My car used to do this after I originally purchased it, it was the O2 sensor not reading (never had a code for it either). Wide-band did the exact same thing you describe, right down to randomly going crazy lean while cruising. It idled fine, but never quite stoich so it took me forever to figure out too.

Turned out the previous owner actually cut the wire for the O2 under the dash to hook up a narrow-band gauge :/ Soldered the wire back together, put in a new O2 sensor, and BAM everything ran great and the problem disappeared forever.

First thing's first, how are your long-term and short-term fuel trims? Do you have Evoscan to look at it? It'll tell you right away if it's messed up, because they'll be way over +/-5% (mine were completely maxed out). A logger/scanner would be the first place I'd start, but I'd bet it's going to be an O2/fuel trim issue. Good luck

Thanks for the reply and it’s just a stock tune. I have swapped ecu’s with a known good one and same thing happens. I checked my 02 sensor connections as well and they are all tight and connected. I have no logger or evo scan. My brother has dsm link in his 97 talon but not sure if I would be able to plug and play that into my 95 6 bolt swap. I have a reader that I use for codes and it is not throwing anything related to fuel trims. Only my usual mdp sensor code and an alternator code. Have had this issue before the swap.

I'm also a 6-bolt swap into my '98; it did it on both motors until I fixed the 02. I can explain why the ECU does it if you'd like including why swapping ECUs won't change it. I never had any error codes at all for it. Too much to type from my phone to explain everything.

If you're looking to quickly test my theory a logger would be best . I'll look into logging options for the '95 once I'm home. You could also hook up your wide-band's secondary output as a narrow-band sim and see if that corrects it for something to try for free.

Yeah I would love to know more about this! I will look more into this over this weekend and give you an update! A way to log things has been on my list of things to get for a while I just have never “needed” one and now I have an excuse to get one. This really sounds like it might fix it if I can’t get my hands on a logger soon I will set my wide band set up for a narrow band sim as well!

Alright I'm home, so info dump incoming! Assuming the issue is indeed the O2 sensor and therefore fuel trims:

The long and short of what's happening is that the ECU is leaning out the fuel when it goes to closed loop due to insufficient/erroneous O2 input, but lets dig a bit deeper:

What exactly is happening here? As you're probably aware the ECU operates in two modes when it comes to fuel trims: open loop and closed loop. There are two values it stores that affects your closed loop fueling in all cases, (it even affects WOT, but it's very minimal; the ECU determines idle (low), cruise (mid), WOT (high) via a load threshold table) the Short Term Fuel Trim (STFT; current O2 feedback) and Long Term Fuel Trim (LTFT; saved O2 feedback). These two trims are carried as a percentage, and adding them together gives the total fuel adjustment; anytime the ECU goes to closed loop it starts from the saved LTFT and uses the STFT to slowly adjust it (the LTFT is re-saved roughly every 4 minutes). Anywho, from what I've seen, if there is no O2 feedback but there is no CEL, the ECU will adjust these anyway as though there is a rich condition, and stores the LTFT as -25% so anytime it tries to jump into closed loop it goes ridiculously lean. I actually watched mine switch to closed loop and instantly go lean on the logger while cruising.

Why do different ECUs still do it? Since all 2g ECUs store the fuel trims and record them the same way, a different ECU will quickly max out the LTFT values since there is no O2 feedback. I tried multiple 2gb ECUs in mine, and they all did the same thing. Why they all default to seeing a rich condition when there is no O2 feedback is a great question.

Why is it so random? I haven't disassembled the ROM personally, so I can't 100% explain why it does it seemingly completely at random, nor why it doesn't do it at idle. That said, I can certainly pose some guesses. There are multiple/many tables that the ECU refers to in order to decide when closed loop is active: TPSvRPM, Temp threshold, LoadvRPM, etc. with varying values that must be met, and the ECU is constantly checking them. Meanwhile, the ECU is aware that the LTFT is maxed out and there's likely a limiter to not introduce such a hard crossover at idle as it can stall the engine, keeping it in open loop. However, since it needs to eventually test this without stalling, cruise would be a good choice for safe testing, so on occasion it does attempt to activate closed loop which goes goofy lean and stays there as there is no actual O2 feedback. That's about as good a guess as I have, as we know the ECU will test things on occasion to clear CELs and such, so I'd imagine if there's a limiter when the LTFT is maxed out it still tests closed loop every so often.

Why does it go away after I push the clutch and go back to stoich? Why does it usually cruise at stoich? When you push the clutch and release the gas the ECU cuts all fuel injection (decel fuel cut), which instantly puts the ECU back into open loop, so it will directly follow the stock fueling table. This will cure it for a bit until the ECU decides to test closed loop again. Assuming you have no vacuum leaks and the MAF is reading correctly, the fueling table is set for stoich, so that's roughly what it will be while idle/cruising in open loop. My idle was a bit rich because my Walbro was mildly overrunning my stock FPR at idle (you'll notice a trend of the previous owner f*cking up every mod he did to my car before I purchased it...). I've attached the fueling table for a 2gb DSM so you can see what I mean; most of the map is set for stoich (14.7:1) at low/mid load/rpm, so that's what you'll get.



If you'd like more information on the tables within the ECU as well as conditions for open/closed loop you can look at Merlin's ECUFlash Tuning Guide. Most maps and triggers are very similar between DSMs and Evos, the DSM ECU is just dumber and less capable XD

Hybrid ECU Logging Options: The hybrid ECU has always been a pain to log as it outputs MUTII protocol over an OBD2 interface, but there are a couple options available. I'd really suggest thinking about your plans for the car's future as you'll save money in the long run getting something you can use then and now.

DSM Link (ECM Link) is obviously the primary choice for logging and tuning on this platform, and it can't be recommended enough! Of course the small price tag of ~$600 is a tough pill to swallow when the 2gb flash ECU and Evo8 ECUs can be purchased/tuned for less than half of that, with the Evo8 ECU being almost as capable (and in some aspects, more-so). Still, it's a phenomenal all-in-one package that will likely do everything your DSM will ever need, and then some.

Hand Held Halo (HHH) is a great choice if you're never going to mod/tune and just plan to stay relatively stock. The hybrid model is no longer in production, but they pop up for sale on occasion. There's either the palm version, or the PC version; both would do the job. It cost about $200 new, but you can often find it for half of that.

Evoscan is a good option if you plan to move up to a 2gb flash ECU or Evo8 ECU, though it will require you to have a laptop. It runs $25 for the software and ~$60 for the 1.3 cable (there are 3 different 1.3 cables; I've read that D works and supposedly U works too. I'd check the forums or email them before buying though). If you plan on flashing a 2gb flash ECU you'll need the 2.0 cable and flash adapter which is about $215 (note that the 2.0 cable has roughly half the logging speed of the 1.3 cables, flashing with the 2.0 and logging with the 1.3 is suggested). Gatecrasher on 3Si also made schematics for a hybrid cable that you can DIY for like $10, it supposedly works, but I've never built/used one.

BlackStealth over on 3Si offers a couple options that work for the DSM platform as well as the 3/S platform, they also conveniently work on 2gb ECUs, Evo8 ECUs, pretty much anything that runs MUT protocol. The BlackStealth LCD Boost Controller is freaking awesome if you want to install it in your cluster and it outputs live data on top of being an electronic boost controller. It's not too cheap though, as you'll be dropping ~$250 for the unit and adapters, but remember it's also an electronic boost controller that can pull back boost if you get knock which is pretty great for the price. Be sure to email/PM him before you try to buy anything so he can ensure it will work and get you the proper adapters/firmware.

BlackStealth also makes a MUT to CANBUS OBD2 Converter which will allow you to use any OBD2 reader, specifically it's designed to use a bluetooth/wifi adapter and Torque. It will run you about ~$120 for the unit and another $30 for the bluetooth/wifi adapter and Torque Pro. I use Torque for all of my vehicles, so I got one for my Stealth to see live data on my Android head unit. Be sure to email/PM him before you try to buy anything so he can ensure it will work and get you the proper adapters/firmware.

That's all folks! Well I hope some of that helps you out; it just wouldn't be one of my posts if it wasn't a freaking novel! I know some of it is just my speculation, but it's probably about as good as you'll get unless Ceddy decides to roll out of his grave and do more ROM disassembly... I'm fairly certain it'll end up being the O2 sensor as your symptoms are exactly the same as mine were. In any case, hope this helps you out man, and good luck

This is so helpful Melbowski!! It really does make sense now. I also got the chance to put my outdated and cheap elmscan to use on my way back from the gym tonight. Your theory was 100% correct. It never showed a cel until it hit the lean condition and it would throw an o2 circuit malfunction code. I’m also able to see my fuel trim values and o2 sensor data on this software (it’s not the best software by any means) I just use it to check codes which it does pretty well. I’m not exactly sure what I was looking at but they were changing a little here and there if they weren’t just steady at 0.

Props to the previous owner for taking out the cel light before I bought it. Trying to pull a fast one on me.

On to the solution now. Should I replace my front o2 sensor and check the wiring with my multimeter or option 2 use my wideband to simulate the narrowband reading to the ecu?

My wideband sensor is right after my flex pipe welded in. Should I move it to the front o2 housing if I decide to go that route?

If any of my jargon doesn’t make sense I apologize I’ve never had the chance to learn how to use tuning software and correct terms and such!

I also forgot to mention it’s an 95 eprom Ecu. So it’s not if but when I decide to shell out the money for link that is the route I will be going.

You want to keep the Wideband out of the O2 housing if you can. As long as you have no leaks, where it is at will narrowband sim fine.

So a little update: Start with the good news first. Replaced front o2 sensor with a denso unit. It fixed my rich idle problems and all cruising speeds are stoich now. Still having a problem with it going lean at higher cruising speeds. Will still peg lean and now sometimes, but not always it will fix itself within seconds. Sometimes still have to press clutch in drop revs. Maybe a wiring issue gonna dig deeper into that. So not a 100% fix right now but helped a lot.

Bad news now is that the past week I started to notice a slight knock from 2-3k range. Very hard to notice and wasn’t there all the time. During the time diagnosing this lean condition it’s gotten worse. Had some nightmares cause of course first thought is rod bearings going. Due to my paranoi I dropped the oil pan, cut the filter open and bam I got glitter in the oil pan. I could physically wobble cylinders 4 and 2 rods on the crank. Cylinder 1 was on its way as well. It did not spin any rod bearings I caught it farely early. Hopefully my built head will be alright. So I’m now at a stand still of course! Car is parked and waiting for rebuild. Yay. Very disapointed cause it was supposedly a solid 6 bolt with only 10k on it. Got the engine for a good price down in Minneapolis. I only put around 100-150 easy going miles on it. Very Very fishy to me. ARP through out the whole block 1g rods. Like they say shit happens and I’m not even that mad but what I foUnd in the oil pan is actually very heartbreaking to me. I pulled out what looks like the metal seal to a jug of oil in the pan. This was not me I also pulled some other random crap out of the pan as well. Very careless previous owner seems to me. Then again I should have taken the pan off before I swapped it in my car. Kicking myself right now.

Guess I’ll be doing an engine build in the future. I got a good 4g64 block and also a 7 bolt 4g63 block and now a 6 bolt. Might build a stroker who knows money is tight right now so it’s gonna be sitting for a while.

Any advice for me regarding measuring clearances on the crank and what not for rod bearings and if there is any reusable parts. Mostly concerning my rods. Need some advice from experienced builders. Any info is appreciated.

So it’s been a couple months since I’ve had the Dsm running. Got the old 6 bolt rebuilt and it’s strong as ever. Onto fixing this leak condition I’m still having. As I was breaking the engine in on the first 20 miles I did notice it do it once again randomly while cruising for just a couple seconds. I also before breaking in the engine I wired my wideband to simulate a narrowband signal on the o2 circuit. This didn’t not help I’m assuming since it’s still happened. What now!?