variables.

I think federal safety guidelines assume .1 mJ is capable of non-pressurized ignition of volatile gases, and the one full-size Ford Kettering ignition that I analyzed back in the seventies happened to put out 8 mJ. I've read, but not personally verified, that some modern day full-size ignitions are using 75 mJ to help meet clean air regulations. - Terry
Thanks for your efforts and observations.
I assumed (as usual) you had done a thorough analysis of systems along the way. One of the other reasons they've gone away from CDI systems in cars is that a conventional coil when discharged gives a (relatively) long burn time as opposed to the very fast (albeit high voltage) spike of a CDI. In troublesome conditions the long energy release can be beneficial. I think this is also why the CDI systems went to multi-spark. Since the longer spark /arc was beneficial but not possible with a CDI the old saying "if at first you don't succeed...." so they give out multiple quick sparks. Better ??
Ford has what they call "a multi-strike system" using coils at low rpms where I suppose it might help cold starting.
As I mentioned I'd like nothing more than to use my driver and a COP coil but when the ignition system is bigger than the engine I don't like the looks of it so I've resisted.
That said George Britnell uses an external ignition box (some using my drivers) and you don't seem to notice it along side his wonderful workmanship. So maybe I should change my thoughts on spending so much on the S/S CDI units.
Thanks again for your observations.
The spark energy meter was described in the epe magazine february 2016 .
It used 30 zener 5watt diodes , wich cost abt half a $ each .
It looked quite an interesting circuit , so I was tempted to build it , but never got round to it .
I must have the article somewhere as a pdf should any potential builder like to take a look at it before purchasing the magazine .

Last edited:
Hi Terry, What's the size of those capacitors you use on the "Trigger Indicator Board"? You don't have them in your schematic. Thanks - Branislav
Looking at the wiring diagram, the hall device is powered when the PWR switch is turned on, Not switched on with the IGN switch, any reason for this? Just thinking out loud.
Looking at the wiring diagram, the hall device is powered when the PWR switch is turned on, Not switched on with the IGN switch, any reason for this? Just thinking out loud.

The PWR switch powers up the Hall device so I can see the trigger indicator without powering up the CDI which is what the IGN switch does. The Hall device remains powered when the IGN switch is turned ON allowing it to fire the CDI. - Terry
The head will probably be the most complex part on my version of Ron's Offy. It's eight multi-angled surfaces and arrays of internal passages require lots of patience and attention. My confidence in getting it right the first time was so low that I decided to run two parts in parallel. One, a 6061 'mule', will be used to verify my setups and test the risky operations before machining the 7075 'production' part.

A few changes made to the original design included shaving the head's bottom surface to accommodate a .020" head gasket. Modifications made earlier to the block to pick up additional gasket material around the cylinders have also altered the locations of several of the holes. The thread depth of the head mounting screws was increased a bit, and I added stock to the head's top surface for integral flanges that will increase the thread depth of the water pipe's mounting screws.

Ron appears to have used VR2L spark plugs with 1/4-32 threaded bodies. The documentation shows them mounted high up in the head with their electrodes not quite inside the combustion chamber. This appears to have been needed to keep a portion of the plug's wrench hex above the top surface of the head. There isn't enough clearance between the internal coolant passages for a recess around the plug to handle a standard socket. Smaller diameter Viper Z1's would eliminate the problem, but in the end I stayed with the VR2L's. I was able to un-shroud the plugs by using a smaller recess that will require a turned-down socket for plug installation.

Construction began by squaring up a pair of blanks with the head's outside finished dimensions. The conical combustion chambers and assorted holes for the head mounting screws and oil and water passages were the first features to be machined. The lengthwise internal coolant passages were later drilled and connected with four cross-drilled passages. The open ends of the cross passages were sealed with close-fitted Loctited aluminum plugs that were additionally secured with steel pins.

The exact angles of the 36 degree cam box deck surfaces are very critical as their accuracy will later affect the mesh between the camshaft and their drive gears inside the gear tower. To insure the precision of this machining step, a pair of 36 degree angle blocks were first machined. These blocks accurately cradled the head in the mill vise for these operations and provide a precise reference point for them. Since both deck surfaces are identical, the head was flipped, around and the exact setup and steps were used to machine both sides.

After completing and verifying both cam box deck surfaces on the mule, I managed to install my 'production' part upside down when I milled its second surface. Mules aren't supposed to be able to reproduce, but so far with a 60% completed head I now have two and counting. I hated to lose my first production head because I didn't have anymore 7075 material. - Terry

Work continued on the head with sealing up the two long internal coolant passages still open at its rear. As with the cross-overs, the ends were capped with Loctited aluminum plugs that were additionally secured with steel pins. Final machining blended the plugs invisibly into the head except for the pins' color differences

The two parallel sides remaining after the cam box machining were used support the head during its topside machining. Excess stock on the head's upper surface was used to add a few extra features. The drawings show Loctited inserts around the spark plug wells to simulate water shields. A portion of this stock was used to machine the shields directly into the head. I also added three flanges for the water outlet pipe that allowed me to raise its mounting screws out of the internal coolant passage running directly below it.

For some reason, during the machining of the combustion chambers on the bottom side of the head, I drilled the holes for the spark plugs. This was a mistake as I've made it a rule to machine each plug's mounting surface just before drilling and tapping its hole (with no deburring) in the same setup. I've learned the plug must be perfectly perpendicular to its mounting flat or its cylinder will be plagued with leaks. Fortunately, I was able to recover from the error since the upper and lower surfaces of the head were perfectly parallel. After re-registering the spindle over each hole from the topside, the plug well was machined and the hole tapped using a spindle tap holder.

A couple loose ends included tapping holes in the head's front surface for attaching the gear tower. As indicated in the drawings, the clearance holes for these mounting screws actually go through the centers of gear shafts inside the gear tower. The internal passages just inside the head's front surface for carrying oil to the top end were also drilled. Because of a ripple effect of the split crankcase, these had to be altered from those in the original drawings and angled past the just-drilled mounting holes in order to meet up with the relocated transfer port on the rear of the gear tower.

The last two surfaces to be machined were those for mounting the intake and exhaust manifolds. These surfaces are 96 degrees off vertical, and the axes of their ports entering the combustion chambers are at 112 degrees. The intake and exhaust plenums are mirror images of one another. The head was cradled in an adjustable angle block during the machining of the first surface, but with no orthogonal surfaces remaining afterwards, a pair of custom angle blocks supported the head during the second surface's machining.

This will be my first experience with a multi-valve multi-carb set-up, and the current plan is to use four O.S. 25LA carburetors with .21" diameter throats. Being more conservative than Ron, I reduced the diameters of the ports at their intersections with the valve cages from 5/16" to 3/16". Each port is blended smoothly into its plenum with a slightly oval intersection with its cage.

I would normally have installed the cages before machining the ports, but this head has been so problematic that I didn't want to wind up scrapping sixteen finished cages if I screwed up the head during one of its final machining steps. As it turned out, the third one made it to the finish line. Now, a fixture will have to be created to machine the openings in the sides of the valve cages before they're installed.

The last step was to plug all the openings and then bead blast and scrub clean the head's exterior to match the rest of the engine's simulated cast appearance. I'm not sure why it didn't occur to me earlier, but if I were starting over, an indexed 4-axis setup would have eliminated the need for so many unique setups. - Terry


Latest posts