My Hodgson 9 Radial Final Assembly

[jixxerbill]

I dont know about model engine carbs but I know plenty about Holley's.. I do know for sure that a carb set up for gas will not work for alcohol.. It takes almost two times as much alcohol compared to gas to go thru the carb.. Im only mentioning this wondering if the problems people have with using a carb designed for use with nitromethane and trying to run gas thru it. Just wondering.. Maybe the search for the RIGHT carb is actually a search for a carb that has enough adjustment to get the air/fuel mixture correct.. I have wondered about this since seeing a thread back a couple months ago...Bill

 

[RonC9876]

Terry: Glad to hear the engine wasn't seriously damaged and that you are getting it back to where it needs to be. I was thinking it might be better to hand crank the engine until you get more familiar with the fuel flow. You would be less lkely to do damage to any internal parts that way. I have had some experience with model airplane diesel engines and you never use any type of starter with those due to the danger of hyraulic lock. Of course it requires a good glove on your hand to protect the fingers from backfire. With spark ignition and a four stroke engine, backfire is not likely. You mght wear your arm out cranking though until the thing actually starts. Best of luck. Ron Colonna

 

[Lakc]

Full size rule of thumb is count 6 or 8 blades rotation by hand before using the starter. (3 or 4 blade prop). This basically means you rotate it twice by hand, but 4-6 times would be better. If you ever had the pleasure of hand propping a real radial over you can feel the bottom cylinders when they hit tdc.

 

[swilliams]

Thanks for your thoughts on the ignition Terry, they are very helpful.

Looks like better days with your engine are just around the corner
Good luck, I'm hoping it goes well for you

Steve

 

[mayhugh1]

Day2 ...
I tore the engine completely down and am now in the process of re-assembly. I did alot of thinking about what, if anything, I should do about the cheek clamp screws. I started thinking about the ugly possibility of tightening them so tightly that I broke one of them. I'm not a mechanical engineer, and so I'm learning as I go, but after a little research I was surprised to find out that common stainless steel screws are half as strong as those made of steel. I used 18-8 stainless steel 6-32 SHCS for the cheek screws in my mild steel crank mainly because I bought an assortment of them at a show and they look cool. I found the maximum recommended torque for 18-8 SHCS published at 9.6 in-lbs. This number goes all the way up to 30 in-lbs for a common steel SHCS. I borrowed a miniature torque wrench and ran some experiments in a 3/8" thick steel plate in which I had drilled and tapped a number of 6-32 holes. What I found was that for oiled threads all the 18-8's consistently broke at 30 in-lbs and all the steel screws broke at 50 in-lbs. The steel SHCS that I have on hand are military spec parts and they have a tiny fillet under the head instead of the sharp corner of the hardware store variety. I used the torque wrench to do some comparative measurements with the hex wrench I had used to tighten the front screw. I estimate that I was hand tightening the 18-8 stainless SHCS at 12-15 in-lbs. I replaced the rear screw with a steel screw and tightened it to 30 in-lbs. I also used blue (medium strength) threadlocker. I also did some experiments on both the purple and blue threadlockers and found that neither seemed to increase the break-away torque above the 30 in-lbs for a steel SHCS tightened to 30 in-lbs. They did add friction for much of the unscrewing process, however. For the front screw I bought a steel Torx SHCS because I had a long Torx driver that would fit through the crankcase to finally tighten this screw using the torque wrench inside the engine.
After assembly, the the crank turns freely in the front and rear main bearings but when the front cover is added there is still a slight bind. I have always had this bind even before the crash. I believe it comes from the fact that when the detail in the front crankshaft section forward of the front bearing is machined the shaft deforms from relieving stresses in the metal. Some of this detail is not symmetrical around the crank center axis and includes a keyway and a transverse flat needed for assembly clearance of the jackshaft. This bind is still about the same after my crash, and so I don't think the crank was tweeked during the crash. I thought about trying to straighten it - it's real tempting - but I finally decided to leave well enough alone.
After re-installing the rear spacer and impeller I temporarily added the rear section containing the distributor to verify the near zero backlash I originally had in the distributor drive. What I found now was some 10-20 degrees of backlash. After checking I found that the screw securing the magnetic disk and rotor to the distributor drive shaft was loose allowing the trigger disk and rotor to slip on the distributor shaft. There is no reason for this screw to loosen on its own. There is near-zero load on all the rotating components of the distributor. I suspect that when I originally timed the magnetic disk to the Hall device when I built the distributor over a year ago I did not finally tighten this screw after aligning it. There is a very good chance that the inertia of the disk and rotor allowed them to rotate and take the whole engine out of time sometime while I was trying to start the engine. It might have been when the engine first fired and tried to run but wouldn't sustain, or it may have happened when the engine was abruptly stopped by the hydrolock. Anyway, I'm glad I found it now because it would have been a big problem later. On the fronts of both of my other engines I engraved timing marks with which to compare with my flashing dwell led when the flywheel is turned over by hand before attempting to start them. I don't have these marks on this engine because of the prop and so I felt blind when it came to verifying the timing 'in the field'. I learned this evening though that comparing my dwell led with the position of the rotor as seen through my transparent distributor is just the thing I'm looking for.
When I originally timed the camshaft I marked the cam ring, the integral cam drive gear on the crank and the jackshaft gear with witness marks so I could re-assemble the engine, if necessary, without going through the trial and error process of centering the cam between the intake and exhaust lobes on the intake stroke. I lined up my marks and checked the centering and the timing agreed within 1 degree of my original value.

I'm also adding some construction photos that I had taken during the machining of my oil pump because if you're like me you like to see pictures instead of boring text.

- Terry

 

[petertha]

Once I get the engine running on methanol with the Super Tiger carb I will certainly give gas a try.... Terry

Terry, I wish I could offer something more tangible than sideline cheering & admiration of what you've accomplished thus far. But I'm confident you will solve the issues one by one in the same methodical manner you built it.

Not sure if you've seen this link or maybe the fellow visits this forum, but its a double row Hodgson. Looks like he selected an RC-ish looking carb for whatever reason, or at least for initial running.

https://plus.google.com/photos/111407870409657577971/albums/5278304464310065009?banner=pwa
https://plus.google.com/photos/111407870409657577971/albums/5311594071549777217/5592492649961509586?banner=pwa

I think Ive seen others too, so maybe there is more than one way to skin this cat. I must confess I've hydro-locked the odd fuel flooded inverted 4S RC engine in my day, especially with pressurized fuel systems. Sometimes with a bent rod or sheared crankpin to show for it if the starter battery or gear ratio was strong. Those engines are nowhere near as complicated to dismantle, so the procedure was remove the glow plug, spin it up & behold the shooting fuel spray geyser. I wonder if a starter assembly using a cordless drill could be adapted & use its adjustable clutch to 'spin out' at some set point you deemed below an unsafe torque level?

 

[mayhugh1]

Petertha,
You make a great point. My drill does have a variable torque setting. When I bought it I just set it for its maximum and forgot about it for normal use. It never occurred to me to drop the setting down to the minimum needed to turn the engine over. Thanks!!! - Terry

 

[t.l.a.r. eng]

Just thinking out loud, most if not all radial engines have updraft carburetors if they don't use fuel injection, and that prevents hydro locking for the most part. If you ever witness a flooded radial on the ramp at the local airport, you will see excess fuel running out the cowling, but not the exhaust.
I used an updraft on my 1/6 Kinner and it is virtually impossible to flood.

 

[mayhugh1]

Day 3...
The main bearings, the oil pump, rear seal, impeller, timed-cam, and front cover are now all installed and the crank once again turns freely. Next I'm re-installing the pistons and cylinders. When I disassembled the engine after the crash, I bagged the piston and cylinder/head assemblies together and marked them so they can go back into their original positions. Here is a photo of the tools I use to re-assemble the cylinders and the 72 small pattern 4-40 nuts, washers and lock washers back onto the studs in the extremely limited space I have to work. The object is to not drop a fastener into an opening in the crankcase. I borrowed the idea for one of the tools from Ken-ichi Tsuzuki's website.
http://homepage2.nifty.com/modelicengine/h9index.htm
He's the extremely patient guy who had to remake his crankshaft four times (and he just finished finally finished his engine a week ago). Its is a simple but remarkably effective tool to start the 4-40 nut onto the stud in the limited space around the stud. At it tip is a short 4-40 stud (enough to to capture two threads and hold the nut. The tool is held above the cylinder stud and then with a toothpick the nut is spun onto the stud and then tightened with a tiny 3/16" open end wrench. For the bottom cylinders the sump interferes with tightening a pait nuts on each cylinder and so I ground down a closed-end box wrench for those two nuts. Also shown in the photo is the socket I use to tighten the sparkplugs. In order to get clearance around the 10 mm sparkplug the socket has to be ground down to its very minimum diameter even with the extra clearance aI added to the head. It is also important when designing the stand that there is sufficient room for the ratchet and plug socket to be able to remove plugs 5 and 6. My stand supports the crankshaft 6.6" above the deck and this gives just enough clearance for my tool.
When I installed my spark plugs during my orginal assembly I used a dab of copper-laced anti-seize lubricant (sold for this purpose) on the threads of the spark plug. I've always used this on my full-size engines but it doesn't look like such a good idea here. Nearly all the heads ended up with some of this on the inside surface of the combustion chamber. I cleaned this off with a long reach Q-tip. I don't think I want it ending up embedded in the valve seats.
I re-assembled all the cylinders to the crankcase and while doing so I examined the rings with a jeweler's loop. Except for one piston, both compression rings show shiny wear patterns 360 degrees around them where the annealing oxide had worn off. The cylinders, themselves, are mirror polished but there is no sign of wear in the bluing. The one piston showed less than a 10 degree area of little or no wear, so fa,r on one of the rings. The oil control rings, however, all had two small 5 degree or so non-wear areas at the 5 o'clock and 7 o'clock positions with respect to the ring gap. Since these results were consistent among all the oil rings, this must have something to do with their construction. My lathe on which I bored the cylinders has a slight taper, and so when I bored the cylinders I made sure the larger diameter was at the bottom of the bore. Therefore, I expect the oil rings will take longer to seat because the diameter of the cylinder over a good bit of their travel is slightly larger in diameter than optimum; and, of course, the oil rings don't see much combustion pressure.
After installing the rockers and pushrods I re-measured the compression to compare with my original measuremens:
now previous
#1 71 psi 70 psi
#2 66 psi 63 psi
#3 67 psi 66 psi
#4 64 psi 66 psi
#5 70 psi 77 psi
#6 72 psi 80 psi
#7 75 psi 72 psi
#8 65 psi 72 psi
#9 71 psi 72 psi
The largest changes seem to be in cylinders which took a real beating with fuel flooding during the 'first pop'. I decided to not change the rings at this time but instead to wait to see if they improve after the engine is actually running. I have three spare head/cylinder/piston assemblies as well as a number of spare compression rings.
I'm stopping with any further assembly at this point. I'm making arrangements to get access to a scope to properly examine the operation of my ignition. I'm still bothered by the low dwell that I chose, and I may make a new trigger disk. But, I want to look at my primary coil current and secondary arc-burn waveforms with a scope to help me decide before I add the rear section, distributor, and wiring. This scope has the capability of recording waveforms and dumping them to a flash drive as a .jpg, and so I hope to include them in my next post.

 

[Swede]

A few thoughts - carbs designed for glow fuel will work with gasoline, but some caveats must be kept in mind.

Most of us at one time or the other in our lives have operated a glow engine on an RC plane or similar. They behave well, and are generally easy to start. Importantly, the mixture needle allows operation over a pretty healthy range, usually a turn or three. When used with gasoline, the needle valve becomes much more sensitive, and the range in needle valve clicks or motion narrows considerably.

That was one of the problems I faced when experimenting with a new/different glow carb and gasoline fuel... where to set the darned needle valve. If the needle is over a turn off from where it needs to be, the engine simply won't sustain, and the standard glow procedure of getting a rough run, then tuning the mixture for best running by ear, doesn't work as well with gasoline.

Probably the easiest method would be to open the valve two turns or so, prime, and crank. If the engine fires on the prime, then dies, open the needle another 1/4 turn. Try again. Eventually, it runs a bit more than just the prime fuel. When this happens, you're getting close. begin working with smaller increments, like 1/8 turn. Eventually, the engine sustains, and more often than not, just a few clicks of the ratcheting needle valve will find the sweet spot.

One good part I've found - if your fuel pressure is constant, such as from a bowl, the needle valve will hardly ever need to be touched again. Much better behaved than when using glow fuel. I don't think I've messed with my needle valve more than two or three times since I found the sweet spot.

Oil - the H9 oil pump is extremely effective. Now what I am about to say is pure opinion, so take it with a grain of salt - the scavenge pump, being larger, should in theory do its job. But if the metering to the pressure pump is open too much, there's too much oil being delivered... the path that the used oil takes to get to the scavenge tank is very tortuous, and what happens is that oil rapidly accumulates in the front cover area, and in the lower cylinders. During some early runs, I found oil being blown out the front ball bearing, just behind the prop. Waaay too much oil, and as you have seen, you can hydraulically lock the engine.

Now, I run with my oil supply needle valve open only 1/2 to 3/4 turn or so. In fact, here is how I start and run my engine w/regards to oil, and guys are going to cringe, but it works.

Engine is cold.

1) Open oil needle supply valve 4 or 5 turns.
2) Switch off. Hand prop a dozen turns or more. We're pre-oiling a bit.
3) CLOSE the oil supply valve.
4) Prime the engine.
5) Crank it. Yes, we start with the oil supply OFF.
6) Once idling, open the oil needle valve 3/4 turn.

More open than that, I notice smoke, sometimes significant, from the bottom exhausts. This tells me I've got too much oil entering the engine.

I've been running this way for years, with no noticeable wear. When I miked a cylinder after this time, the wear was close to zero. My cylinders are also 12L14.

Every engine is different. Maybe my oil rings aren't as good as yours, maybe my pumps are different. My point is that the design seems to run fine with less oil than our instinct wants to give it. If you use transparent plastic lines for pressure and scavenge, you can see the oil flowing, and rather than seeing solid oil in the tube, I am seeing oil, then air, then oil, or a frothy mix of air and oil.

This engine (at least my example) does not get very hot. I've been running it a bit leaner than I normally do lately, and have been rewarded with hotter cylinders and heads, especially the lowers, which pleases me. I found that a larger prop also improved running and heat characteristics.

As you're discovering, it can take some time to get the engine running. May I respectfully suggest skipping the drill starter. Wear a leather glove, and hand prop. It's more labor intensive, but when dialed-in, the engine starts with 2 to 3 hand flips.

Good luck with it! You'll have it running soon. One last thought - consider a large external battery for these early attempts... it removes one variable from the equation, namely, weak current and voltage to the ignition.

 

[Swede]

For reference, here is the carb I am currently running with... it is a Super Tigre carb, but I don't remember the exact model. Off a .30 or so glow engine. It looks ridiculously small, but works very well.

Over the years, I have measured full-throttle RPM with various ignition and carb schemes, and with the original prop, which was way too small, BTW, the highest I saw was 6,200 RPM. That was using a larger carb and hall-effect ignition. But mid-range and idle suffered... there was not adequate air velocity at idle to properly feed/atomize the fuel. By going with a smaller carb, I lost about 400 RPM off the top, but it idles nicely.

This is your thread... I don't want to distract from it, but some of the issues you are having were identical to mine. In the end, you know what is best for your engine, and how you want to operate it. You've done a tremendous job with it, and I especially like how you packaged it all into a neat, tidy setup.

 

[PeterB]

This is amazing!!!
I can feel my breathing frequency change when I'm looking and reading your posts :bow::bow::bow:

 

[mayhugh1]

I was able to get hold of a Tektronix TDS2012C digital scope and a compatible current probe. This scope is capable of recording one-shot events and is nearly ideal for what I am trying do. I made a capacitive coupler for one of the x10 probes in order to sample the voltage signal from the high tension secondary winding of the ignition coil during an actual sparkplug firing. I experimentally sized the coupler to give a reasonable amplitude without damage to the scope or probe. The distributor is electrically connected to my radial engine's TIM ignition, but physically it is currently unattached to the engine. In these measurements I'm not passing any high voltage through the distributor. I have a sparkplug connected between the ignition coil secondary and system ground. The plug will fire nine times per revolution of the distributor rotor. The only part of the distributor I'm using in these tests is its magnetic trigger disk and Hall sensor. In order to capture the waveforms I want to study my plan is to arm the scope and just spin the pinion gear of the distributor with my finger while the ignition is powered. The current probe is around the lead going to the primary winding of the coil and will measure the current waveform of the primary winding. I'm using a third, ordinary, x10 scope probe to capture the coil primary voltage waveforms. I'll trigger the scope data collection with the rising edge of the coil primary voltage or current. This scope is capable of displaying only two channels at a time but can record and save them to a USB flash drive. - Terry

 

[Lakc]

It takes a very brave friend to loan you a scope to work on potentially damaging voltages.

 

[cfellows]

Terry, I hadn't seen this string of posts before and didn't realize you were so close with the hodgson. All I can do is shake my head at the work you've done. Hope to see you at Rudy's tonight!

Chuck

 

[mayhugh1]

The first pair of waveforms that I captured is the coil primary voltage and current. It takes a little trial and error to capture the information that I'm looking for because, being manually spun, the rotor rpms are constantly changing as it spins up to a high speed and then spins down. And, I want a time scale that optimally shows the waveform detail while the sensor is in dwell. "TEK0003.BMP" shows the coil primary voltage in green at the top and the coil primary current in yellow at the bottom. When the output TIP42C transistor switches ON the voltage across the coil's primary instantaneously switches to just over 5 volts. As can be seen later in TEK0000.BMP the leading edge of this voltage waveform appears in the secondary voltage waveform transformed by the coil turns ratio into a harmless low energy glitch. In a Kettering ignition with mechanical points this glitch would not occur because the capacitor across the points would slow the rise of this edge so that it would not be seen by the coil. The coil current rises exponentially toward its maximum which, in this case, is about 4.4 amps. Dividing the coil voltage by the coil current at the end of the dwell, after the effects of the inductance have died out, gives the coil resistance which, also measured independently with an ohmmeter, is about 1 ohm. The exponential current rise gives me an opportunity to finally calculate the primary inductance of my 'old style' Exciter coil. The time constant of this rise is the time it takes for the current to reach 63% of its final value. From the current waveform this time measures to be about 1.8 ms. The time constant for a series LR circuit is tc =L/R and so L = 1.8ms x 1 ohm = 1.8 mH. The time for the coil to reach its maximum (for all practical purposes) current is 4-5 time constants, and from this waveform we see this takes about 6 ms. We can also calculate the energy stored in the coil at the end of the dwell period as E = 1/2(L * I^2) = (1/2)(1.8mH)(4.4^2) = 17mJoules. This will be important later because this is the energy that will be available to the sparkplug to ignite and burn the air/fuel mixture in the cylinder. This will be a key parameter that will be used later to determine my dwell requirements. We can also calculate the power dissipated in the TIP42C during the dwell period. In my circuit this is (.6V vce x 3.5 amps avg) = 2.1 watts (I'm roughly eye-balling the average dwell current). I'll correct this dissipation for duty cycle as a function of engine rpm later. There is also a series diode in the emitter of the output transistor of Jerry Howell's TIM design whose only job is to limit the coil voltage and take some of the dissipation away from the output transistor. I need to keep an eye on the dissipation of this diode also. This dissipation = .75 Vd x 3.5 amps avg = 2.6 watts and, again, I'll correct the dissipation for the duty cycle later.
When the output transistor switches OFF at the end of the dwell period, the coil primary voltage spikes to about -26 volts as the coil tries to continue the primary current flow. This is the so-called flyback voltage and its value depends upon how quickly the output transistor switches OFF. To a first order, the amplitude of this spike can be as high as L x (delta I)/(delta t) where delta t is the switching time of the output transistor. Plugging in values for L, delta I and .5 us for the switching time for the TIP42C and ignoring strays we get a maximum of almost 16kV! However, this voltage is clamped to a much lower value when the plug fires. Without an sparkplug connected to the secondary it is this high voltage spike that destroys a solid state ignition when the flyback occurs. What is the value of this clamped primary voltage? To find this out we must use an important physics work called the Paschen Curve which is a curve of the breakdown voltage of an air gap as a function of the gap length and pressure of the gas surrounding it. Plugging in .018" for my spark plug gap, and 14.7 psi air pressure into this curve gives 2500 volts for the plug firing voltage. This means that the plug will fire as soon as the secondary voltage rises to 2500 volts. Dividing this initial firing voltage by the coil turns ratio of 100, gives 25 volts which is very close to the observed 26 volt primary clamped flyback. Assuming a compression ratio of 5, a resulting cylinder pressure of 73 psi, and the same plug gap the Paschen curve predicts a plug firing voltage of 8kV. Dividing this by 100 gives a primary clamped flyback voltage of 80 volts. The Vce and Vcb breakdown voltages for the TIP42C are 100 volts and so we are safe but with little margin.
The coil current drops to zero when the output transistor is switched OFF, and this marks the end of the dwell time. There is some additional structure in the primary voltage waveform that occurs after the end of the dwell. This 'noise' is coupled back through the transformer from the high voltage secondary to the now high impedance primary and is created by the electrical storm going on in the plug gap.
I selected this waveform pair out of many that I captured because they represent an engine rpm where the dwell is such that the coil just reaches its maximum possible current. If the engine is running faster, the coil current will peak out at a value limited to less than its maximum possible, and the energy available to burn the fuel in the cylinder will be lower. If the engine is running slower, the current will not be significantly higher; and the net result will be the output transistor and the coil run warmer.
The waveform "TEK0000.BMP" shows this effect. Here, the top green waveform is now the secondary voltage. Although the vertical scale says 50V/div this is relatively meaningless because of the the uncalibrated capacitive coupler that I made to pick up the high voltage secondary. We can estimate the attenuation of my coupler by assuming the actual peak secondary voltage is the Paschen value of 2500 volts. Since the scope is recording this spike at 155 volts, my coupler is therefore attenuating the signal by a factor of 16. The bottom yellow waveform is the coil primary current as before. In this waveform, the distributor rpms are rapidly slowing down after being manually spun up; and we can see two adjacent primary current pulses. In the first pulse the coil has enough time to reach the full 4.4 amps, but the second pulse has only enough dwell time to reach only 4 amps. (It seems to be common jargon to refer to the maximum possible current that the coil can reach as the 'saturation' current; but technically the flux density of the core is not actually reaching its saturation point and so I just refer to it as the maximum possible current.)
The large positive spike in the green secondary waveform when the output transistor switches OFF is, of course, the voltage that fires the plug. As explained above, this voltage will rise to whatever the cylinder conditions require (up to the max possible flyback voltage times the turns ratio) to jump the plug gap. The air/fuel mixture is NOT instantaneously ignited at this time. Just after the plug fires the voltage waveform quickly falls to about 150 volts which is the voltage needed to sustain the arc across the plug gap. The voltage remains at this value for about one ms while the total stored energy in the coil (the mJ we calculated before) is dissipated in the plug gap. It is during this time, called the spark time or spark zone, that the air/fuel mixture starts to burn as the energy necessary for ignition is transferred to the mixture. This spark time continues until the coil runs out of energy to sustain the arc in the gap. Hopefully the mixture is explosively ignited before the end of the spark zone. The time it takes after the plug first fires to transfer enough spark energy to the mixture to cause it to explode is dependent upon a number of complicated factors; but this delay is the reason why we typically have to advance the timing in an engine (even our small unloaded models) to get them to run optimally. Statically timing an engine so the plug fires at TDC means only that the plug will instantaneously fire at that time. Additional time is required for the coil to transfer sufficient energy into the mixture to reliably light it up. And this happens during the 1 ms interval after the initial plug firing. In a V-8 running at 5000 rpm this 1ms spark time is equivalent to up to 30 degrees of timing.
The green secondary voltage waveform in "TEK0000.BMP" shows a relatively large negative spike after the arc extinguishes and then a noisy decay back to zero. What is happening here is that when the coil current is once more interrupted but this time by the extinguished arc and another flyback spike occurs, this time in the secondary, as the coil tries to maintain the sparkplug current. In a Kettering system with mechanical points and a capacitor this shows up as another nicely damped sinusoid as the remaining coil energy oscillates back and forth between the coil and the capacitor.
The final goal is to put all of this together in order to come up with a value for the dwell time. This dwell time will determine how much energy is required to reliably ignite the air/fuel mixture over a reasonable range of engine rpms and cylinder conditions. To do this we must know, after the plug fires, how much coil energy is required to start and continue the mixture burn. There have been several white papers written on this subject especially over the past 40 years when the auto manufacturers have been interested in cleaning up engine emissions. What I have found is that a minimum of 0.2 mJ is capable of reliably burning fuel at a stoichometric ratio of 14:1. For very rich or very lean mixtures as much as 3 mJ may be required. The high energy ignitions in modern full-size engines can produce as much as 70 mJ or more, but those high values are required to reduce emissions under demanding cold start and high load conditions. Supercharged race engines use ignitions capable of producing more than 100 mJ. For my model radial I'll likely set an energy requirement of 1 mJ or so. I hope my next installment will have the final data to tell me if my present trigger disk is adequate. - Terry

 

[Lakc]

Its nice to have the right toys, its downright frustrating to use my old analog scope with a broken trigger circuit. Nice writeup, thanks.

 

[mayhugh1]

From the waveform measurements I compiled two tables - one for a 10 (distributor) degree dwell trigger disk and one for a 20 degree dwell trigger disk. My radial currently has a 10 degree trigger disk. Each table shows the spark energy, the output transistor dissipation, and the coil dissipation as a function of engine rpm. I realized that I have not yet shown a schematic of the TIM ignition and so I am including a sketch of it also. This circuit was first published in the June/July 1992 issue of Strictly IC magazine and has been successfully used by many model engine builders. Jerry Howell added the series diode in the emitter of the output transistor in order to reduce the coil current so that a 6 volt battery could be used for a power source. This exact circuit is in both my V-twin and V-4 engines. (In fact the V-twin uses two of them instead of a distributor.) My tables also include the power dissipation of this diode as a function of rpm. From our last discussion a 1mJ spark energy should be adequate to ignite an air/fuel mixture that this engine is likely to encounter (once the carburetor problems are sorted out). The tables show that either 10 or 20 degrees of dwell should be capable of running this engine beyond 5000 rpm with the 20 degree dwell providing some healthy margin. And so, the component dissipations will be considered next.
Ignoring, for a moment, the issue of zero rpm i.e. if the engine is stopped with the Hall device triggered ON, the heat-sinked output transistor and the dropping diode dissipations are within reason. Things get a bit toasty for the non-heat-sinked diode at idling and mid-range rpms with the 20 degree dwell, however. The coil dissipations are markedly higher with the 20 degree disk. The coil is rather physically large, but the epoxy potting likely has a high thermal resistance, and so the core and windings will likely see high transient thermal stresses that will affect the reliability of the coil.
And so my conclusion is that I will, for now, stay with the 10 degree disk that is currently in the distributor. My engine will likely spend most of its time between 2000 and 4000 rpm and this seems to be a sweet spot for this ignition combination at 10 degrees of dwell (and assuming all this theory is correct :wall.
The real concern is being careful to not allow the engine to sit still with the coil current ON. With my brief experience with this engine so far, the compression bumps tend to cause it to settle in a position with the coil current OFF. The diode and coil seem to be the biggest at-risk components. I am going to look at replacing the diode with a higher current version or adding some kind of heatsink.
It would seem that ideally this circuit could be further optimized for this engine by adding a second dropping diode and then changing the dwell to 20 degrees to compensate for the drop in current caused by the second diode. The second diode would divide the dissipation between two devices and it would limit the coil current at zero rpm with the output transistor ON. The increase in dwell would compensate for the lower maximum coil current at higher rpms.

One last comment concerns the gap between the rotor tip and the high voltage tower contacts of the distributor. The tip of the rotor generally doesn't actually touch the tower contacts as this could leave metal particles in an area where they can cause a lot of havoc. And so the distributor must designed so there is a gap between the rotor and the tower contacts. It is important to make this gap as small as possible because there are two gaps to which the ignition must supply energy. The energy in the spark zone will be divided between the rotor gap outside the engine and the sparkplug inside the engine. The rotor gap energy does nothing to help burn the air/fuel mixture in the cylinder. It just erodes the rotor and tower electrodes over time and widens the gap to make the problem even worse. My plug gaps are about .018" and so my design goal was to have a maximum rotor gap less than .002". Jerry Howell's otherwise great distributor was designed to use the Satra distributor cap which uses metal rivets for contacts which stick down vertically from the towers. The rotor must be designed so that its flat top surface rotates below these rivets. The problem with this design is that it is difficult to maintain a consistent vertical clearance between the rotor and the tower contacts. This clearance is affected by the bearing scheme used for the distributor shaf, but even more important it is affected by the run-out of the meshed pinion gear set driving the distributor shaft. A much better design for the cap is the one used by full-size distributors back in the days when distributors were used. In these caps the tower electrodes extend down from the tower insulators at the periphery of the cap. After the electrodes are pressed into the towers, the cap is put in a lathe (or on a mill and rotary table); and the electrodes are bored out to about half their diameter for a depth sufficient to clear the total height of the rotor. The diameter of this bore is equal to the diameter of the rotor. In this scheme it is not important to maintain a consistent or non-varible vertical clearance of the rotor since the rotor gap is now horizontal instead of being vertical. It is only the run-out of the rotor and the centering of the tower boring operation that determines the maximum gap. I made my cap in this manner and included a registered periphery to insure consistent mounting to the distributor base.

The next step is to finish the assembly and try for a 'second pop.' - Terry

 

[mayhugh1]

I've included the Paschen curve I used in my previous post. I forgot to add it to my previous discussion of the dwell requirements for my ignition.
I finished the re-assembly of the engine, timed the distributor, and made sure the trigger disk screw was tight this time. I've decided to use the Super Tiger carb for this second start attempt since I think my chances for success are higher. I'm also going to start with a 15% mix of methanol and Crowne camp fuel since I will be safely outdoors. The methanol I'm using is from an Auto parts store. It is sold as a fuel line dryer called 'Heet'. The yellow bottles are nearly pure methanol while the red bottles are nearly pure rubbing alcohol. If I can get the engine running to a point where I'm satisfied with it I'll then try gasolene in this carb per Swede's comment. I hope to return to the Walbro later. The carb that I have is a Super Tiger model 12163146. It was salvaged from a second hand RC helicopter that my son didn't have much success in learning to fly. It was designed for high revving glow engines with displacements between .60 and .90 cubic inches. This sounds like an ideal carb for this engine but it this carb was designed for engines that rev up to 16,000 rpm. As others have pointed out in this thread, a carb this size is likely a bit large for this radial which has a per cylinder displacement of just under .9 cubic inches and maximum rpm of 5k to 6k rpm. This carb is spec'd with an internal throat diameter of .41" but I measure a venturi diameter of .35".
So I decided to pre-set the needle adjustments to the lean side before mounting the carb to the engine. To do this I connected a length of clean plastic tubing to the fuel inlet on the carb. With the throttle wide open I blew into the other end of the tubing and adjusted the high speed needle while listening for air flow through the throat of the carb. My goal was to find the point at which the air flow stopped and then open the needle 1/2 turn as a starting poin for my radial. However, I could not stop the air flow no matter how far in I turned the needle. I couldn't see any damage to the needle itself and so I don't know if this is the wrong needle or if the carb was designed so the high speed mixture cannot be completely turned off. I performed the same test with the low speed needle but this time with the throttle nearly closed. I was able to then set the needle 1/2 turn open from the point that the air flow ceased.
At this point I'm beginning to remember that the engine on that helicopter didn't actually run all that well, and so I went down to the local hobby store to buy another carburetor. The only one they had in stock was a Super Tiger 12163145. It is designed for glow engines with displacements between .40 and .51 cubic inches. It has the same .41" internal throat diameter as my salvaged carb but, for some reason the outer diameter of the throat is .050" smaller than the carburetor around which I designed my adapter. So I made a .025" brass bushing for my adapter so I wouldn't have to make a new one. With this carb I was able to completely shut off the air flow using the high speed needle with the throttle wide open. In fact, I tried this needle in my first carb and it also shut off the air flow in that carb. This makes me think the stock needle on the first carb was designed to always allow some mimimum amount of fuel flow past the high speed needle. This is probably something I really don't want for this engine. With the throttle fully open I opened the high speed needle 1/2 turn from the point at which I could here no air flowing through the carb. With the throttle nearly closed I opened the idle needle 1-1/2 turns from the point where I could here no air flowing when I blew softly into the tube. I marked both needles with dabs of paint so I can keep track of their settings.
Next, I attached the bowl to the carb and hooked it up to my fuel pump before bolting the combination onto the rear of the engine. My last experience has made me overly sensitive to potential flooding problems and so I wanted to make sure I don't at least have this problem. Unlike the Walbro, in this set-up the fuel pump just pumps fuel into the bowl which tries to maintain a constant level with the excess being returned to the tank through a return line. The level in the bowl is set by the height of a vertical return tube soldered to the bottom of the bowl. I moved the bowl to its lowest position on the carb adapter to reduce any chance of flooding and this position sets the fuel level .25" below the carb spray bar. Unfortunately when the fuel pump was turned on and the carb primed the carb flooded. with no air moving through the venturi. The fuel pump has a rheostat so I can adjust the voltage to the pump, but I had only a very narrow range of voltage around three volts where the pump would pump without flooding the carb. I had not done much testing on this set-up since I expected to be using the Walbro with the pressure regulating bowl that I had designed and thoroughly tested. This Super Tiger was pretty much an after thought and a very distant plan B.
The cause of the flooding was the turbulent high speed flow of fuel into the carb was causing the fuel level to rise all the way to the top of the carb in one single corner of the bowl. And this corner happens to be where the outlet to the carb is located and so the fuel level seen by the carb was above the spray bar. I think the carb was probably even seeing pressurized fuel when the fuel in this corner hit the lid on the bowl. This was happening even though my fuel line to the bowl contains a .022" restricter. I tried adding various baffles to the bowl to try to drop the fuel level in this corner, but I could not find a solution in the small volume of the bowl. Eventually I shortened the height of the drain pipe inside the bowl. This allowed me to get a non-flooding pump zone over a range of 3 to 5 Volts pump voltage - not perfect but good enough. - Terry
　

 

[Lakc]

I appreciate all the work you have put into the math involved regarding ignition systems, its very timely.