Discuss all things tuning in this section. News, products, problems and results.
This is my first time tuning a rotary and i've been fighting some issues with my afr fluctuations recently and can't figure out what's going on.
Setup is a 13b rotary, bosch 1000cc primaries, elite 1500, and aem wideband/bosch lsu4.9 sensor. All the correct injector deadtime info is supplied and triple checked, but the afr to me seems incredibly unstable, no matter what I do I seem to come up with a 0.02 to 0.08 lambda fluctuation.
If anyone can please check the log I attached and zoom in to the idle areas you'll see a fairly consistent rpm, and MAP, but the lambda is far too jagged for what I would expect to see.
Only thing I can think is this is either normal for a rotary with more overlap than standard, I've got a distorted/ interfered signal, or ive set something up wrong.
Thanks in advance for any assistance
Can you describe the exhaust system - length, diameter, connections (V-bands, flange/gasket, slip fit?), where the Lambda sensor is located - what is the distance from end of exhaust?.
I often find idle Lambda readings are a challenge due to the low air flow, and non-combusted air pulled into the exhaust. The closer the sensor is to the tailpipe (or slip connection) the worse the problem. Sometimes you just need to tune by ear -- give the engine what it wants to run smoothly. Assume the wideband is probably reading leaner than the actual mixture.
You might try using a more aggressive averaging filter in your analysis software to smooth out the fluctuations. I found the signal was improved in MLVHD with a Smoothing Factor of 11, but even a minimal factor of 3 removed much of the apparent noise. I think for tuning, I would probably use a smoothing factor of 50 (which I think is a rolling average of 50 ms.
The manifold is one of my own making so there is every potential for some small pinhole leaks, but I did a leak test before I installed it and couldn't find anything
Wideband is about 8" from the turbo, all v-banded, with an egt bung a few inches closer to the turbine.
All the v-bands are 100% I'm confident of that atleast.
In the end of the day, as you say also, I'm not too concerned with the lambda at idle/ less than 1500-2000 rpm, as the car will only ever be there during warm up and pit cruising, when it's running it's high rpm/on boost the whole time being a tarmac rally car so I think as long as those cells are ok and stable enough I can extrapolate enough to make a safe tune.
My biggest concern was that amount of discrepancy at say 8000rpm and 120/140kpa could be a 10.5-11.8afr jump and i would prefer not to see those apex seals until I pull the motor down
OK, I missed the part about it being a turbo... are you using the pressure compensating AEM wideband? I think that is one of the issues with running a Lambda sensor pre-turbo is the backpressure can affect the readings. AEM made (not sure if it's still in production, couldn't find it on their website) a 4-channel wideband controller that has an input for a pressure sensor to do compensation. They specifically refer to improving the accuracy pre-turbo.
Sorry David, it is a post turbo 02 sensor, about 8" post turbine in the downpipe, it's a 3" dump from a gtx3582r
In your opinion if it is an exhaust leak creating a false lean spot during the exhaust pulse effectively scavenging fresh air at idle through the leak that should disappear on boost if the exhaust manifold pressure exceeds the ambient? And if it doesn't then there must be another cause, does that seem like decent logic?
I think you have the gist of it, but I'm not sure the source is an exhaust leak, from the rotor overlap, or somewhere else.
Here's an experiment you can perform.
What happens to the Lambda readings if you put your hand or a rag over the exhaust pipe exit while idling? Does the Lambda reading stabilize? If so, that indicates the source of the extra O2 is probably the tailpipe. My experience has shown that can result in a smoother / slightly richer reading from the O2 sensor.