A short study on ignition timing and combustion

Are we confusing static squish/quench clearance with running clearance here? When people talk about setting squish clearance on a Commando, most are talking about the static clearance, i.e. with the engine on the bench and not running. Allowing for crank flex, rod stretch, and piston rock, as well as expansion at operating temperatures, the actual clearance at high rpm is going to be well less than the static clearance, particularly for those who think red line is wherever the valves float. I normally set Commando race engines at .040" clearance, but that's just my preference. That's a number I got from Axtell back in the '70s for standard stroke 750 engines with modified stock crankshafts and rods. I've run that configuration engine as low as .035" without problems, but I like the safety margin of .040". Like Jim Comstock says, the safe limit depends on your engine build (as well as how much you plan to abuse it). In the days of one piece crankshafts, steel or titanium rods, a variety of bores and strokes, long and short rods and pistons, one size probably doesn't fit all, at least if you're trying to push the limits.

I think the engine Jim Schmidt is referring to that broke the cases at less than .040" clearance was the one I rebuilt for Martin Adams' Commonwealth Norton back in '86, after it blew up at Laguna Seca. It was an 850 overbored to 79 mm, with Nourish one piece crankshaft, Crower long (6.200") titanium rods, and Omega pistons. Martin had insisted the original engine builder put it together at .020" squish, because he thought the crank and rods wouldn't flex and stretch as much as stock parts. He loaned it to a friend of mine to race at Laguna Seca, and he broke the cases big time on the track. When I tore it down, it was clear where the pistons had been hammering the head. Martin insisted I put it back with .020" clearance, so I agreed, but set it at .040" anyhow. The engine ran great at Daytona in '87, and Martin was happy. It was the only race the bike ever finished. After the race I told him about the .040" clearance, and he was so pleased to have a result that he wasn't upset about it. Years later I acquired the bike from Martin, and when I tore the engine down there was no sign of the pistons touching the head.

Fascinating stuff for gearheads. I've never tried seriously experimenting with tighter squish clearances like I'm sure JIm Comstock has. The results of guessing wrong kept me from it, but if you are really looking for the last bit of horsepower, you'd presumable want to get it as tight as possible. The difficulty is finding that limit without some expensive failures. The cost of finding that last bit of horsepower can be pretty high. When I was racing, I eventually realized that I'd get more return from spending money on track time than I would from trying to get just a little more horsepower out an already reasonably well developed engine.

Ken

With Carrillo rods, full skirt pistons and aluminum barrels I could set the squish somewhere between .010 and .012 when measured above the pin.
With hard racing that would just start to leave signs of contact. With 12.5 compression and mag timing at 27 degrees the power would increase a couple horse just about the point there was almost contact.

I raced a couple times where I had more than light contact but never suffered case damage because of it. Pinched the upper ring land once however. Jim

Indeed Ken, whenever we set squish clearances statically we are only really guessing / compensating for the vagaries of expansion, stretch, flex, etc.

Norman White told me that when he worked for HRC on the big 4 cylinder racers, they set the static squish to zero! But I guess that makes sense as they had super no. Flex cranks, titanium rods, lightweight pistons and an alloy alloy top end, so the gap would be bigger when hot. I know on his racing Commando (with light pistons and titanium rods and possibly alloy barrels) he set it below .020".

I raced Norton cranks in Triumph engines and as my notes aren't to hand I can't be totally precise, but I learned that at / close to .030" clearance, the pistons would touch, not badly, but touch nonetheless. At more than .060" I'd get carbon build up on the squish area, indicating that it wasn't working correctly. So I ended up aiming for .045" (which in reality meant anywhere between .040" and .050").

As my Commando has light pistons and steel rods, I ended up being under .040" so will see what's what when I remove the head (sometime soon).

One thing I don't like about the composite head gaskets is you have to guess the amount its going to crush, I'm quite uncomfortable about this as you don't really know what you've got until you strip and measure it, but then you have to use a new gasket, so you're back to square one!

Do you know if the composite gaskets crush a reliable and fixed amount every time?

We learn something every day on this great site! It amaze's me that the petrol/air "bomb" explodes so SLOW! The crank/piston swings through 50-60 degrees quicker that the Bomb explodes completly! ie the spark starts at 29 Degrees before TDC and the the max pressure happens 15-30 after top.
Most American gun users would consider that "Damp powder burn" I would have thought a compressed and heated 15-1 fuel/air mix to be micro seconds burning? I suppose Burn is the correct term and not exploding. Great post and i like these in depth replies.

[video]http://www.youtube.com/watch?v=sEf8va1S7Sw[/video]

Jim.
Thats just amazing! Not what i would have imagned,,,,,the flame looks "bity" not the sudden blast i expected..Many thanks for posting.

comnoz said:

[video]http://www.youtube.com/watch?v=sEf8va1S7Sw[/video]

Click to expand...

john robert bould said:

We learn something every day on this great site! It amaze's me that the petrol/air "bomb" explodes so SLOW! The crank/piston swings through 50-60 degrees quicker that the Bomb explodes completly! ie the spark starts at 29 Degrees before TDC and the the max pressure happens 15-30 after top.
Most American gun users would consider that "Damp powder burn" I would have thought a compressed and heated 15-1 fuel/air mix to be micro seconds burning? I suppose Burn is the correct term and not exploding. Great post and i like these in depth replies.

Click to expand...

There is a difference between gun propellant and high explosive. With HE you are only interested in detonation. With gun propellant the rate of pressure rise is important (brisance). The action of changing to a higher octane fuel appears to have a similar effect as having different brisance. What you usually find is that the burn doesn't happen at a constant rate, more like an acceleration. The measure is rate of pressure rise divided by the time. I think you would find that with all other things being constant, the burn time in a combustion chamber stays fairly constant regardless of the revs, that is the main reason you need an advance curve to get peak pressure at the same piston position regardless of the revs. There are two things to consider when looking for acceleration out of corners - the advance curve and the shape of the needle in the carburettor - balancing act ? As I've previously suggested a more sensible way to go might be to calculate a curve from the geometry and estimated burn time, then jet to it as a fixed algorithm. If you are playing with two variables at once you are in the middle of nowhere.
The burn time can be estimated by setting fixed advance at 29 degrees, and thinking about the degrees ATDC where max. pressure should be considering the leverage aspects of the internal geometry.
The only thing left to play with then would be the jetting.

john robert bould said:

We learn something every day on this great site! It amaze's me that the petrol/air "bomb" explodes so SLOW! The crank/piston swings through 50-60 degrees quicker that the Bomb explodes completly! ie the spark starts at 29 Degrees before TDC and the the max pressure happens 15-30 after top.
Most American gun users would consider that "Damp powder burn" I would have thought a compressed and heated 15-1 fuel/air mix to be micro seconds burning? I suppose Burn is the correct term and not exploding. Great post and i like these in depth replies.

Click to expand...

There is a difference between gun propellant and high explosive. With HE you are only interested in detonation. With gun propellant the rate of pressure rise is important (brisance). The action of changing to a higher octane fuel appears to have a similar effect as having different brisance. What you usually find is that the burn doesn't happen at a constant rate, more like an acceleration. The measure is rate of pressure rise divided by the time. I think you would find that with all other things being constant, the burn time in a combustion chamber stays fairly constant regardless of the revs, that is the main reason you need an advance curve to get peak pressure at the same piston position regardless of the revs. There are two things to consider when looking for acceleration out of corners - the advance curve and the shape of the needle in the carburettor - balancing act ? As I've previously suggested a more sensible way to go might be to calculate a curve from the geometry and estimated burn time, then jet to it as a fixed algorithm. If you are playing with two variables at once you are in the middle of nowhere.
The burn time can be estimated by setting fixed advance at 29 degrees, and thinking about the degrees ATDC where max. pressure should be considering the leverage aspects of the internal geometry.
The only thing left to play with then would be the jetting.

Jim, About the throttle position sensor. If you are playing with more than two variables you get into the realm of pattern recognition and mathematically comparing data sets. Unless you can automatically gather the data, you have a much bigger problem to consider. Most of us cannot imagine anything more than 3 dimensions. The mathematical situation with the number of variables involved in road racing a bike is a multivariate space. A lot of the maths on pattern recognition was done by Jurs and Kowalski at MIT in the 1970s. I'm too stingy to buy their book.
http://ww2.odu.edu/sci/tisenhou/homepage/RESCOMP.HTM

Have you read anything on 'games theory' ?

acotrel said:

Jim, About the throttle position sensor. If you are playing with more than two variables you get into the realm of pattern recognition and mathematically comparing data sets. Unless you can automatically gather the data, you have a much bigger problem to consider. Most of us cannot imagine anything more than 3 dimensions. The mathematical situation with the number of variables involved in road racing a bike is a multivariate space. A lot of the maths on pattern recognition was done by Jurs and Kowalski at MIT in the 1970s. I'm too stingy to buy their book.
http://ww2.odu.edu/sci/tisenhou/homepage/RESCOMP.HTM

Have you read anything on 'games theory' ?

Click to expand...

A little bit some years ago.

With the TPS input you are generally looking for better strret economy and a little better mid throttle response. Generally to use it you would build a curve at no load and another curve at full load and then use the TPS input to graduate between the two curves. Jim

Are you suggesting there should be a load cell sensing the torque reaction ? It might be possible to change the front isolastic mount to an oil filled cell with pressure sensor, and feed that input along with the throttle sensor in to the computer in the ignition module, and adjust the variables seeking a maximum.

acotrel said:

Are you suggesting there should be a load cell sensing the torque reaction ?

Click to expand...

No????

No Load = engine running through the rpm range with no load other than the drivetrain and tire to road loss. Engine speed is controlled by the throttle position -slowly changing speed either up or down

Full load = wide open throttle -engine speed is controlled by the dyno brake as the engine speed is run through it's rpm range -slowly changing speed either up or down

john robert bould said:

We learn something every day on this great site! It amaze's me that the petrol/air "bomb" explodes so SLOW! The crank/piston swings through 50-60 degrees quicker that the Bomb explodes completly! ie the spark starts at 29 Degrees before TDC and the the max pressure happens 15-30 after top.
Most American gun users would consider that "Damp powder burn" I would have thought a compressed and heated 15-1 fuel/air mix to be micro seconds burning? I suppose Burn is the correct term and not exploding. Great post and i like these in depth replies.

Click to expand...

Ignition Speed Potpourri - more thoughts on the slow nature of the combustion event.

Just as Comnoz has talked about kernel size, and the growth thereof, to describe the initiation of the combustion event, other technical literature dissects the phases of combustion by plotting cylinder pressure against crankshaft angle (as the link hobot provided shows) and divides the combustion event into the following time periods.

Burning Angle - The period between ignition spark and peak cylinder pressure
Lag Angle - The period between ignition spark and a rise in cylinder pressure that exceeds the engine's motoring-pressure curve.
Effective Burning Angle - Burning angle minus lag angle

The "lag angle" is by far the slowest part of combustion, where very little flame travel or increase in cylinder pressure occurs, and the region that high energy/mutli-spark/multi-plug ignitions attempt to improve upon (Comnoz took 4 deg out of this region with his high energy multi-spark single plug ignition). There is little that can be done to accelerate the "effective burning angle" since this portion of combustion occurs when flame propagation velocity is high, accompanied by the rapid rise in cylinder pressure which further accelerates the burn rate.

To bring the above back around to the real world, let's consider an example - the ignition timing employed on the often hot-rodded small block Chev V-8 engine. This engine has a wedge combustion chamber that is nothing special in terms of efficiency, and has historically employed ignition timing of ~ 36 deg BTDC. Why is this range of ignition timing chosen? A wedge combustion chamber is not particularly efficient, thus it requires significant spark lead to get the ignition event started. In general we could view the combustion process in this particular case as having a lag angle of ~ 15 deg, so it takes basically 15 deg of crank rotation just to get the fire started. By the time the "effective burn angle" starts the crankshaft is at 20 BTDC. A typical "effective burn angle" occurs over ~ 40 crankshaft deg and for best power is evenly spread on either side of TDC, with peak pressure occurring ~ 20 ATDC. So in support of JRB's range of 50-60 deg for combustion to occur, a suggested break down of how 36 deg of ignition timing actually requires ~55 deg for combustion to complete.

Engines that employ best power ignition timing under 30 deg have one or both of the following attributes - improved ignition and/or improved combustion chamber design.

Another aside relating to means of reducing ignition timing is a literature example of a flathead test engine fit with 17 sparkplugs evenly distributed across the combustion chamber. With the timing optimized for best power at 2000 rpm, the combustion time using a single plug was 2X (3 milliseconds, 36 deg) the combustion time of firing all 17 plugs simultaneously (1.5 milliseconds, 18 deg). In this latter case the "lag angle" must obviously be negligible.

Thinking out loud and visualizing the lag angle in my minds eye, in the worst case scenario the "lag angle" must be akin to striking a match somewhere out in the dark universe and waiting to see what happens as the minute speck of faint light ever so slowly begins to grow. In contrast (again in my minds eye), Comnoz' high energy multi-spark ignition system must be like striking a 200A arc welder that creates a brilliant yet short sparkler (like the head of a meteor) with the burn initiating along this ignition line. At any rate I think he's got the lag angle down to about 4 deg, and at 24 deg total timing, is about the shortest ignition timing I've ever heard of. My $0.02.

PS - Hypothetical question. What if Comnoz had the same ignition in a dual plug format? How much more timing could be taken out of the equation? And yes, I realize it would take a truck alternator to power such a beast, hence I said hypothetically.

Believe me, I have looked hard at the possibility vs the difficulty of installing a second plug. It could be done and I have enough alternator to power it. It would not be easy and some head finning would have to be sacrificed. Since the ignition I have has done a pretty good job of delaying pressure before TDC I am not sure how much could be gained.

I actually have a plasma ignition unit here that I want to play with first. It will do much higher frequency than my modified MSD board. It will require going to individual coils instead of wasted spark as I have now, so some ECM changes and a cam sensor will be required. That will also let me run synchronized injection. Jim

comnoz said:

acotrel said:

Are you suggesting there should be a load cell sensing the torque reaction ?

Click to expand...

No????

No Load = engine running through the rpm range with no load other than the drivetrain and tire to road loss. Engine speed is controlled by the throttle position -slowly changing speed either up or down

Full load = wide open throttle -engine speed is controlled by the dyno brake as the engine speed is run through it's rpm range -slowly changing speed either up or down

Click to expand...

What I was suggesting is that if there was a torque reaction measuring device where the front isolastic mount is on a Commando, the output from it could be used by the microprocessor to automatically optimise the advance curve against throttle openings for max. torque for the various carb needles. The programme would have to do a lot of testing and toing and froing, however if it could feed a storage device, the conclusions to be made from it might be very useful.
I think that is what you are effectively doing with the dyno, however you are not actually riding the bike when you are doing it. That load cell would measure the force as the front of the motor rises under acceleration. You would probably need to use only one other isolastic mount and no head steady.

Torque measuring? thats so vairable, what would be the datum point? I remember seeing a torque scale on early Aston Martins ..there was a fixed scale on the body and a arm from the engine,when reved the arm measured the engine's "twist" . Moving the dizzy back and forth caused the measureing scale to move.

Is there a need for datum point if you are comparing the current torque reading to the the one immediately prior to it in a damped system, and then adjusting other things to get a maximum by adding or subtracting ignition advance ? Surely in the programming you only test if the next torque value is larger or equal to the last, then act to adjust the other parameters if it is less ?

The responce to how much power to fire two plugs, 40 years ago i drilled /tapped a second plug hole in my BSA single, simply split the HT lead , the differance to tick over was large! pull one plug and it slowed,hooked back on [rubber gloves] and it speeded up..so even a 6 volt battery had enough poke to fire both plugs...carnt say i conducted precise tests I guess it was like setting fire to a barn...blow lamp or two matchs...the barn burns quicker when a match is applied same time at Both ends!

john robert bould said:

The responce to how much power to fire two plugs, 40 years ago i drilled /tapped a second plug hole in my BSA single, simply split the HT lead , the differance to tick over was large! pull one plug and it slowed,hooked back on [rubber gloves] and it speeded up..so even a 6 volt battery had enough poke to fire both plugs...carnt say i conducted precise tests I guess it was like setting fire to a barn...blow lamp or two matchs...the barn burns quicker when a match is applied same time at Both ends!

Click to expand...

Hmmm...

A mate of mine was a serious Velo racer in BHR. He set up one of his Velo race bikes for twin plugs. He found exactly the same as you did John.

We both thought 'well there you go, that's PROOF that it is better'... Simples innit?!

But when he put it on the dyno it produced LESS power with twin plugs!

I'm not sure if he ever got to the bottom of it. I simply concluded that the whole twin plug thing, in fact the whole sequence of events that happens inside the combustion chamber, is a tad more complex than I had realised!

Just to muddy the water a bit, have you ever seen Honda cars stratified charge engine set up? Like on an 80's Accord, they use a thimble sized chamber in the head , feed fuel and air through a pinky nail size valve,with a port exiting into the main combustion chamber. The carburetor had a seperate pinky size throttle bore to feed the thimble. The spark plug sets off the charge in the thimble which in turn blasts the flame front into the main combustion chamber. These engines could pass emissions tests of the era with very few other added devices like cats and air pumps.