Engine unstable

[brashear]

Hello,
I am converting a ~300cc single cylinder air-cooled engine to EFI. When running at around 3000RPM with a light load on it, the leanest the engine will sustain a stable air fuel ratio is around 13.5 +-0.5 variation. If I reduce the fueling even just slightly (a few percent less), the AFR will go up to about 16 with +/-1 variation. This problem does not occur at higher loads. The most I can advance the spark is about 25 deg due to the placement of the crank trigger. It seems like it's a problem with misfire at the lean limit, but I would expect the engine to be able to sustain at least stoich, even at light loads. Any insights on how I can absolutely confirm that it's a problem with misfire? How much leaner will I be able to run the engine by advancing the spark further?

 

[NickJ67]

Take a look at your injector pulse widths. I suspect that you may be hitting the minimum (typically <2ms) which is why control is lost. Basically this would mean that your injector is too big but altering settings (if available in your ECU) for open/close times might win you a little. What ECU are you using?

Nick

 

[brashear]

I switched over from crank timed to cam timed injection (from 1 injection per revolution to 1 injection per 2 revolutions) and the problem still persists.

 

[bradrs]

If you are limited to 25 degrees, that makes me think you are still using the OEM trigger? And the OEM CDI? If so, you need to investigate ways to increase spark duration. You can run leaner, if you have a longer duration spark. OEM ignitions can be weak at idle too.

Another thing that might help would be trying to make the mixture more homogenous inside the chamber. Better injectors with a finer spray pattern might help as well.
More details on your setup would help.

 

[BrianPetersen]

I presume that you are measuring air/fuel with an ordinary lambda probe.

Is there an audible misfire in conjunction with the display showing lean? Even if there isn't, it could be that combustion is not going to completion. Either way, the lambda probe is probably indicating a false lean condition - the true lambda ratio in the chamber is not lean, but the unburned air is showing up on the oxygen sensor.

What type of engine is it, what is the valve arrangement and what type of combustion chamber does it have? If the combustion chamber is a crappy design (slow burn because of not enough squish action, not enough charge turbulence), that is not going to help matters.

What's the exhaust system like? Excessive exhaust residual can lead to slow combustion and partial misfire. It could be because of too much back pressure, but also if the pipe arrangement has a resonance at that RPM that the engine doesn't like, pushing a pulse of exhaust back into the cylinder during the valve overlap period. Problem goes away at higher load because the higher pressure in the intake system resists the nasty pulse better and the extra volume of air and fuel reduces the effect of exhaust dilution.

Where's the injector? It should be aimed at the backside of the intake valve and the spray pattern shouldn't hit the side of the port too close to the injector. (Down close to the intake valve is OK.) If the spray pattern is hitting a surface that is not hot when the engine is running, you could be getting poor atomization.

Most engines want more timing advance than you've got at light load, but there's not really any mathematical way to relate how much better it will work if you can get more advance than that, this is purely experimental in the real world.

 

[brashear]

I was trying to see if there's a way to figure out whether or not the problem can be attributed to misfiring other than just putting on a new crank sensor and coil. If it is indeed a misfire problem then advancing spark / longer spark duration / things to promote faster burn rate at the same lambda (more homogeneous mixture, introducing swirl) would likely indeed help.

 

[brashear]

The engine sounds off when the O2 sensor is lean, but I can't say for sure that it sounds like a misfire.

I should've taken a pic of the head when I had it off the cylinder, but I doubt there's a lot of squish. The compression ratio is 8.5, which is quite low for an engine this small so I don't think the designers optimized the squish or else they could've increased the compression.

Not really sure about the exhaust but good points.

The engine was not designed for EFI and as such has a long length of intake. You can't see the back of the valve from the intake port. I've oriented the injector so that as much of the spray is hitting the engine casting as possible, subject to packaging constraints. I would say at least 90% of the spray goes to the hot engine side and not the insulated injector holder side.

 

[Tmoose]

First I'd watch how bright and steady the ignition is using an inductive timnig light.
Then I'd dig out the old oscilloscope and see if I could capture the primary and secondary spark patterns.

http://www.picoauto.com/waveform/

http://www.picoauto.com/waveform/Ignition/Primary/wave65.html

http://www.picoauto.com/waveform/Ignition/Secondary/wave84.html

 

[bradrs]

Those waveforms are for an inductive ignition. He hasn't shared whether he has an inductive or CD ignition, but most small engine ignitions like that are CDI. An inductive ignition would have a much longer duration spark, but he sounds like changing the pickup and coil is something he wants to avoid. Changing from CDI to inductive is more work than that. So, probably not high on the list of things to do.

It is possible to get more power out of a stock CDI too, or maybe improve the energy in whatever CDI he is using. But it is difficult to describe how to do it, without know what he is working with.

 

[Lou Scannon]

How are you applying load?
If you can do a AFR sweep while holding torque or speed, and throttle constant, you should see a peak rpm or torque somewhere richer than stoich, then a gradual decline as you go leaner, until you reach the lean limit, when torque or speed falls off sharply.
Another way to go about this would be to fix the torque (or better still rpm, if you have a dyno that can measure torque while holding constant rpm) and injector PW, then vary the throttle opening over the usable AFR range (about 10:1 up to misfire limit). Do this for a series of pulse widths in the operating area of interest. From the data thus collected, you can plot torque as a function of AFR at more-or-less constant (3000) rpm, and see the misfire limit.

"Schiefgehen will, was schiefgehen kann" - das Murphygesetz