I've noticed just how tweaky the LC1's nb signal is and what it does to prop control. I've read a few threads from a while ago explaining ways to deal with it, like changing the sampling rate or narrowing the signal bandwidth (both ineffective). Maybe I'm missing something here, but wouldn't it be easiest (and best) just to increase the signal filtering? It could be made much more stable if the slow O2 error window were widened and the filtering increased.

I see people using the LC-1 for NB. I am just wondering what you are trying to gain over using the stock O2. I can see adding in a three wire sensor to get the heating element upgrade so that you can go closed loop faster.

Are you using it in the stock location? I think that you want the NB close and the WB far from the header collector.

i just graphed the o2 output in tunerpro and slowed the lc1 output till it looked clean and smooth. seemed to work for me.

at least its easy that was.

My exhaust's nowhere near stock, and I wanted to simplify the sensor location. I don't trust the stock sensor for anything anymore, too unreliable. The LC1's much more accurate & 1000X more responsive than the stock sensor which makes it a much better 'switch' for the ecm to find stoich (and I can change the output to make it run any afr I want, just about anywhere I want) but the responsiveness is where the problem lies.

Stock programming doesn't have enough filtering to use the simulated signal properly- the INT's always chasing the 'tweaks' in the simulated signal. Infact the LC1's quick enough to track individual puffs of gas leading to false rich/lean signals. Slowing the sample rate (response rate on the LC1) doesn't fix the problem either, it just slows the number of 'reports' to the ecm, giving a more generalized but less accurate image of what the fueling's doing. Increasing the filtering increases the averaging of what the ecm sees but does it at a more active rate.

Imagine someone walking down a flight of stairs. They get down the stairs at a fairly constant rate but they spend alot more time out of synch with the general direction of travel due to the jerking motion of stepping down to each level. That would be in effect what happens when you slow the sampling rate. Now imagine someone walking down a hill. They move down the hill at a much more linear rate because they don't have to jerk between steps. That's what happens when you increase the filtering.

Both different means to the same end, one's more linear though. I've noticed a dramatic change in fueling stability on my datalogs (with reduced prop gains) vs changing the LC1's response rate. What this means for me is a cleaner signal of where the engine's running making tuning that much easier, and improved fuel useage.

I always meant to research how the lc1 slows the responce time. like if it adv it or just slowed the sample rate. but never did. where did you change the o2 fliter value in tuner pro?

Could you post a screenshot of the selected output in your LC-1 software?

I also had the same problem with mine and ended up going back to the stock sensor for the ECM. I believe it was because the stock o2 sensor output is more like a sine wave, where the voltages drop/rise more rapidly during rich/lean conditions, and the LC-1 seemed to be more linnear. I never had a chance to mess around with it after going back to the stock sensor since I grenaded the engine.

The table is at LC047D called slow 0 error filter coefficient vs airflow. I still haven't found what it uses for filtering outside the 02 error window, I'm starting to think it doesn't.

I've changed my O2 tables & the LC1 to straddle 500mv as stoich, so you might not want to use it. Here's a graph of the stock sensor's response curve.


Full on to full off is 14.5 to 14.9 at .850mv and .050mv, respectively. The stock sensor's signal tapers off at the extreme ends of it's range in more of an 's' fashion than a 'z', unlike like the LC1 signal. The stock sensor also responds slower than the LC1, closer to 1/12 sec, but like I said earlier the LC1 changes the sample rate. Leave the simulated output at instant and reduce the INT delay vs error table (LC044B) some to correct it, in addition to increasing the filtering and widening the O2 error window (I have mine set .1v wide).