Miata Turbo Forum - Boost cars, acquire cats.

Miata Turbo Forum - Boost cars, acquire cats. (https://www.miataturbo.net/)
-   MEGAsquirt (https://www.miataturbo.net/megasquirt-18/)
-   -   Long term trim experiences? MS3 (https://www.miataturbo.net/megasquirt-18/long-term-trim-experiences-ms3-84861/)

Zaphod 06-15-2015 12:56 PM

Long term trim experiences? MS3
 
Has any of you guys experiences withe the Long term trim stuff in the newer MS3 firmwares?

I would like to see if or how this works for you.

patsmx5 06-15-2015 01:23 PM

I read about it and they say to get the fuel table dialed in well before turning it on. Thus I have not enabled it yet. But it looks pretty sweet, and I will use it once my fuel is spot on.

cyotani 06-15-2015 02:19 PM

<p>It should act like VE analyze autotune always running in the background and adjusting your long&nbsp;term fuel trims based on your short terms. Every modern OEM ECU has this feature and it's nice to see it being added to the MS3. No direct experience yet but I'll be sure to enable this and see how it works out.&nbsp;</p>

18psi 06-15-2015 02:21 PM

I'm planning on messing with it, whoever beats me to it post up, I too find it interesting

JasonC SBB 06-15-2015 07:01 PM

Does LTT adjust the entire fuel table, or a single number?

gjsmith66 06-15-2015 08:49 PM

RTFM

:idea:

Will have to give it a try if I ever upgrade the firmware...

richyvrlimited 06-16-2015 04:32 AM

I'm not sure I see the point. I fairly regularly tweak my VE table , (say once every 2-3 months), anyway due to changing other components, or generally trying to improve the tune. LTT is to take into account sensor drift/clogged up injectors etc which takes quite a while - years I suspect.

DNMakinson 06-16-2015 08:18 AM

I read the manual through once, but it was hard for me to interpret what was automatic and what took intervention to make the adjustments active.

What I am doing now is trying to tune for perfection with winter gas, and let things go rich in the summer.

For me, the LTT would take the place of that, and seasonally adjust.

Of course, EGO adjusts now. My thought is that if the O2 sensor goes out, I'd rather error rich than lean if everything defaults to the table values.

That being said, I do not have limp set up and am not sure what the MS3 would do if the O2 sensor did go bad. I think it already ignores a non-input and defaults to 100% EGO, as it should.

Neither EGO or LTT will work properly if the O2 sensor drifts and gives a value, but a wrong one.

18psi 06-16-2015 10:27 AM

I think only rev's CAN units make it default to something like 14.7, having the wb wired directly it will cause the car to do all sorts of weird things if it goes out, unfortunately.

But yeah, I basically have it exactly like that right now, needs about 10% to be able to compensate between 60* and 120*

Ben 06-16-2015 10:46 AM


Originally Posted by 18psi (Post 1240855)
I think only rev's CAN units make it default to something like 14.7, having the wb wired directly it will cause the car to do all sorts of weird things if it goes out, unfortunately.

No, MS3 has sensor error checking as a standard feature. EGO sensor failure detection will cause the MS3 to disable EGO and flash CEL code 7.

18psi 06-16-2015 10:50 AM

Oh really? well that's cool

Reverant 06-16-2015 11:06 AM


Originally Posted by Ben (Post 1240871)
No, MS3 has sensor error checking as a standard feature. EGO sensor failure detection will cause the MS3 to disable EGO and flash CEL code 7.

See, there's the problem. The AEM for example has a lower limit of 10.0. This means that if you unplug it (power off, burnt fuse, analog out disconnected), the MS2/MS3 will hapilly report an AFR of 10.

If you enter 10 as the trigger value, then any time that you hit 10 (not difficult on a boosted car with a less than perfect tune), the MS3 will light the CEL. Not ideal.

JasonC SBB 06-16-2015 11:22 AM

WBO2's don't do a slow drift error in output like NBO2s sometimes do?

DNMakinson 06-16-2015 11:58 AM


Originally Posted by Reverant (Post 1240883)
See, there's the problem. The AEM for example has a lower limit of 10.0. This means that if you unplug it (power off, burnt fuse, analog out disconnected), the MS2/MS3 will hapilly report an AFR of 10.

If you enter 10 as the trigger value, then any time that you hit 10 (not difficult on a boosted car with a less than perfect tune), the MS3 will light the CEL. Not ideal.

Actually, a CEL is not so bad if, as you say, the tune was "less than perfect" as a root cause.

Then, your recommendation, or preferred implementation is? LC-2 and a limit of < 10?

I think a CEL and a limp mode to ignore O2 information is better than a false 14.7 that EGO will then try to enrichen. Well, except my EGO limits in boost are pretty narrow. Still at 14.7, I will get fuel cut due to AFR protection. OK, that's not so bad. Puts me into a default Limp mode.

All these possible settings and interrelationships.

DNMakinson 06-16-2015 11:59 AM


Originally Posted by Ben (Post 1240871)
No, MS3 has sensor error checking as a standard feature. EGO sensor failure detection will cause the MS3 to disable EGO and flash CEL code 7.

I'll have to look at how I have mine set. I am using some of the error checking features, but not all.

richyvrlimited 06-16-2015 01:58 PM


Originally Posted by Reverant (Post 1240883)
See, there's the problem. The AEM for example has a lower limit of 10.0. This means that if you unplug it (power off, burnt fuse, analog out disconnected), the MS2/MS3 will hapilly report an AFR of 10.

If you enter 10 as the trigger value, then any time that you hit 10 (not difficult on a boosted car with a less than perfect tune), the MS3 will light the CEL. Not ideal.

the CEL stuff you can put a deviation amount in too, a permanent flatline of 10:1 with no deviation you're pretty likely to have an issue of some sort.

That said the documentation yet again sucks. The deviation numbers have no explanation of what they relate to.

Ben 06-16-2015 04:05 PM


Originally Posted by Reverant (Post 1240883)
See, there's the problem. The AEM for example has a lower limit of 10.0. This means that if you unplug it (power off, burnt fuse, analog out disconnected), the MS2/MS3 will hapilly report an AFR of 10.

If you enter 10 as the trigger value, then any time that you hit 10 (not difficult on a boosted car with a less than perfect tune), the MS3 will light the CEL. Not ideal.

That's true, but not the whole story. The ECU is looking for a minimum value and lack of fluctuation. If these events are met, the ECU determines faulty sensor and disables EGO correction.

If Sensor Validation is not enabled, you are correct -- the ECU would read 0V and attribute that to the 0 volt AFR value. Best way to combat that is limit your EGO authority to a reasonable value. No different than an OE car.

Something else that you can do. Program your wideband analog output to 0:5V :: 1.5 lambda:.5 lambda. This is the inverse of Innovate's default curve. Then if you had a sensor failure, the EGO would react to a lean condition, no different than what happens with an OE ECU and the narrowband goes out. EGO authority should be limited to a reasonable amount, no matter the curve or sensor type.

Reverant 06-16-2015 05:49 PM

Last time I checked, the AEM and the Zeitronix can't be adjusted, and the Innovate can't inverse the curve's direction. You can change the slope, but can't go from rising to falling.

Ben 06-16-2015 07:25 PM


Originally Posted by Reverant (Post 1241028)
Last time I checked, the AEM and the Zeitronix can't be adjusted, and the Innovate can't inverse the curve's direction. You can change the slope, but can't go from rising to falling.

I will check on the Innovate. That would be lame if so.

DNMakinson 06-16-2015 08:28 PM

Ben / Rev: Can you discuss the effects / choices of combining the serial CAN box and sensor error detection?

In other words, if using the CAN input, if the sensor goes bad, does it indeed send 14.7 (Lambda = 1) to the MS3, and therefore sensor error will not be triggered?

Reverant 06-17-2015 03:09 AM

The module is configurable. That means you configure what the digital output will be on either warmup or error. I have 14.7 for warmup and 7.5 on error, with minimum reported value set to 10 (this means that if the wideband reports lower than 10, the module will report 10). My errror trigger value is set to 8, so it will actually only trigger on an error.

DNMakinson 01-03-2017 04:36 PM


Originally Posted by JasonC SBB (Post 1240715)
Does LTT adjust the entire fuel table, or a single number?

So I am reviving this as I plan to play with it some. I have a set of questions on the MSEXTRA forum, but I get answers there on about half of my inquiries.

It appears to me that LTT is going to work similar to VEAL with manual or automatic writes. It also appears that it is writing on a bin by bin basis (256 different corrections).

To me, this is not as useful as a single number, created by some weighted averaging. That's because my use for LTT is to adjust for Summer vs Winter fuels.

I don't want only cells that have been recently touched to get updated. If the Soichiometric AFR has changed in the fuel, I want the entire fueling to adjust, not some of the cells.

Still, I have set up EGO to not come into play until 175F, as I do not want corrections during warmup to be recorded in the LTT table. I'm not wanting adjustment for bad WUE. That should be a separate issue.

I have also set EGO authority to "0" all across row for 27kPa, and then normal from 28kPa up. That way LTT will not try to adjust over-run cells. It will, however, try to update idle cells. This may not be needed as time there is brief before Fuel Cut occurs, but that is how I set it up. May change it later to see if that is an important set-up parameter.

I will post up some LTT tables:
Start will be all zeros.
Then see what happens after some drive time, with no write to Flash.
Turn off car and re-start and see if table 1 goes back to zero or not.

Settings (other than mentioned above on EGO parameters):

https://cimg9.ibsrv.net/gimg/www.mia...f076b5e4a0.png

DNMakinson 01-03-2017 07:06 PM

So, after driving home, this is Table 1:

https://cimg5.ibsrv.net/gimg/www.mia...8a750c0fbe.png

I suppose it turned RED because one cell changed (by -0.2%). As you can see, this is with the car still running. After I turned it off, then KOEO, the same table pulled up. Tomorrow, I am taking a different car to work, so I won't get a further update.

Unfortunately, my tune is such that only the one cell changed. I'm thinking I will change the Stoichiometric AFR from 14.7 to 14.4 so everything will be about 3% rich. Then we can see what happens with a poor tune or a changing system.

cpierr03 06-21-2021 01:50 PM


Originally Posted by DNMakinson (Post 1384539)
Unfortunately, my tune is such that only the one cell changed. I'm thinking I will change the Stoichiometric AFR from 14.7 to 14.4 so everything will be about 3% rich. Then we can see what happens with a poor tune or a changing system.

Necroing this a second time.... @DNMakinson, did you go any further in testing LTT? I'm not finding much data out there on it.

DNMakinson 06-21-2021 02:11 PM

I did not. Seemed complex, and hard to tell what state things were in. I think a good tune removes the need as well.

As an aside, (the original subject of the thread) I also do not use CEL for WB02 sensor checking... Why, because every time fuel cut comes on, the AEM goes to steady 18.2, which triggers CEL due to no fluctuation. Have not found a way around that.

WestfieldMX5 06-22-2021 04:27 AM

Maybe use a loop function to trigger the check engine light with 2 conditions (lower or higher than target) and duration for X seconds.
Not even sure if that's possible 'cause loops make my brain protest ... just thinking out loud.

DNMakinson 06-22-2021 07:44 AM

I’ve not found a way to include loops into CEL functionality.

deezums 06-22-2021 11:56 AM

More half-baked code from MS3. LTT and CEL both.

Someone could fix the CEL in the code, easy as adding a flag to cancel CEL flagging while in overrun or whenever EGO is conditioned off, but to do it they'd also need to step about 6 years back in time and through about 6 years of bullshit basic ass bugs and shitty code.

DNMakinson 06-22-2021 12:54 PM

@deezums Correct. I live with whatever is in 1.4.1, because I don't want the newer stuff. They finally fixed a LONG running bug where limp mode did not properly lower either boost cut or rev limit (I forget now which it was, but I won't upgrade to get the fix.

I suppose I could learn to code, and modify 1.4.0, but, really, what I have works Okay for me.

How is your Haltech coming? Is it treating you well?

deezums 06-22-2021 01:24 PM

Before I bent two rods it was going ok, still out of commission for the time being unfortunately. Didn't really have time to get into the fancier features just yet. Wish I had more time to play with the stock engine first, but I got tired of hitting boost cuts on the binding turbosmart wastegate and stupidly just raised the limit to "fix" it.

I know I loved the WUE taper, I have mine linked to RPM. No more deciding between running pig rich or idling wicked lean while cold.



All times are GMT -4. The time now is 07:07 AM.


© 2024 MH Sub I, LLC dba Internet Brands