kristopher_d

I don't understand the point of going open loop. I'm sure I'm missing something, but all my reading doesn't explain why you wouldn't want to be adjusting fueling based on AFR at WOT. I do understand going richer at WOT to provide the safety margin, espcecially when you're really just guessing at how much fuel you need to keep things safe when you're open loop. But again, I don't see the point of open loop at WOT. Given the processing capabilities of the stock ECU, it has more than enough flops to calculate and adjust at any RPM the engine can spin to safely.

To me, and I know I'm probably missing the point, which is why I'm asking, it would make more sense to progressively richen the mixture as load increases, but still tune for target AFR's.

Of course, it's not as much of an issue of the injectors are flowing like they ought, but since there's a wide varience, and the ability to adjust accordingly, why is open loop the WOT paradigm? If we were closed loop at WOT, there wouldn't be so much issue creating "richer" or "leaner" maps for cars that need them, would there.

arghx7

Here, read this: http://en.wikipedia.org/wiki/PID_controller . It's technical and math-intensive at some points, but you can get some general ideas from it.

Now, is this a general tuning question or a specific question about the design of the Rx-8 PCM? I think on many OEM vehicles there is no point as it adds more complexity to the design of the PCM and fuel injection system while the benefits are limited.

First, Cars generally are open loop under WOT from the factory first because narrow-band sensors can't read AFR's that rich accurately, and most cars still have narrow-band o2 sensors designed only to keep the car at ideal catalyst efficiency (14.7:1). That's the biggest thing first of all. But some cars have widebands stock, so why not go closed loop under high throttle conditions?

There are two issues with WOT closed loop operation I'd like to talk about. First, you better trust the O2 sensor if you are going to depend on it to make minute corrections to a fuel map that should already be pretty well designed (like on a factory car). Common production widebands like the Bosch LSU sensor were not designed for long term closed-loop operation under rich conditions... They have a factory calibration resistor in them, but If a PCM is making corrections to the fuel at WOT using a sensor that has 80k miles on it then there will be less accuracy. You are introducing a source of error. Some aftermarket wideband systems have a calibration feature, and I've heard of some PCM's that self-calibrate their widebands occasionally.

Second, you have to tune self tuning! On the Megasquirt EMS for example, if you want to run closed-loop WOT you have to set a bunch of different parameters to get precise correction. Ever tried to tune a self-learning electronic boost controller? It's easy to get slow response or correction that over-shoots. From an OEM's perspective, it's just one more thing to spend development dollars on. To what benefit though, for an OEM car?

On factory cars that are all pretty similar, it's just way simpler to program the PCM to say "ok, if x grams of air come in, inject this amount of fuel." They don't need another expensive failure point for the PCM when most people hardly ever go WOT. Besides, if the engine knocks, factory PCM's all have closed-loop knock control systems that build timing trims. GM LSx engines for example have a high octane and a low octane map (the "dumbass put in 87!" map), plus additional algorithms to adjust timing beyond just switching maps. Remember also that on a system with a MAF sensor, if extra air comes in the MAF will signal the PCM which will know to inject more fuel. Speed density cars like Hondas and 93-95 Rx-7's basically guess how much air is coming in based on manifold pressure and a table of engine efficiency.

So in summary, from the factory closed-loop setups under WOT just aren't worth it, especially since most drivers hardly ever go WOT. It's more to develop and more to fail, it requires precise o2 sensors with excellent longevity, and if knocking is the biggest concern, they already have closed-loop knock control. As far as aftermarket standalones go, well some people run closed loop with some success but many people don't bother.

zoom44

as argh mentioned- its not the power of the PCM but the long term accuracy of the O2 sensor

kristopher_d

Both general and RX-8 specific.

Thanks for the explanation.

So, on the RX-8, it's trust-the-injectors over trust-the-02-sensor because it works well enough for Joe Average? That being said, if one were to re-tune the factory ECU, ideally the WBO2 would be calibrated in the process, the injecters would all be flow matched, and you'd set the open loop threshold to 110% throttle?

arghx7

I don't know all the details about such a conversion, but you've gotta hope the control algorithm in the PCM would actually work well under those conditions. You can tell any self adjusting system what the targets are, but that doesn't mean it won't just oscillate worse. So if the target AFR is 13:1, the ECU reads 13.3 on the O2 sensor, it's entirely possible that it would either react too slowly (maybe only richen it up to 13.2:1) or overshoot and give you 12.5:1, then on the next calculation cycle it overshoots or undershoots again, etc. That's what COULD happen, and that's the classic problem of any closed loop system such as an electronic boost controller, a thermostat in your house, anything.