Ian,

Thanks for your time.

The car was chocked--on the ground, not moving, emergency brake was probably not on. I did notice the laptop reading 23 mph once, and we certainly were not moving. I did not even know there was a VSS that needed calibrating. In looking at TS, VSS 1 is Off, VSS 2 is Off, Shaft Speed Sensors 1 and 2 are Off, and VSS Output is off. My guess is it has been set up like that since I got the MS3 Basic from Rev four years or so ago. (She's been tuned four times at two different places. Initial naturally aspirated tune, retune for squaretop, retune for the Rotrex install, and retune when I changed the intake tubing to the Rotrex.) If VSS is Off, does that still indicate a ground issue? If I need to turn it on, please let me know which one(s) and anything else useful. What would VSS being off cause, supposing everything else was working fine?

I checked the ground to the tranny/block beneath the heater hoses, the ground underneath the throttle body, and the grounding pack under the dash. All seem fine. Are there others?

TS says Mainboard Injectors Bank 1 Dead time @13.2V(ms) is 1.000, with Correction Curve 1. It also says MS3X Injectors (at the top of the pulldown) Dead time 0.630, Curve 1. I'm running the Flow Force injectors, new beginning of last year with the Rotrex install.

I just installed new Magnacore plug wires. They're maybe a week old.

Thanks again,

When you say you checked the grounds, does that mean you wiggled them or completely unbolted both ends and scuffed the contact surfaces and replaced them?

I pulled and replaced the engine. Both the grounds I mentioned under the hood were completely removed. I cleaned them then, which was only about a month ago. Yesterday, I just checked they were secure. I can do what you suggest again. The car is basically waiting for its next event, Feb or Mar 17 at the earliest. Plenty of time to check things.

Are there other grounds susceptable to the elements that I have missed?

Thanks,

The calibration is done in the "speed and gear sensors" menu, mine is set up as type "digital", input "PE2 Flex", diameter "1.9", type "pulses per mile", pulses "40,000", and "smoothing" of 50. I dunno if that's what all NB MS3s use, but it's something to look at it. I would expect that if the VSS is configured as "off" then the software should always output "0" for vss, but the MS3 code doesn't always follow the best software development practices... Might be worth trying to set it up and see if it makes the random values go away, it's possible that even when set to "off" it's running in some kind of analog mode that's trying to make sense out of random noise on the input line.

Those are all of the dedicated ground wires that I'm aware of in the car. OTOH, there are ground wires running to the ECU and all of the sensors in the car, and if any of those is damaged it could produce strange effects like this too. IIRC the VSS sends an analog signal to the gauge cluster, which interprets it and then sends a digital VSS signal to the ECU. I assume the speedo needle was sitting on zero and not wandering around while idling?

--Ian

I just noticed under Gear Detection in the Speed and Gear Sensors Pulldown, VSS/RPM is selected. and the number of gears and their ratios are input.

Yeah, that's so that the MS3 can know what gear you're in, by comparing VSS and RPM, it's used for boost-by-gear or other things like that, I think. I'd be surprised if it had any effect on this.

--Ian

If I were to turn it on and mess it up, would anything bad happen?

AFAIK the MS3 doesn't actually use the VSS value to compute anything critical, it's only used for extra features like displaying mpg and for logging. (It is quite useful in datalogs, though).

--Ian

I have not had a chance to look at your logs, but something helpful you can do when logging is hit spacebar in TS. This will cause a mark in the log so you know where to look. If you have not been doing that I would suggest giving it a try so we have specific places to look in the log. A good thing to do would be to hit the space bar a certain number of times and jot down what is going on. So then we could go to that specific part and the log and be able to see what you think is happening.

In post #55 there's a log from 14:22. Please take a look at the scatter plot. Does that mean anything?

Thanks,

Calculate gear using RPM and VSS.

EDIT: Duh, that's what you just said a few posts up. In this one you said, "nothing critical"... Agreed.

Like many technical people, I'm reading from the bottom up.

I mentioned in a previous post I replaced the coils with an unused spare set that had been given to me my the previous owner six years ago. I can't call them new. I don't know how old they are and was concerned one or both might be malfunctioning, so I obtained another used but functional coil pack. Installed last evening. It was dark and I decided not to take her for a ride. These logs are just idling in the driveway. No other changes or fiddling but the coils.

When she started idling rough, I got out the timing light again. #3 still had a weird pattern, but this time it was flash, flash, flash, long flash, flash, long flash, instead of flash, flash, blank, flash...

I've got the package tray carpet out and the cover off the fuel pump area. When it started running rough, there was no change in the sound coming from the pump. Can I assume the fuel pump is good?

Any advice would be greatly appreciated,

Gents,

I'm out of ideas. I really need a hand here.

I spent what seemed like forever cleaning/polishing/kissing and loving the grounds under the hood. She seemed fine when I started it up, so I was going to go for a ride and log some data. Unfortunately, she started running like crap before I even started out of the driveway. Very labored idle. I revved the engine slowly, and at about 1300 rpm she seemed out of balance. I fanned the throttle a couple of times, and at the end of the log, you'll see it just died. The fuel pump kept running for a second or two afterwards.

To recap,she's got new plugs and wires. I've tried three coil packs, three different fuel pumps, and two alternators. I've cleaned all the grounds. The crank position sensor is new. Rotrex and Flow Force injectors installed early 2015. I just swapped in the bottom end from a Japanese Domestic Market importer, replacing my 143,000 miler with one at about 60,000. When she's running right, she's a beast.

Thanks,

Sooooooooooooooooooo, I replaced the pretty new crank position sensor with the one that came with the JDM motor. I was dancing in the garage after about 20 minutes of smooth idling, but my dreams were dashed. She started running like shit again. Dammit. Now, I'm completely outta spare parts to try...

Please take a composite log when it's running shitty and post here.

I don't think it's a sync loss issue. It's showing lean on the wideband, even though it claims to be injecting fuel. Either it's misfiring with a bad spark, or maybe there's an injector that's misbehaving.

--Ian

There's the -not 100% reliable- report from the timing gun that spark is flaky. So it's likely to be spark related.
My best guess at this point is wiring.

That last log finally captured the engine dying. Are there any clues in that?

A few other questions.When I was cleaning grounds, I found that part of the wire bundle to the wideband was scorched by the header. However, no wires were showing, and the wiring was well clear of the header during that last run. If that wiring is actually damaged, could it cause the problem?

I modified the harness to extend the wiring to the water temp sensor when I did the coolant reroute. Is there anything strange in that last log related to coolant temp?

Anything else I can try?

Thanks,

Are you running sequential ignition?

Rev,

I hope you and yours are well.

Under Ignition Settings, in the Ignition Options/Wheel Decoder tab, the Spark Mode pulldown has Miata 99-05.

Are the three connectors hanging off the front of the engine still in place? If so open them up and look for corrosion. Water has a way of getting in there and raising hell.

What I meant to ask is - are you using the stock coils?

Yes. I've tried three different coil packs. All give the same behavior.

Here's another composite log. I cleaned all the male blades inside all the connectors I could get to easily with contact cleaner and a q-tip. I did not do the water temp sensor at the back of the block. I didn't try to clean the female coil connectors in the harness either. That's next.

Looking back, I did not have these problems before I did the engine swap, but they didn't happen right away. I want to ask again, if the wiring inside the insulation layer of the O2 sensor has been damaged by heat, could shorts there cause the engine problems? My initial feeling is no, since the O2 sensor is just a gauge and not part of the tune, but I'm going to cut away the outer insulation layer and see what I can see. I need to get the car up off the ground.

Thanks much for everybody's help so far. I'm sure this will become a much longer thread.

I really don't think this is a cam/crank sensor issue, your logs of when the car is running crappy don't show any sync loss. If sensor errors were making it run badly, it would be constantly incrementing the sync loss counter. Composite logs are only useful for debugging cam/crank sensor issues.

When it's running badly, the normal laptop logs show the AFR going very lean, even though the pulse width isn't changing very much. I can think of three possible reasons for this:

1) The wideband is lying. If the car is running closed loop (which it usually is at idle), then a bad wideband sensor will make it run badly, because it'll put in the wrong amount of fuel. OTOH, the pulse width value isn't changing all that much, so I don't think this is likely to be the cause. Also, if the wideband is lying when it says that it's lean, the ECU would be compensating by adding lots of fuel, which it doesn't seem to be in the logs. This would make it rich, and you need to go pretty rich before it will have trouble idling. It's much easier to make it run crappy by going lean.

2) The fuel injectors are not injecting the amount of fuel that's being commanded. This could be caused by bad fuel pressure, by a bad injector, or by bad wiring to the injectors. If you can lay hands on a spare set of injectors, you might try swapping those in. If you stay out of boost, you could even try this with your stock injectors, although you'll need to use the big adjustor to compensate the map (req fuel, I think?). IIRC you checked the fuel pressure while the car was running poorly at idle, right?

3) The engine is misfiring. When this happens, the fuel doesn't burn, which means that the unburned air and fuel are both blown into the exhaust manifold, where the wideband O2 sensor sees the air and indicates lean. Normally I would say this is the most likely, because there are lots of common failures in this system. Poorly gapped plugs, bad plug wires, bad '99 coil packs, all can cause this, but it sounds like you've swapped all of those already. Bad wiring could also cause this.

Bad wiring will make the car run like crap in a bunch of strange and non-predictable ways. If it started when the engine was swapped, then it's theoretically possible that some wire got pinched while installing the engine and partially broken.

--Ian

Ian,

Let's call this an aside.

My understanding has always been the O2 sensor has nothing to do with the way the engine runs--nothing to do with the tune. It's just a sensor and can't actually change anything. If it did change the tune, why would I need a MegaSquirt?

Megasquirt supports EGO correction, where the wideband reading is used in a feedback loop (closed loop fueling) to hit the target AFR. It is an option and may only be on in certain regions; some people do not use it at all as it is off by default with many base maps.

Depends on if you are running EGO closed loop correction. Ian assumed you are, and stated that it was an assumption. The O2 sensor does not make changes, but the MS certainly can make changes based on the O2 sensor's input. Same with MAP, CLT, etc.

I really hope you solve this.

The ECU can run in two modes.

1) Open Loop. When running in this mode, the computer looks at the two axis of the fuel table (MAP and RPM, generally), applies trim factors due to other sensors (coolant temperature, battery voltage, etc), computes a fuel amount, and runs the injectors at that level.

2) Closed Loop. When running in this mode, the computer does the same thing as above, except that it also has "feedback". It has a target air fuel ratio and it compares that to the actual AFR as measured by the wideband sensor and calculates an error. That error is then used to modify the fuel computed using the fuel table + trims, either adding or removing fuel to try to move the measured wideband value closer to the target.

So if you "tune" you mean the values that are in the fuel & spark tables, then no, the wideband output does not modify those values directly. It does, however, modify the way in which the computer interprets those values when it calculating how much fuel to actually inject.

There are a bunch of parameters in the megasquirt that enable or disable closed loop under various conditions. Generally speaking, most people set it up to run closed loop at idle or in cruise, but open loop in boost.

--Ian

I understand (or maybe I don't) that when things are bad, the OEM computer makes changes based on O2 sensor inputs. But, I also thought it didn't cause subtle changes to the tune. I also didn't think the O2 sensor was anything more than that with the MegaSquirt. However, this is all magic, smoke and mirrors to me.

Everything was so simple before boost...

My AFR/EGO Control pulldown is set at PID with a bunch of other settings. EGO sensor Type is Wideband,

In the spirit of my initial posts, could any of this be causing my problems?

I bow to everyone's superior intellect on this. I'm just trying to understand.

Thanks,

'PID' is closed loop. The 'P', 'I', and 'D' values are parameters to tune the way that the AFR error is fed back into the fuel input.

When closed loop is enabled, the ECU is constantly tweaking the looked up value using the error. However, when the measured AFR is close to the target, the error is small, and thus the correction value is also small. If you look at "EGO corr" in TunerStudio or the data log, it will tell you how much this control system is altering the fuel.

As the saying goes, Garbage In, Garbage Out -- if the wideband sensor is bad, then the closed loop will definitely screw up the fuel. You can test this fairly easily though, if you get the car so that it's running poorly at idle, flip it from "PID" to "No Correction" and see if it starts running better.

--Ian

Excellent. I'd like to try as many things as possible before I start taking her apart. It will probably have to wait till the weekend.

Thanks much,

Gents,

I sat with the car while she idled in the driveway, waiting for the rough running so I could engage No EGO Correction. The engine did run differently, but it didn't fix the roughness. I actually saw about a half point richer idle AFR with no correction. I took another log.

I had my non-contact laser thermometer handy. While she was idling, I hit the center of the straight pipe coming out of the flange of the Racing Beat header with the beam. Front to back roughly, 360, 300, 370, 410. The numbers danced around a bit, but I did it several times, measuring at the same part of the header each time and the numbers were consistent. When she started running rough, the numbers went to 370, 330, 410, 380. Does it mean anything?

I'm shooting in the dark, I know, but the next step is to start checking wiring, which means I have to pull the top of the intake manifold and remove the injector harness. I'm willing to put that off as long as possible.

Thanks,

It's interesting you would see a 100*F difference in temps between cylinders. What do your spark plugs look like? Any variations in color between them?

As a diagnostic measure, I would be inclined to switch locations with the number 2 fuel injector and another and see if the low temperature follows that injector.

Have you run a compression or leakdown test on the engine recently?

IR thermometers don't work very well on polished stainless steel, they can give widely varying results depending on how dirty/clean the surface is, level of polish, etc. So I would be hesitant to draw many conclusions from that data. To get good data you need a thermocouple, ideally an EGT sender in each of the two tubes that you can read simultaneously.

(has to do with something called "emissivity", which the IR thermometer assumes is 1.0, which is a good assumption for most things except bare metals).

--Ian

Ian,

I can't refute your point, but I have to ask... could it give widely varying but consistent results? That first inch or so coming out of the flange is not (or no longer) polished, and the color is pretty much the same for all the cylinders. I measured each cylinder a bunch of times, and the readings were consistently as posted.

Regardless, I pulled the O2 sensor to check the wiring underneath the melted insulation. There is no visible damage to the wiring. I was concerned that maybe the internal wiring was shorting and causing some weirdness in the MS3, and since the problem was getting worse, I thought it could be explained by the wiring touching the header causing a little more damage every outing. That doesn't mean the O2 sensor isn't bad, but I no longer have concerns about that melted spot. I've had the O2 sensor go bad before, and it didn't affect the way the engine ran, only how the gauge displayed. This could be my opportunity to upgrade to the new version of the Bosch sensor, but there's got to be a way to figure this out without replacing everything one piece at a time to see if the problem gets solved.

As to Sixshooter's question, the newer used bottom end produces a noticeable increase in power, along with significantly reduced oil consumption and smoking. I also replaced the valve seals when I swapped my rebuilt head onto the Japan market block. The current configuration is so much better than her former self when she's running right I can't believe there's anything wrong with compression or leak down. Neither of those would be intermittent.

I'm on board with moving the injector from the #2 cylinder. Can I swap in injectors of a smaller size to see if she doesn't run rough at idle without changing the tune? I have some OEM injectors, and a set of unused Deatschwerks 350s that came with the Kraftwerks kit.

Thanks,

Is the damaged wiring in the wideband O2 sensor or the factory narrowband?

If they were dirty/discolored/etc in exactly the same fashion, then yeah, you should be able to compare numbers between them, but it's possible that they might be sufficiently different to throw off the reading without it being visually apparent. I don't know.

For injectors, if you swap in a known-good set of smaller ones and adjust the big number in TunerStudio (req fuel, I think it is? I don't remember) by the relative amount, and configure the proper dead time for the new injectors, then you should be able to start and idle the car fairly easily.

--Ian

There was no damaged wiring, only melted outer plastic insulation. The asbestos heat shield beneath was still intact, and the individual wire insulation was unharmed. I only have the wideband--no factory narrowband sensor.

I meant to more simply switch locations between the number 2 and the number 1 injectors and see if the cooler temp follows it. Don't get crazy with changing the whole set and introducing other variables.

I understand. I was just thinking if I move the #2 injector and the temps also move, and the car still runs rough (which it should given no other things are changed) then the next step, besides checking the injector harness wiring, would be to change injectors, especially since I have some on hand.

Another approach would be to just send the set of injectors to be tested and rebuilt. I've used Marren before. While the injectors are off to Marren, I can be checking wiring. However, those Flow Force injectors are not even two years old. Could an intermittently bad injector could cause the engine to die?

I would think that an injector that is bad enough to cause that would have roasted that piston under boost by going lean by now.

I've exchanged some emails with Nigel at Flow Force. Great guy. He thinks it might be a bad adapter pigtail to an injector. I'm awaiting parts.

Here's a screenshot of the AFR/EGO Control pulldown.

Anything on there I should change?

Thanks,

You have controller authority limited to +/- 3 percent. That seems pretty small to me, the wideband loop is barely allowed to do anything. I'm not sure what's normal, but mine is set at +/- 30%.

--Ian

Alrighty then... I assume the car should be tuned with that set to Zero, and then some correction percentage is chosen. Is that right? I read the book by Matt Cramer, "Performance Fuel Injection Systems", and I reference the MegaSquirt website often, but I haven't found the kind of help I usually need. Here's my tune.

Is there anything in the AFR or VE Table that's heinous--that would prevent me from progressively increasing the EGO correction after I get the car actually running right?

Thanks much,

You don't need to change that value to tune -- just turn on VE Analyze Live and let it do its thing. It can read all of the current fuel data from the megasquirt, back out the correction factors, and figure out what to put in that cell.

The EGO control authority limit is to stop the MS3 from doing something truly stupid if the wideband sensor fails. Arguably, 30% may be too large a value for it (since if you were at 11:1 and knock off 30% of the fuel, it's now dangerously lean). I don't know what most people run it at, and I'm not sure where the 30% in my table came from, I don't recall changing it, but I might have.

I don't think the control authority limit is causing the problems you're seeing here, it's just something I noticed when looking at the table you posted.

--Ian

I was referring to an initial dyno tune. It seems to me you'd want that Off. Regardless, I've posted my tune before for other reasons. You are not he first person to comment that 3% was low. But, I assume (although I could be wrong) that value has never been changed.

Typically people tune fuel on the street and then use the dyno for doing timing. Dyno time runs about $2-3/minute, and there's no real advantage to doing it there if the car is streetable.

--Ian

This is what I use. 15% in idle is to work out hot restart. Still, it is likely more than I need.
Small numbers in boost make that region practically open loop, which is OEM common. I may make them a bit bigger in the future, as I did have an unexpected AFR fuel cut yesterday. Then again, my AFR Error table is pretty tight.
The low kPa are small because, when trailing throttle, where's the danger, AND, i was tuning that section manually and did not want any confusion.


https://cimg6.ibsrv.net/gimg/www.mia...1ed894bd21.png

So, my philosophy, not yet compete, is that 30% is too much and 0% is too little. It also may be, that if one wishes to set VE and leave it alone forever, then you need the EGO to account for crappy winter gas (low BTU) vs good summer gas, and that can need either long term trim activated, or more range on the EGO correction. I'm under the impression that winter to summer is about a 3.5% differential.

Another safe thing to do is to tune on winter gas and let things run fat in the summer, but that will give poorer gas milage. Track car, you probably don't care.

Sir,

I cannot find that table in my version of Tuner Studio.

Oh, yes I see. What Firmware Revision are you running on your MS3?

1.3.4 is pretty good since you are not running EBC. Otherwise Reverant is recommending 1.4.0.

Also, I'm not sure this is your issue.

Your MAT Air Density Table is messed up:

https://cimg9.ibsrv.net/gimg/www.mia...34edc4ef91.jpg

In that you are not at 100% where you likely tune. However, the relative numbers are not too bad, so, though a bad tuning practice, I don't think this is your problem.

Do you have an intercooler? If so, where is the MAT sensor?

I didn't see anything else out of place in your tune.

Yes, I have an intercooler. The GM temp sensor is in the aluminum tube just below the 90 degree silicone connector going into the throttle body.

If I were to slide the numbers down so 100% lined up with, say, 110 degrees, would everything get 4% richer than it already is?

Thanks,

Yes. Basically, I'm saying that where you tuned (typically around 70F) should have initially been set to 100% along with the rest of the curve. Then, as other MAT temperatures affected AFR, those points should have been adjusted, leaving the base at 100%.

Now that you have tuned everything, it would likely not be worth the effort to redo.

If you were a purist, you would multiply the whole VE table by .96 (or whatever matches the present correction at the temperature where you did most of your tuning of VE), and then redo the curve. I'm not suggesting that, only observing.

Again, these are side issues. I don't see where the actual cause of your problem has been found. Intermittents are tough.

Nigel at Flow Force sent an install kit with the newer pigtail version. At first glance, they seem identical to the ones I had, but closer inspection showed the blades in the connectors to the OEM harness are much longer. I don't know if there are other differences. I installed those, as well as the new aluminum spacers that sit atop the three plastic isolators. The spacers add 2.5mm of height to the isolators. I assume the guys at Flow Force figured out the injectors needed a little more room. I repaired the burnt insulation on the O2 sensor wiring, and recalibrated the sensor since I already had it disconnected. I also swapped the locations of the #2 and #3 injectors. (See posts 94 to 101.)

She's still up on jackstands, and it's cold and rainy here. I elected to leave her up in the air and just let her idle. She ran smoother than ever. I ran her till the coolant temp got up to 192 and the fan kicked on. No rough running. I hit the header flange with the laser thermometer when she'd warmed up -- 340, 310, 340, 360. #2 was still and consistently the coolest, and #4 the warmest, but there wasn't the 100 degree difference like there was before. (I have no idea if those numbers mean anything. I had the laser thingy handy, and it gave me something to do while I waited.)

I plan to take her out tomorrow. Wish me luck.

Thanks,

I think that did it.

It's 74 here right now. I took her out for over an hour, about 50 miles. Slogging around in my little town, some 6800 rpm pulls, highway driving, traffic -- no issues. Running smooth and cool. I don't know what else I could do. If there were an HPDE tomorrow, I'd go.

I do have one concern. I was reading another thread about issues updating the MS firmware and I checked mine. TS says I'm running 1.2.0. I haven't updated the firmware since getting the box from Rev a few years ago. There never seemed to be a driving need to do it. I'm a don't-fix-it-if-it-ain't-broken kinda guy. Now, I'm running obsolete firmware. If I were 100% sure she was fixed, I'd take on updating. I went to the MS website and could not find any version change notes, so I don't know what I'm missing, but I need to hear back from Rev.

What do you all think?

Merry Christmas,

Awesome!

As far as software updates go, there's a big improvement in the EBC algorithms in 1.3.x. If you're using EBC, you'll want this upgrade, but be aware that you'll need to spend an hour or two retuning it once you do.

Personally, I'd wait til you're confident that the car is fixed before messing with a software update. Drive it for a month or so.

--Ian

It is the 1.4.x firmware that has updated EBC controls.

I would highly suggest upgrading to the 1.4.1 firmware. There are tons of very good changes that have come over the years. There was a thread recently though about another MSLabs customers upgrade process. I would take a look at that and follow in his steps while upgrading.

I've been keeping up with this thread, so congrats for finally figuring it out.

So in sum, it looks like your car had issues misfiring/giving fuel etc, just from one of the injectors not getting inputs correctly. I just got the FF kit in the mail. Hopefully the issue is gone. That sounded like it was nightmarish to diagnose. Hopefully your saga is over.

Wasn't that TorqueZombie?

My Friend,

I can neither confirm nor deny it was a single injector. After living with this for a while, I speculate it was sometimes more than one. Sometimes... it would run like crap but run well enough to get me down the road. However, there were a few instances where I thought I would need a tow. Once I was only about three miles from home, but I didn't think she'd make it. I'm hopeful the issues were all related to the injector setup.

I owe Nigel at Flow Force Christmas beers.

Merry Christmas,

DAMMIT, DAMMIT, DAMMIT.

She's not fixed. I decided to go through the idle adjustment procedure again and had the car idling in the driveway. Coolant temp was upwards of 180--I have a 180 degree thermostat and the big TSE radiator. I stepped into the garage to grab a beer and the car just stopped running. There was no indication of anything wrong. One second she was running, the next she was not. I restarted her and she was running like crap, just like before, like she was out of balance.

I would have bet a month's pay she was fixed, and would have gone to an event if there were one this late in the season. I was sure the problem had been solved.

I took a couple of logs afterwards. They're not long and probably don't reveal anything.

Thanks for taking a look and best wishes for a Happy New Year to everyone here,