[AlexE]

Hello,

I took my friends mx5 (2001 mk2.5 1.8vvt) to dyno to make test what can I achieve on stock car using my PnP board. And I found following issue: car runs unstable especially on high rpm, see attached dyno plots. Above 5000rpm there was massive instability.

First I checked #sync loss and found it is 0 all the time. And car runs really bad. Changing fueling, spark, advance, vvt parameters, dwell did not help at all.

came back home and started to dig in the logs. Only weird thing I found was RPM spikes to 10000+ on some dyno runs. But only on some, there was no good (I mean good measurements) dyno run out of 10 trials I performed

On second day I tested this car and also 2001 mk2.5 1.6 (non vvt) and on both found the same issue - cars were running instable, sometimes even stalling for few seconds (tachometer drops to 0, then comes back). And #synclosses still 0 on both. Tested both 202310 and 202402 firmware, the same issue

tried changing trigger filter settings, "resync every cycle" OFF, did not helped

Workaround was ... switching to semi sequential. Immediately both cars started to run smooth

I can provide logs and tune

[AlexE]

I uploaded logs and last tune

and dyno plots looked like this (also inconsistent)

mx5 1.8vvt problem.png (44.61 KiB) Viewed 481 times

[iltheo84]

Looks like the same issue I had on my MK1 (which also ran full sequential).
I never managed to find the issue and I have reverted to the stock ECU

[PSIG]

@AlexE, looking or clues. I see single misfires that are affecting your dyno data, but also unstable fuel PW and DC. Note the PW declines as rpm increase, and the traces are unstable. The way to diagnose this is to look at the data, then test anything that looks weird. Fuel looks weird. Pressure, electrical, something else? Lots more clues in there, but these jump out.

Fortunately, sequential has little to no effect at higher loads and rpm, so it won't hurt performance at all to run in semi-, quasi- or paired batch modes looking into these issues, if necessary. I'll also add that while everything looks generally smooth, the O2 corrections are working hard and at limits frequently. This indicates the target AFRs are not being reached as-expected at various points. This could be from a secondary issue, or simply the areas have not been tuned to-target yet.

After each change or test, run the logs through MLV to see how it affected operation. Take more clues from that.

AlexE_20240325.jpg (157.35 KiB) Viewed 362 times

[AlexE]

I took another mx5 (nbfl 1.6 with some intake/exhaust mods) to dyno and problem is the same. On full sequential there is instability, immediately cured switching to semi. See attached plots, only difference is full (bad) vs semi (ok)

nbfl1.6 semi vs full.png (37.33 KiB) Viewed 261 times

so there are 3 cars now, all runing ok on stock ecu, and all with the same problem on Speeduino and the same workaround. All 3 cars bad? Ok, they are all 20+ years old, but why all running ok on stock ECU and also on Speeduino when semi-sequential?

yearlier I thought there is something wrong with new revision of my board (bad routing causing noise on CKP or CMP signals for instance). Or my tunes were corrupted in a meantime. Ok, I know shit happens. But yesterday I used old board and very old tune, only I updated 202305 to 202402fw. And found problem to be identical.

during all testing I was using "enable per tooth timing" feature ON. Didnt tried is something happens when I switch it OFF. Maybe someone will try?




After succesfully finishing dyno session with some performance gain over stock

nbfl 1.6 mystery 2.png (259.18 KiB) Viewed 261 times

we drove for an hour or so fine tuning partial load and testing. No problem at all.


Im not sure I want to dig further this problem. I have to pay my money everytime I take car to dyno, spend my time etc ... having working solution is good enough. I dont have to know everything, I just want to have happy customers

[PSIG]

I imagine you are frustrated that it's running better than stock, but not how it should. If wanting some direct answers, I'd suggest using the Discord link (top of each page), where you'll find the project owner and some of the coders. Perhaps some answers there, as you appear to be speaking for other MX-5 users, and possibly more. Best success to you!

[AlexE]

oh God, Im an old guy who llikes forums, accepts Facebook but Discord seems to be still beyond my capabilities