MS3X NA6 - "Misfire" at high rpm related to Timing Error?
 

I installed my DIYAutoTune MS3X earlier this year and have been street tuning the VE table as time allows. I’ve autotuned and tuned via logs from 40 to 80 kPa and have been working on 80-100 kPa recently. I think I’ve gotten about as far as I can with tuning fuel until I sort out this issue...

After the first startup on the MS3X and setting base timing, I went through some trial and error setting the cam and crank input pots on the MS3 and Extension boards when stationary with a warm engine. “Good enough” was revving the engine in neutral to 6k rpm producing no misfires, though 6k to 7k still had a misfire or two with the tach swinging wildly when it happened…like tach input was lost momentarily. I continued tuning load fuel below 6k and gradually increased to 7k w/ ample fuel in those ranges, but the “misfires” continued, not always at the same specific rpm either. Every time the engine accelerates under load above 5.5k, this happens.

Logged the issue (quite repeatable), and from what I can tell, it seems to be evident in RPM and Timing Error. There are a couple blips in the attached log file at 2716-2717 sec in RPM (6400-6600 rpm) that don’t follow the tach needle movement 1:1 (needle swings maybe +/-500 rpm). You can also see the change in RPM from RPMdot in the same area of the log considering TPS was 99-100% through the occurrence.

I vaguely understand due to my Ignition Options/Use Prediction set to 1st Derivation Prediction, the ECU is predicting ignition events, and the difference between predicted and actual is Timing Error. Since it increases linearly with RPM, any error I have at idle/low rpm gets exaggerated near redline. Is this enough to cause these misfires?

Spark plugs are still fresh and properly gapped. Coils checked fine when I had a no-start issue. I will double-check resistance on CAM IN, CRANK IN, and SENSOR GND wiring, though I recall 0-1 ohms on all adaptor harness wiring through the MS3/X harnesses.

Setup Details
• `90 Chassis, stock internal 1.6
• I/H/E, AFM-delete, sequential fuel, stock coils/CAS
• DIYAutotune Pre-assembled MS3X (not PNP) V3.57
• MS3 1.3.4 Release Firmware
• MS2E v3.2.1 MSPNP2-MM9093 base map MegaSquirtPNP by DIYAutoTune.com (in lieu of finding a genuine MS3X 1.6 base map)
• 64-pin DIY adapter connector (MS3/X DB37 to 26- and 22-pin factory harness plugs), wired per Frank’s guide
• AEM 35-4110 UEGO

Any guidance would be appreciated. (.msl and .msq attached)

I have no idea which pots you are playing with, but I don't think those matter at all.

You're problem is probably having the car run so damn rich for one. If you don't have a flawless VE table you shouldn't be running EAE either.

Have you checked spark latency, because that's about all you can do to correct high RPM timing error if I'm not mistaken. No care what TS says the error is so long as it's actually firing right.

These were the pots I adjusted based upon using the composite logger. DIYAutotune published some decent instructions for setting them.

I do see above 6k, I'm 12.0 or less at 95-100 kPa so ~.5 more rich than target. I've been working on this and a few other areas over the last few days.

Per DIY's recommendations, EGO Correction is off and AE is 100% TPSdot while tuning. I think EAE came from Frank's base map, maybe...

Spark latency...I tried increments of 5 usec from 0 to 20 when VE was rich in these areas, no improvement. I may slowly increase this if leaning out 95-100 kPa cells >6k doesn't remove the misfires. Did verify w/ timing light that timing was suddenly retarding 5-10 deg w/ Fixed Timing.

Thanks for the sanity check, I'll update after more tuning time...

If he's observing wild swings of the tach needle when this occurs, then that indicates this isn't a mixture problem. It's either a problem with the CAS pickup (hence the pots he's fiddling with) or an ignition problem (as the tach on the 1.6 car is driven by the igniter module.)

Check the datalogs. If this is a CAS pickup problem, you'll see a very sharp glitch in the RPM indication when it occurs (you can tap the spacebar while logging to insert a marker that's easy to find later. Hit it when this misfire occurs.) Given that old CASes and the default input circuits in the MS are known to not get along well, my bet is that this is where you'll find the problem.

Stop screwing with spark latency. That's not the problem.

why did you have to go through trial and error on the pots?

sounds like you didn't know how to set them correctly and still have yet...

Based on my experience tuning multiple MSes on multiple NAs, there are some CASes that just can't be made to work reliably with the old comparator-style input circuit, no matter where you set those pots.

On my first Miata, this resulted in me manufacturing a crankwheel setup. On subsequent cars, I learned of the MAX 9926, which is the chip that DIY now uses on the MS3Pro series.

I've never had an issue, and I use the VR circuits exclusively for the crank/cam input. Built at least 50-75 units that way, never heard one complaint. (only initially when I setup the cam pot incorrectly with NB sensors)

not discounting that the max9926 is much better.

the only thing i notice in the logs is that the your decel enrichments are pulling fuel by 5% in the time stamp you mentioned, RPMdot & MAPdot pretty spikey at that point--nothing else stands out.

are you sure your "misfire" isnt just the accel enrichments false triggering, pulling some fuel, and you feel it?

Quoting the OP:


If the tach needle is suddenly dancing like Elaine from Seinfeld, that's not a fuel problem. That means that the coil isn't firing, either because the MS didn't command it to or because there's some electrical problem with the igniter.

Igniter problems don't typically manifest themselves only at certain RPMs, nor would the OP having adjusted the CAS input trigger pots have made the situation better.

The facts in evidence lead to the conclusion that the ECU is losing input sync. When I have time, I'll install MLV on this computer that I'm at and take a closer look at the log.

im looking at it from excel. lost sync on the log is 0.

Coils can fail to ignite if the mix is wrong, and if you aren't driving the tach with the megasquirt it will certainly dance. My stock coils did it all the time, even without a lot of boost.

MS2 doesn't have adjustable pots, and I've never had a problem. I only get lost sync hits when the laptop is plugged in.

Hmm. I'm looking at the log now, and I also don't see the sort of RPM glitch associated with lost sync.

I hate it when a really nice-sounding theory gets proved wrong by something as trivial as reliable data...

mkrazit, where in this log did the problem occur?

msq and logs both look good as far as i can tell here.

I saw a rpm glitch and no lost sync, but that was a week or so ago.


Joe:
2716-2717 sec in RPM (6400-6600 rpm)

Deezums has it -- 2716-2717 sec, 6400-6600 rpm. May be other areas, too, but I'm sure about the time mentioned. (Keep forgetting spacebar for marking...) There's not much of a loss of rpm in the data vs. movement of the needle. I'll try to log this again in the morning and video the tach movement.

Logic tells me it is ignition related. Original CAS and ignitor, AFAIK. I'll confirm again they check to the FSM when I have a chance. The more I thought about the rich mixture, the more I'm doubtful it's a fueling issue; 11.7 AFR should still ignite w/ good spark.

UPDATE 10AUG2015
 

It’s been a little bit due to family obligations but I managed to tune the VE table more this past weekend after adjusting AFR targets again (still learning). The misfire I'm experiencing has enhanced itself to an intermittent and complete cut of spark and fuel, likely due to RPM signal being lost momentarily. Seems like this would be related to the high-rpm cut-out because of the tach behavior when this occurs. (The other high-rpm hiccups are still there, too.)

I was able to log this yesterday…complete cut occurred at about 1370 sec, before the marker. RPM drops to 0 but RUN = Y at 1370.216, then 1370.284 RPM = 1 (?) and RUN = N. TPS and MAP were still active.

https://www.dropbox.com/s/ynhyqwwo5i...utout.msl?dl=0

(Sorry for the Dropbox link, short on time to read how to crop/edit my log file to get under the 4.77 MB limit.)

Other "lighter" misfires (the kind that started this thread) were 1473 - 1476.

Before this log, I checked the coils; checked out fine per FSM. Pulled the spark plugs; nothing abnormal there. Checked the igniter; couldn’t check the mega-ohm values (e.g. tach output signal), but all other resistances with respect to ground checked out against known good values I have in my files. I’ll see if I can borrow a DMM that can check for the ~5.26 mega-ohm value on the tach output signal pin of the igniter before I rule it out. (Seems odd, though, that this issue came up after the MS3X installation…if the igniter was failing, would the MS3X make its failure more obvious?)

UPDATE 12AUG2015
 

Tested the igniter at work today w/ a Fluke 289 DMM capable of reading up to 30 Mohms. (See link for known good igniter values in Post #21) Terminals VB and TAC were reading ~500 kohms, far off from the ~5.25 Mohms for known good.

I'll be ordering a used igniter tomorrow from one of the Miata community's recyclers and will report back. Still uncertain how this crept up after the MS3X install but we'll see what happens w/ a known good igniter installed...

UPDATE 21AUG2015
 

So I've got to question the known-good values I found (I think from Bill Strohm?) because the used 1.6 igniter I received checked out the same as my original igniter... At any rate, no change in behavior w/ tach output or misfires at 5500+rpm, and not losing sync… (see attached screen capture last nights log...markers indicate tach events)

Dropbox link for tach event/misfire video: https://www.dropbox.com/s/ut29gcymmv...12209.mp4?dl=0

I also changed the blue NGK wires that were installed few a used set of Magnecors I had in the basement to rule out more hardware...

Because this issue started immediately after installing the MS3X, I still suspect I don’t have the trigger and hysteresis voltages set correctly to get a good rpm signal above 5500. I didn’t have these set 100% when the MS3X started the car for the first time, so I got them close to start tuning fuel. I have no measurement equipment except a multimeter, so I set the voltages per posts from MS3X users after setting them per the MS3X manual by turns (unsuccessful) – I’ll see if I can dig up those links for reference. (I’m sure my neighbors wanted to beat me after revving to 7k repeatably!)

Is there an easier way to reliably set the trigger and hysteresis pots?

UPDATE 14SEP2015
 

Since I last posted, I removed the MS3X and measured input and output voltages for the VR conditioner circuits per DIYAutoTune's tech article. I ordered and received a JimStim to more readily facilitate this. Input voltages appear fine to both the Crank and Cam inputs, however, output voltages, after following the instructions to check output voltages (pin 2) to pin 3 of the U7 op amps, I concluded that since I was not getting near the output voltages specified, both U7s must be faulty. Unless I somehow flubbed where I probed w/ my DMM, but I triple-checked DIY's instructions...

Contacted DIYAutoTune, been working w/ Matt to rule out possibilities. I've mentioned the output voltages a couple times; it wasn't said but seems we should rule out TunerStudio settings first (understandable). A change in dwell was suggested, more -- I had been running Nominal Dwell = 3.5 ms, Spark Duration = 1.0 ms from Frank's basemap. Running MS2PNP settings from DIY of 4.5/.7 (Nominal Dwell and Spark Duration, respectively) made the misfires occur just over 3500 rpm, though I did adjust the VR pots while making the dwell changes, so not sure which contributed more. Again, running stock 1.6 coils and igniter.

Going to grab a composite log to see if the signals are proper and try higher dwell tomorrow on the way to work...

UPDATE 16SEP2015
 

Attached is a data log from my commute home today w/ markers (late in log). This was w/ Nominal Dwell Duration = 4.5 ms and Spark Duration = 1.0 ms. The engine was not happy above 3k rpms, very lumpy and unsmooth in power delivery, didn't want to rev. I know it shows I'm lean near the markers, but I'm not so sure that isn't a side effect of what's happening w/ the ignition. Everything I see says I should be running ~5 ms dwell w/ the 1.6 coils. :confused:

Also attached is a composite log from this morning's commute, easy acceleration.

This behavior of the engine w/ an unsmooth powerband/loss of torque/feels like spark retard was much more mild and higher in the powerband prior to pulling the MS3X to check VR circuit input and output voltages. As I mentioned, I did adjust the pots to get what I could on pin 3 of the U7s, but the output on pin 2 was non-existent. I understand we're checking if software settings are contributing to the issue I have, but since my issues started immediately after the MS3X install and adjusting the pots after first start and no-load fuel tuning began to eliminate the misfires and allow the engine to rev more freely, I can't help but think time should be spent there...

UPDATE 25SEP2015
 

I've been in contact w/ DIYAutotune for some help on this issue. I've confirmed the following:

• Igniter is good
• Coils are good
• Spark plugs gapped at .035" or .039" make no difference; not fresh but hardly worn
• Two sets of plug wires confirmed good
• TDC mark on pulley aligns w/ piston 1 TDC using socket extension
• Base timing checked again, confirmed at 10-deg BTDC
• Dwell of 2.5-3.5 ms appears to move torque loss/misfires above 4k rpm
• 1.4.0 firmware made no difference (was 1.3.4)

I continued to play w/ dwell settings as a part of troubleshooting. Reduced dwell duration to 2.5 ms, keeping spark duration between 1.0 and 1.5 ms, and added fuel between 2.5-4.5k where it appeared slightly lean before. These settings seem to be closer to how the engine was running before removing the MS3X to measure and adjust the VR circuits (14SEP update post)...but it's still misfiring...

Tach behavior during "misfire"
(Dropbox link)

While confirming base timing, w/ TunerStudio still in Fixed Timing = 10 deg, I slowly rotated the throttle plate w/ the timing light still on the crank pulley to determine if timing was retarding. I was expecting a gradual change if it truly was retarding as rpm increased, though what I saw was a sudden step-change. The crank pulley TDC mark suddenly ADVANCES from 10-deg BTDC w/ Fixed Timing (900-2600 rpm) to 0-deg TDC between 2700-3300 rpm. I checked this last night and also saw that the mark went back to 10-deg BTDC from 3300-4100 rpm then started retarding ~5-deg at ~4100 rpm (spark hardware latency = 20 usec).

Crank pulley
(Dropbox link)

The .msq's spark map doesn't have any sudden changes that would retard suddenly through the powerband. If this retardation of spark timing is occurring when the spark map is being used, that would explain the loss of torque I've noticed along w/ the misfires. Trouble is TunerStudio only logs what spark timing is commanded vs. actual, otherwise it could be useful to log spark timing through these occurrences.

VR Circuits
I haven't ruled out the Cam and Crank inputs yet. Testing (can share measurements if needed) showed the output pins of the U7s on each input NOT providing the voltages DIY cites (mentioned in the 14SEP update). Is this worth investigating??

UPDATE 03JAN2016 (The FIX!)
 

The warmer and not-so-snowy weather has been a treat and allowed for additional troubleshooting of this misfire issue I've documented here.

Discussions w/ Joe Perez and DIYAutotune left me w/ the symptom of a poor CKP signal causing a loss of spark at above 4.5k. If the signal was bad from the stock CAS, then it needed to be filtered. I'm not into making circuits, so I arranged a Dual VR v2.1 from JBPerf (Jean Belanger). It arrived last week, so this past weekend, I reviewed the latest (2015-10-12) MS3X V3.57 hardware manual as Jean's instructions detailed needing a pull-up resistor for the input, which sounded similar to what I had read in the MS3X setup documentation (https://www.diyautotune.com/tech_art...azda_miata.htm).

I determined that my DIYAutotune MS3X did not have the internal 1k pull-up at R57 on the main board, so I installed one early December since Jean's instructions also called for it.

I reinstalled the MS3X w/o adjusting the VR pots and logged this:

https://www.miataturbo.net/attachmen...ine=1451856612

Since I was losing sync above 5k, I was fairly certain I wasn't meeting the voltage threshold required to create a CKP pulse to fire the coils. Adjusted Main and Expander pots to get 2.5V w/ 5V applied (at top of R54 and ZC test point, respectively), I got the following (log was huge, so just a screenshot):

https://www.miataturbo.net/attachmen...ine=1451856612

Nothing crazy, just shows pulls to 7000 rpm…but after months of not being able to do this and not being able to figure out why, it feels like a small win!

This allowed me to continue tuning the VE Table (as you can see, I had conservatively WAY too much fuel above 5.5k) and start exploring closed-loop idle tuning and AE for throttle tip-in.

I appreciate the help along the way in troubleshooting this issue. The solution forced a much deeper understanding of the MS3 and how it triggers the Miata’s coils via the CAS, but in the end, the idea was to better understand engine management…so another win!

Key words: MS3, MS3X, high-rpm misfire, no tach signal, no rpm signal, no sync loss, R57, pull-up resistor, VR circuit, crank input, CKP

You have a sync loss. Can you post a composite log of it?

I definitely had sync loss from the 1st screenshot in yesterday's post. This was w/ previously set crank and cam VR pots set to what gave me the most rpm range before the "misfire" and w/ only adding the R57 pull-up. Once I reset the pots per the MS3X hardware manual, I no longer had sync loss and could increase rpm to rev limit (2nd screenshot is R57 pull-up w/ pots set per hardware manual).

The curve colors are washed out on those screenshots, but I have not logged any sync loss since the R57 + setting pots. I simply wanted to document what happened to me here for anyone in the future.

I had a similar issue, turned out to be a slack timing belt