I have a 1987 TurboII with Rtek 2.1, firmware up to date (1.2.1 I believe) and a ZT-2 Zeitronix wideband controller. I'm having problems with the simulated narrow band output. I ran the SNB output to the O2 sensor input to the ECU and disconnected the factory O2 sensor from the harness and set the controller to output the SNB signal switching at 14.7AFR. I spoke with a Zeitronix rep and I told him what the FSM says about the factory O2 sensor's output and he said the SNB should match that, but under light throttle I would see very lean AFR's (17-20) and the car would buck and hesitate. I disconnected the SNB and reconnected the factory O2 sensor and that issue is completely gone, but I want to be able to use this feature so I need to figure out how. Does anyone else use this feature successfully? I searched quite a bit and found only one person who said they run it with it switched a tad richer than 14.7 (so probably 14.6 or 14.5, he didn't say exactly). I didn't dare give it heavy throttle when using the SNB, if it runs crazy lean under boost I'm screwed for sure. I saw that some have had issues with other widebands' SNB outputs, one said he slowed down the refresh rate for the output and that solved thhe problem, but I don't think I can do that with the Zeitronix unit (if I can someone please let me know how). I will take a look at some of my logs for more hints and possibly post them up here.

 

With SNB (ignore user input on this log, it was disconnected):


Without SNB (user input on this log was hooked to the controller's own SNB output):

 

I'm running a zeitronix/rtek combo also, but on an na. I was running mine slightly leaner at 15:1 in an attempt to increase fuel economy, but I didn't notice any difference. It didn't have the hesitation you are experiencing, but it was jerky in throttle transitions. I find that the car runs smoother when I run with the narrowband output disconnected (no closed loop).

Is it possible your airpump is on and giving false lean readings?
Also, the ecu won't look at the O2 under full load, it will be in open loop and only look at your fuel map.

 

I have virtually no emissions equipment: no air pump, no ACV, no split air pipe... I do have a CAI: https://www.rx7club.com/2nd-generation-specific-1986-1992-17/my-cheap-cai-884103/
and complete Racing Beat REV TII exhaust.

The car wasn't at full load as pictured when the SNB was connected to the Rtek. The markers show the throttle being at 24%, 24%, 53%, 49%, 48%, 0%, 25%, and 35%. All at low RPM, and the corresponding AFRs were 13.6, 17.3, 17.3, 19.8, 18.2, 21 (maxed out), 13.7, and 17.3 respectively. That can't be right. And it felt VERY wrong, too. To add to that, it couldn't have been using only the maps, because once I reconnected the stock O2 sensor and disconnected the SNB the issue disappeared. With the stock O2 sensor connected instead I had no such light throttle problems. I would sometimes see 16 AFR, but that's fine compared to 19+.

 

Can you adjust the refresh rate on the wideband? I do the same thing with my Rtek 2.1 and Innovate LC-1.

I had to slow down the refresh rate for the narrow band simulation. I was having bad part throttle lean hesitation because the wideband was too fast.

 

Why's the RPM so low in the first log? Were you actually driving it that low? I'm not sure of how the RTEK does closed loop, but if it's trying to correct on incorrect feedback it could likely be causing your issue. Do you have a graph that shows the Narrowband simulated output range?

 

I don't know if I can or not, I haven't seen anything in Zeitronix's documentation about it. The Zeitronix rep said he'd call back if he found out something that could help, but he might not find anything.

 

I was just trying to get started from a stop/slow roll, I was afraid to push it higher or harder since it was bucking and running so ******* lean. I drove it two blocks away and came back. I'm pretty sure the Rtek works just like stock in how it uses the O2 sensor reading. And sort of, the user1 input on the second picture is the SNB output voltage. It ranges from 0-1V. I do wonder though if the SNB output has a different enough curve to cause this, but I wouldn't think so.

 

The output looks wrong to me. The 14.6 vs 14.7 are almost .3v apart! Below is the NB Sim that techedge uses (and i use often). As you can see yours is reading all wrong in comparison, it should have better resolution between about 13.5 and 15.5, yours is jumping way too much between just 14!




Edit: It almost seems like the range it's reading (or actually sweeping through the 1v range) seems to be somewhere around 15-16:1, you could try to push the narrowband switch point lower (say 14:1 or 13.7:1), and repeat that same datalog to see if the outputs better match what the stock ECU is expecting to see.

 

I've searched for typical stock O2 sensor plots and they look more like the Zeitronix SNB than yours, and I don't know how true to scale that Zeitronix plot is. It would be MOST helpful if someone actually using this feature with their ZT-2 could shed some light on this.

I'm going to compare my Rtek log with my ZT-2 log and see where their NB plots differ. I had considered that maybe it's a little off, on the same idea as when you mentioned switching it to a richer switching point. I'm going to do a little more research before I just start trying things though.

 

The plot I posted is based off the standard Bosch sensors. It's pretty much the same as the ZT one, except it's not quite as smoothly curved on the extremes (A real Bosch isn't either, but it shouldn't matter much at those values). In any event, yours is definitely reading incorrectly because it's reading anywhere from .7-.98v for 14.7:1 when it should be closer to .5-.55v. The problem is your ZT isn't following the picture shown, which I can't quite figure out as to why. The only idea I'd have is to shift the whole thing (beacuse you can see it starts to step between 0-1v somewhere in the 15's range, which should be happening in the 14's range). I don't know anyone using a ZT with the stock ECU, only aftermarkets so I can't even think of anyone to ask to test it.

 

Ah, I see what you're saying now. Yeah maybe it has shifted for some reason... perhaps poor programming or something wrong with the circuit, because it should be a simple on the fly calculation. The one variable is the wideband's reported AFR, the other is the switching point, then BAM you have a voltage to output. Why in God's name that voltage would be higher than it's supposed to be is beyond me, at least for the moment.

 

What baffles me is how it jumps almost .3v for such a small change (from 14.6 to 14.7), it could just be that it wasn't normalized for that one reading because it was just kind of a blip right there, but the rest of the log really seems to put it somewhere in the 15's somewhere in it's normal operating voltage. I guess you could datalog it once more as it is, mark off a few points to see exactly where it's operating voltage is, then adjust the shift accordingly and repeat to see if you can get it down where it should be. Unless for some reason the datalog is showing the voltage incorrectly, all sign pointing to the shift being wrong.