When you click on links to various merchants on this site and make a purchase, this can result in this site earning a commission. Affiliate programs and affiliations include, but are not limited to, the eBay Partner Network.
Setup : 13b PP, with 2x 50mm itb's, 1300cc's ID injectors, IGN1a x4 running in wasted coil on plug. Trigger wise, i made a custom 36-1 trigger wheel.
ECU : Microsquirt V2 running MS2 Extra 3.3.3.
I am running it in pure alpha-n, 4cyl wasted COP because i found it to be easier to tune at the time, since i run no split at all.
RX8 timing table for ignition advance.
It runs great all in all, made 250whp in my miata.
However, accel enrich are still very difficult to tune.
To tune the upper left part of the VE table, i used mostly Tune Analyze live. It proved to be difficult since the PP bucks a lot under say 2200rpm.
Here is my question : with my accel enrich like it is now, it always goes full lean then full rich. I guess there is just too much fuel.
I think it is because of the VE cell values when i slam the throttle.
But the engine doesn't want to run if i lean this part of the table out.
Can you guys spot something fishy ? I attached my MSQ and a log from today.
Just looked at your tune and log.
I had a setup that was a semi-PP engine with 4x 43mm ITB's and had a similar issue.
Never got it really perfectly nailed, but here are some things I would try:
- I think I would switch from using 2 tables to just using 1 VE table, in my opinion a single 16x16 table are more than enough datapoints for a n/a rotary.
- I would switch from setting it up as a 2000cc 4-stroke, to setting it up as a 1300cc rotary with ignition set to COP-mode and then dialing in some split, around 10-15 deg at lower loads, tapering down to 5deg at higher loads. in my case that helped a little bit. Edit: Thinking about it, I guess on a microsquirt this would mean you lose sequential injection. So maybe you way to have it set up right now isn't that bad after all.
- VE table needs a little tuning and smoothing, in my experience a wideband lambda on a PP engine is of limited use. Whenever the engine is brapping and bucking the wideband is reading false. Also, I had issues with the sensor dying very quickly because of the raw fuel coming out of the exhaust. This is using bosch 4.9 sensors. The problem is that they don't completely die, they start reading false. I added EGT probes to help with tuning, I set the fueling so it would be around 900-ish deg C full throttle, and the lower load reagins I would try a lot of things and set it where it ran best.
I see you have closed loop lambda enabled, and in your log it's doing a lot of correcting. For a PP engine I would disable closed loop and run open loop (Set the control authority to 0%).
- I think the TPS based AE needs more tuning, I don't have my own tune here but I think my values are way different. This is a really good video on acceleration tuning that might help:
- Not sure what your future plans for this engine are but I used a microsquirt for around a year, then switched to an MS3 ecu, and with similar settings actually noticed an improvement, not sure what exactly what caused it, but it ran a little better. Maybe this is because it allows sequential injection and split timing.
- Dwell time on the IGN-1A can be increased a bit, I would try 5ms dwell. Might help igniting a sub-optimal mixture a little easier.
- Another thing that I notice is that you have set 4 squirts per engine cycle, so one injection event every 180 crank degree. Why are you not running 2 squirts per cycle as usual?
Acceleration settings are set in millisecond injection time, in your case since you are injecting twice per rotary engine cycle this gets added twice, so a little enrichment in the software is a lot of fuel in your case, especially with large 1300cc injectors.
This is why during a quick throttle increase in some cases your duty cycle shoots up to 130% in some cases, which means whenever you mash the throttle your injectors are going static and dumping in a ton of fuel. So I think you need to rescale your TPS enrichtment table and reduce the added ms a lot.
Turn off your ae when trying to dial in your ve table. It really screws things up. Just drive with smooth inputs and the lack of ae wont even be noticed. I fought and fought with my ve table untill i did thus. 30 min of driving later i was hitting my target afr everywhere.
I ended up switching to the eae/xtau stuff and found it works significantly better for an itb setup on a rotary.
we tend to need a lot of fuel off idle as we are basically 90-100kpa instantly.
Things i would do until you have a good ve table.
i would set your afr table or 13 everywhere
Id either disable AE or set the tps dot threashhold to something like 100-150
Why are there 2 ve tables? I thought that was only when you were running itb mode or dual table cofig? ve3 looks much better than ve1
50% sensor lag factor in the general settings seems pretty high.
engine and sequential settings look not right to me but im more familiar with ms3pro I would compare yours to aaron cakes rx7 guide.
logs also show your ve table is no where near ok. also noticed your duty cycle is crazy high.
have to head out but will look a little more when i get back.
Hello John, thank you for chimming in. my answers are in red :
Originally Posted by John Huijben
Just looked at your tune and log.
I had a setup that was a semi-PP engine with 4x 43mm ITB's and had a similar issue.
Never got it really perfectly nailed, but here are some things I would try:
- I think I would switch from using 2 tables to just using 1 VE table, in my opinion a single 16x16 table are more than enough datapoints for a n/a rotary. I couldn't tune the cruising part without ve3 enabled because of the fast changes in air flow around low angle TPS value. I needed more resolution up until 20%. To solve this problem, i will try the ITB mode, which basically made for setups like mine. But that includes welding a pipe on my intake runner, connected to a MAP sensor. Gotta find some time to do it.
- I would switch from setting it up as a 2000cc 4-stroke, to setting it up as a 1300cc rotary with ignition set to COP-mode and then dialing in some split, around 10-15 deg at lower loads, tapering down to 5deg at higher loads. in my case that helped a little bit. Edit: Thinking about it, I guess on a microsquirt this would mean you lose sequential injection. So maybe you way to have it set up right now isn't that bad after all. I use batch firing anyway. I just didn't want the hassle of setting up extra programmable outputs on the microsquirt. that's why i went with 0 split and wasted coil-on-plug. Due to the fact that i properly hid away my wiring, this is not going to change anytime soon. By the way, my first post is wrong. I am using ID 1050x's, not 1300cc anymore.
- VE table needs a little tuning and smoothing, in my experience a wideband lambda on a PP engine is of limited use. Whenever the engine is brapping and bucking the wideband is reading false. Also, I had issues with the sensor dying very quickly because of the raw fuel coming out of the exhaust. This is using bosch 4.9 sensors. The problem is that they don't completely die, they start reading false. I added EGT probes to help with tuning, I set the fueling so it would be around 900-ish deg C full throttle, and the lower load reagins I would try a lot of things and set it where it ran best. Thanks for that. I will consider adding EGT's in the future. Though the microsquirt isn't able to deal with them so that means extra gauges, which i am not fond of, but i understand how useless the WB can be whenever it braps or bucks. I have been lucky so far with AEM X-series gauge and bosch 4.9 probe though.
I see you have closed loop lambda enabled, and in your log it's doing a lot of correcting. For a PP engine I would disable closed loop and run open loop (Set the control authority to 0%). Closed loop lambda is only enabled during cruising, up until 40% TPS iirc.
- I think the TPS based AE needs more tuning, I don't have my own tune here but I think my values are way different. This is a really good video on acceleration tuning that might help: https://www.youtube.com/watch?v=HSE3CxaWF-I I will look into it thanks.
- Not sure what your future plans for this engine are but I used a microsquirt for around a year, then switched to an MS3 ecu, and with similar settings actually noticed an improvement, not sure what exactly what caused it, but it ran a little better. Maybe this is because it allows sequential injection and split timing. This is my track only car, and with a baby on the way and other cars to give TLC to, can't afford the time or money for an MS3 (which i happen to have in my other car, and it is awesome).
- Dwell time on the IGN-1A can be increased a bit, I would try 5ms dwell. Might help igniting a sub-optimal mixture a little easier. Will do that.
- Another thing that I notice is that you have set 4 squirts per engine cycle, so one injection event every 180 crank degree. Why are you not running 2 squirts per cycle as usual?
Acceleration settings are set in millisecond injection time, in your case since you are injecting twice per rotary engine cycle this gets added twice, so a little enrichment in the software is a lot of fuel in your case, especially with large 1300cc injectors.
This is why during a quick throttle increase in some cases your duty cycle shoots up to 130% in some cases, which means whenever you mash the throttle your injectors are going static and dumping in a ton of fuel. So I think you need to rescale your TPS enrichtment table and reduce the added ms a lot. Understood. Will try 2 squirts per cycle and report back.
- Are you sure the injector dead-time is correct? 0.595ms seems low, According to this it's almost double that: https://www.miataturbo.net/megasquir...ewidths-92738/ I took this number from Injector Dynamics themselves.
Turn off your ae when trying to dial in your ve table. It really screws things up. Just drive with smooth inputs and the lack of ae wont even be noticed. I fought and fought with my ve table untill i did thus. 30 min of driving later i was hitting my target afr everywhere.
I ended up switching to the eae/xtau stuff and found it works significantly better for an itb setup on a rotary.
we tend to need a lot of fuel off idle as we are basically 90-100kpa instantly.
ill pull your tune down tonight and look
Hello mikey.
AE were turned off (1000% threshold) when road tuning.
About why i used 2 VE tables, the answer is in my answer to John.
Can you tell me why my VE tables are no where near OK ? It is interesting because it drives very good as is, apart from AE of course. Thanks
I'll look at your map again and see what I was talking about.
Eae really helped me have proper acell enrich at those low rpm high load areas.
edit....
here is why i said your ve table isnt ok.
Im sure its drivable but in my non-expert opinion your are probably leaving a lot of performance off the table.
your VE table is way to rough.
Your duty cycle is all over the place. Im assuming the numbers is wrong, but you shouldnt be going over 80% duticycle
This scatter plot shows you have data all over the place also. you should see a mostly straight line going diagonally across the box.
Any time you use autotune it will do that. If you use autotune, let it do its thing, then go into TunerStudio, highlight the whole map, and hit the smoothing function twice. Then go back and go into autotune, and repeat.
You will still need to do some hand correction, especially on a high overlap engine that will not have an accurate O2 sensor signal below 50kpa or so.
Alpha-N is the style you should be tuning in, which is MAP x TPS in which acceleration enrichment is done via TPS. MAP signal isn't solid enough as per Peejay.
That's not what I said.... infact I have used MAP on peripheral and bridge ported engines no problem. Rotaries have sharp enough intake pulses, and only two to choose from with enough overalp between the two, that it works just fine. Need to do some mechanical smoothing not just for the MAP signal but also so the fuel pressure regulator is happy, but I've found that stock port engines need this too for a stable MAP signal. I use a piece of carb cleaner straw epoxied into a vacuum coupler. Works a treat. It DOES put a small amount of lag to the MAP response so I use TPS-based accel. It works okay at low RPM and at high RPM it doesn't matter.
What I said, was due to the rough nature of a high overlap engine, you cannot rely on the oxygen sensor to have any useful input until you have the engine loaded down enough that it isn't misfiring. So you basically can NOT use closed loop at all, and you can not use autotune at loads below 50kpa or so, or at RPM below 2000 or so. You will have to tune these areas by feel. The engine might be getting 12:1 but the wideband will be reading much leaner because of all the raw oxygen in the exhaust due to the stuttering brap that rotorheads love so much.
I kind of want a MAF to make this easier - measures air directly, outputs fuel correctly.
When you have your fueling table correct, you'll find that you don't need very much acceleration enrichment at all. I do miss the "simple" accel enrichment from the original MS firmware, it was a little clunky but it was also very easy to work with, especially if you had your box positioned so you could see the LED as you were driving.
Any time you use autotune it will do that. If you use autotune, let it do its thing, then go into TunerStudio, highlight the whole map, and hit the smoothing function twice. Then go back and go into autotune, and repeat.
You will still need to do some hand correction, especially on a high overlap engine that will not have an accurate O2 sensor signal below 50kpa or so.
Thanks man, that is solid practical advice right here. The car is running good with the VE table as is, but it could run a fair bit smoother for sure.
What smoothing factor do you use though ?
@RGHT and peejay : that's why i went stright to pure alpha-n from the beginning. Barometric pressure doesn't change much where i live, and I don't go up the moutains to have fun. So alph-n is alright.
It could be better with the ITB mode down low, but it's a track car so it doesn't matter much.
@peejay : i got 2 questions for you now : smotthing will not solve my accel enrich problems. What do you think of my low rpm/ high TPS zone ? Do you find it relevant regarding AE formulas ?
2nd question : on overrun, below 3000rpm, the car bucks like crazy. Like undriveable without pressing the clutch. I chose to not activate the overrun fuel cut, because premix and no more OMP. IS it related to fuel or ignition advance, or both ?
@micky_d : wow i didn't see your answer. Your scatter plots - especially Duty Cycle vs MAP vs RPM - are relavant when your tune with MAP/rpm. This way you want a straight line alright. But not with alpha-n. Because of the sudden change in MAP signal, as soon as you touch the throttle.
Ok for the VE table smoothing, you guys seem to agree on that one lol.
About the maximum duty cycle, it is a bug. This engine uses nowhere near 100% in real life at WOT at 9500rpm.
Thanks for taking the time to go through my logs.
I didnt mean to say your scatter plots should look just like mine, i was just illustrating that not much linear stuff is happening in your tune. Values should generally go from low to high.
i have a similar setup, that last 20% of tuning is challenging.
A mistake i made early on is to tune ve and screw with ae at the same time. What happens is you get a **** ve table that presents as an inconsistent driving experience, meaning ae works in some areas and not in others. Your afrs will wildly swing in your logs and you wont be anle to correct it.
That bucking you refer to is likey you are likely lean. As peejay mentioned below 50kpa youre wide-band lies. If your auto tuning it will keep pulling fuel on decel. Just free rev tune that area and lock those cells. Stuck a box fan in front of your radiator so you can hold rom and let things stabilize.
05-05-19, 06:51 AM
