Which tooth to remove?

[ScramblerXLE]

Hey guys,

I race ATVs in the Grand National Cross Country series. These are single-cylinder 4 stroke engines, and we do dead-engine starts, where the flag is thrown, you start your quad and take off for the first corner. To get there quickly, I’m running a second battery on the way to the starter, to step it up to 24V. Doing so allows my machine to start at about 1/3 throttle and start revving in what sounds like 1 revolution.

I’m looking at doing some R&D on an EFI conversion. I’ve got a trigger wheel that has 18 teeth on it and none missing, so they’re spaced evenly every 20° starting with one right at TDC. Since I have to remove one of the teeth anyway, I’m curious as to what everyone feels my best bet would be to get the machine to start on the first revolution. Currently, (with a Kehien FCR pumper-carb) I run my quad at about 1/3rd throttle for about 5 seconds, then hit the kill button about 30s before the flag flies. I’m guessing it fires so quick because there is still a charge in the cylinder. I can do that with the microsquirt as well, but if I understand correctly, the ECU needs to see 1 tac pulse (which I am given to understand is the tooth immediately after the missing tooth in a single cylinder, wasted spark application) before it will give you spark.

My question is, which of these teeth should I take off so that I have the maximum chance of it firing on the gas on the first revolution? I’m hoping that I can map it so that at starter-motor RPMs, no matter what the throttle position, it runs the same injection settings as well as a retarded ignition setting. (maybe 5° ATDC) and we can get it to start at full throttle.

About the most timing I could see throwing at the engine would be around 40°, so I’m thinking I need to remove either the tooth at 80° BTDC or the one at 60° BTDC so that the one at 60 or 40° was the tac pulse.

One other question: If I were to make sure that the MS didn’t lose power when I kill it 30s before the start (maybe tie the kill button into the 12V wire to the coil?) would it remember where it was when I hit the start button? Or does it clear when RPMs reach 0 so that it has to find itself again. I’m guessing that the chance of the engine turning backwards a tiny bit once it stops, would cause an issue, but I’m not sure how it’s coded…

Thanks very much for the help! Sorry for the mile-long post for a simple question lol.

-Walt # 728

[grippo]

With efi the only way to get the car started as fast as a carb is to have different missing tooth patterns, for example 1 missing tooth and 180 deg later 2 missing teeth. That could get you started within half a rev instead of one. But you also have to deal with wasted spark. If the first cylinder to fire is on the exhaust stroke it does you no good you have to wait for the next cylinder.

Better yet add a cam wheel inside the distributor and have 4 teeth of different widths. This is how oems get new cars to stop the engine at a red light and and re-start as soon as your foot goes back on the throttle. MS-extra code supports many of these kinds of wheels.

As far as memory of wheel position, when sync is lost, regardless of whether it's from key off or the engine stopping, it assumes something has gone wrong and will start over from scratch looking for sync.

[GrocMax]

Seems like you would also need to use a zero speed capable Hall Effect sensor and not a VR.

[grippo]

That would make re-start even faster as the ecu would know exactly where it was as long as power wasn't lost. If you kill power you could still save where you were for the next cold start if you had the right type of memory, but there would be no way of knowing whether someone turned the engine over by hand during the shutdown period.

[ScramblerXLE]

Hey guys, (Edit: I started this as a separate thread, which was not the right thing to do, so it got moved to here, and some of the same information as above is contained below, but the question is still open)

This is my second post on this topic, but we’re planning to machine one of the 18 teeth off my flywheel this weekend, and I have to decide which, so here we go… This will be installed on a single cylinder ATV engine which I race dead-engine start cross country races on. Due to the dead-engine start, starting as quickly as possible is preferred. With the current carburetor setup, I run 24V to the starter to get it to crank faster and it will start on the first revolution at about 1/3 throttle.

So here are a few assumptions:

First, I’m assuming that when I shut the engine off 30s before the flag, it will almost definitely wind up on the compression stroke, which is why it repeatedly starts on the first revolution currently. I’m not sure where on the compression stroke it winds up stopping, but I would think it would be before 90°.

Second, I’m assuming that the most timing advance I could ever need would be about 50°. That’s what they use on the YFZ450’s and I’m guessing the KTM engine I’m using would have a little more mellow timing curve.

Third, I’m using an 18 tooth wheel, so the teeth are spaced every 20°. I haven’t checked yet to see if, with the crank locked at TDC, the VR sensor is directly over a tooth or not, but for the sake of argument, let’s say it is. So there is a tooth at 20°, 40°, 60°, BTDC, etc.

So here are a couple of questions:

I read that the first tach pulse (tooth after missing tooth if I understand correctly) will start the first firing sequence, and after that, the next missing tooth will start the second next, etc. I assume that this means that the tach pulse must take place at greater than or equal to the maximum timing advance that I would ever want? Which would mean that I need to remove the 80° tooth so that the 60° tooth was the tach tooth.

I’m guessing this would be the case because, if I were to remove, say the 40° tooth, so that the 20° tooth was the tach tooth, than as soon as the map called for 21° of advance, we’d have a problem, as by the time the next firing sequence started, we would have missed the spark. Do I have that right?

If that’s not the case, and Microsquirt will automatically deal with that so that you don’t have the one dead revolution, then it would make the most sense to put the missing tooth at 40°, so that, assuming you had an ignitable charge in the cylinder, and the crank was at around 100° BTDC, you would get a string of teeth, the skip tooth, then the tach tooth at 20° and be ready to fire at around 5° ATDC and potentially start on that first revolution just like the current setup does.

IF I DO have to leave the appropriate number of teeth such that the tach tooth is always before the maximum timing advance, that’s fine, as I can just take the 80° tooth, but then that leads to another question:

How does Microsquirt know (using a VR sensor) where the skip tooth is. Example: I imagine if I started cranking with the VR sensor in the skip tooth region, it doesn’t know the difference, it would just assume I had started cranking a short time after I actually had. That means that removing the 80° tooth actually causes a dead area of 60°-100° where if the VR sensor happens to be there when it starts cranking, you’re waiting at least 1 full revolution before you get spark. (This would also affect my ability to start the ATV with the backup kick-starter in the event that the 24V total loss system goes dead…) Further though, I can’t imagine that seeing just the 100° tooth, then the big gap, then the 60° tooth would be seen as a skip tooth, so I’m guessing there needs to be a string of some number of teeth before the skip tooth anyway?

Final question, in my other thread, (http://microsquirt.com/viewtopic.php?f=38&t=25633) someone suggested using a “zero speed capable hall effect sensor” because as long as the ECU doesn’t lose power, it would remember where it was. That sounds like a great option, but are such things available that would stand up to the heat/oil etc inside the cases on a race engine? Also, from my reading, Hall Effect sensors detect magnets, not ferrous teeth on a flywheel? The flywheel on an ATV or dirt bike engine is also full of magnets for the alternator to work, would that not mess with the hall effect sensor? I’ve got to imagine there’s a reason all of the OEMs are using VR sensors in that application, but maybe it’s a cost issue?

Anyway, thanks very much for any information ahead of me actually picking a tooth to machine off Sunday. Sorry for the lengthy read, and the likely flawed assumptions… lol

[24c]

Hi ScramblerXLE,
Your application is so specialised I would be tempted to make a dummy crank wheel and spin it by hand from different positions to simulate a dead start.

I know from my own work doing trigger wheels, I can detect cranking and get sync at around 25-50 rpm on a camshaft trigger set up, which equates to 50-100 crank rpms. One of the ignitions I used (not MicroSquirt) would work from 7 rpm, and as soon as the trigger wheel passed the sensor it switched on the ECU and sparked! It didn't have a missing tooth just a sensor & a one tooth wheel.

Obviously you'll be using a wasted spark approach, so knowing a MicroSquirt will sync at 50 crank rpms, the question is what speed does the starter crank at? and then you'll have a likely time.

50º seems a little advanced even for YZFs, however the KTM has good combustion chamber shape & squish, so I wouldn't expect it to be that high. Bit of a guess, but on a part throttle, with a good starter, I'd probably start on 5-10º base settings, and jump to 15-20º at idle (KTMs are around 3K??), but it'll depend on your exhaust back pressure. I would as soon as its started you'll be full throttle, and as the fuel will be burning faster, it shouldn't need as much advance as you think.

Interesting problem
Mike

PS Hall sensors can work in noisy environments too, and only need ferrous wheels. Most have magnets built into them, which can lose their magnetism, or break the tracks between the transistors, due to vibration. A VR is more robust, but produces variable voltages depending on cranking or crankshaft rpms.

[ScramblerXLE]

Mike,

Thanks for the reply! I'm not sure what the cranking RPM is, but with the idle being set at around 1800, I doubt it's anywhere near 50!

I was really shocked by the 50° as well for the YFZ, but that's what I saw on the Dynatek website for their CDI. I think they claimed the stock map was around 48 or so. I was certainly thinking closer to 35° myself for a max.

Any thoughts on the timing advancing beyond the tach tooth? If the Microsquirt can do that, the temptation would be to remove the tooth at 40° so that the 20° tach tooth starts the firing sequence, then you get it to start at around 5-10° advance like you said, but once you get up above idle, you cross the tach tooth, and I would think that would mean that every time that happens, you would have a revolution where you missed a spark. Do I have that right?

To your point, the other way to go would be to take out the 60° tooth (which in the same situation as above, may lock me in to no more than 40° of advance at any given time,) and since it's a keyed flywheel on a tapered shaft, I could simply run move the missing tooth around and experiment and see what happens.

I'm using a slightly modified flywheel off a 500 EXC which was already EFI, and has 18 teeth with one of them being fatter than the rest, so I kinda wanted to hit it on the head the first try instead of having to machine on another $200 flywheel later, though I suppose cutting another woodruff key slot would also be an option...

[24c]

ScramblerXLE,

I'd go with the 60º option, the starter will turn the engine over pretty darn fast, worst case is you move it around a bit afterwards...and like you said you have a taper shaft, the keyway is redundant,it only aids setting, but you can use a dial gauge and get accurate results. Do that stuff with karts, Honda CR & Yamaha YZ in the past, if fitting after market ignitions with stock VR sensors.

Re advance, I reckon it's a 14000 rpm thing, fuel still takes x time to burn, and I seem to remember the old hemi headed Bianci's etc in the pre two stroke GP days having 60º of advance, which caused heat build up & failures.

There is a useful bit of info here too
Mike