For any discussion not specifically related to your project
By jackscr
#51262
Hey guys, it's me again... unfortunately .. can't get a break...
Had this problem a while ago when the car felt like it was running on 3 cylinders ..out of 4.. and going really lean... it came out of nowhere and was gone just the same... nothing changed same settings everything... completely forgot about it for a couple of months... and now something similar where it is going slightly lean 1 AFR point almost can't feel it at normal driving except idle where it starts to break up as it does not like over 15:1 Afr.
First time I did not get a chance to troubleshoot because it basically happened one time and went away.. But this time I went straight and swapped the ecu for the old one and everything is fine... swap the speeduino and it runs rougher and a little leaner but not as bad , maybe 0.5 AFR (it's very inconsistent but it's not going away this time) .....
Speeduino is on 0.4.3 board running seq injection and wasted spark
The old megasquirt ecu is running batch injection and wasted spark

Could this still be the injectors acting up even tough it runs fine on the old ecu (but being batch fire it somehow masks the problem)? Or is there something wrong with my mega 2560 board? ( I ordered another mega board for a spare and will test this)

Would posting the tune help any? I have to stress that there was no change made in the tune to cause this... the car ran fine in the morning and than in the evening this happened... :cry:
By jackscr
#51268
theonewithin wrote: Tue Jun 22, 2021 8:48 am For any troubleshooting a copy of tune and a log during the issue is needed.

There is basically 0 way to know what could be happening without.
The problem with that is I did not get a chance to log yesterday when it was really bad and today magically it's better as I've said still lean especially when I rev it a little it goes crazy lean and stays there and before it would be to target.
Anyway here is the log and the tune
Oh I already said this but besides the Lean AFR the rpm at idle is not as stable ,before it was really nice the idle ignition advance was keeping it nice and stable
Attachments
(80.94 KiB) Downloaded 146 times
(1.61 MiB) Downloaded 144 times
By theonewithin
#51272
Why is your AFR correction enabled when your tune isn't finished yet?

Turn that off...

Everything requires your VE table to be finished as everything either adds or removes from that.

So finish your tuning with your AFR sensor disabled.

Then once tuning is complete, you then enable the sensor and tune it's PID values so it does it's corrections effectively.

AFR target table doesn't look right either...

You are aiming for 18:1 AFR around cruise????

More to add...

Injector correction isn't done either...

You need this stuff sorted before you get into your tuning.

You have WMI enabled???

Might want that disabled until your tune is finished...
By jackscr
#51276
theonewithin wrote: Tue Jun 22, 2021 11:44 am Why is your AFR correction enabled when your tune isn't finished yet?

Turn that off...

Everything requires your VE table to be finished as everything either adds or removes from that.

So finish your tuning with your AFR sensor disabled.

Then once tuning is complete, you then enable the sensor and tune it's PID values so it does it's corrections effectively.

AFR target table doesn't look right either...

You are aiming for 18:1 AFR around cruise????

More to add...

Injector correction isn't done either...

You need this stuff sorted before you get into your tuning.

You have WMI enabled???

Might want that disabled until your tune is finished...
Wait wait wait.... the VE is fine, where I actually tuned it :lol: and that is where it has ran so far ,and it has been running ok.did not have time to tune everything yet. It has run like this for some time and it ran FINE... until yesterday when it went nuts ... did not touch it... same VE same AFR table and spark !
And yes aiming for lean cruise used that for 5 years now getting 52mpg at cruise... don't worry about it :D
WMi is enabled and runs fine thank you very much... that is the reason for the wonky VE table up top... when WMI enables the VE needs to be lower...and I ran two boost levels ,without wmi around 160Kpa and and over 180kpa wmi kicks in... again this is how I've done it for years it's fine don't worry.

Injector correction , you mean the voltage correction? Completely forgot about that..but again it has been FINE had no issues...
I told you the LOG is not going to show anything as it just looks like a little lean because that is what's happening.... but it has nothing to do with the tune... it's some hardware of software issue...but what to test?
By jackscr
#51278
Oh yeah now I know why I forgot about the injector V corr, because I did not know what to fill in that table, other than directly from the injector manufacturers I don't know where to get that data from... in the MS ecu I used 0.2ms/V and that worked great...how can convert that to % for speeduino?
By jackscr
#51281
Ok just realised that the correction value for the injector voltage was set to % of Open time not total PW... :shock:
I adjusted the values for aprox 0.1ms/V relative to the 0.8ms dead time that was measured at 12.8V
Do not know how to determine the actual values without doing injectors deadtime tests at differing voltages
Any suggestions are welcomed .

Thank you
Attachments
Untitled.png
Untitled.png (83.23 KiB) Viewed 4350 times
User avatar
By Mr_Hypnotic
#51283
Search the site. This has been talked about before. For example:
PSIG wrote: Tue Feb 02, 2021 10:28 pm You could, but we try to test for accurate numbers to apply, and with minimal testing. Do a search for testing injector latency. Save a copy of your tune so you can either reload where you were with a click, or update your tune with better numbers. ;) Some set the rpm a bit high (stable area), disconnect the MAP reference from your regulator (if yours has one), set for a consistent Lambda in the area (no change if MAP or rpm changes a bit), then switch number of injections per-cycle. When you do this, you are adding additional dead times (latency) with every added injection, so 1 to 2 squirts per-cycle doubles the applied dead time. If the dead time setting is wrong, the fueling will change the Lambda.

The Lambda will oscillate, but read generally rich or lean if not correct. You can simply read if the Lambda changes and adjust, going back and forth as both will change, which works but takes a lot of time. Or, you can read the pulse width each time the Lambda crosses your tuned value in your logs, and do a bit of simple math for a specific result of dead time setting at the current voltage. You can imagine how useful logs would be for this. <hint> Again, do a search for those guidelines.

To begin your important :!: voltage corrections table, load the electrical system to lower the system voltage, and do it all again for a new number at the new voltage and convert to a percentage of your first one. Extend those values to create your corrections curve. I have a modified 70A server power supply for this type of stuff if I want a full set of accurate 10V - 16V results quickly, but the simple DIY methods will get you in the ball park with some work.

To answer the question you didn't ask - yes, it's worth it, on every project. However, you won't know how much grief it saved you unless you were already suffering. :lol:

[EDIT] PS: your fixed-pressure regulator isn't helping your situation. If idle or lean cruise are not stable after this, I'd go for the MRFPR next.
By jackscr
#51289
Don't know why I asked for suggesting because I know full well there is no going around measuring the actual dead time ... but I hate the Squirt method for testing it... I will hook up my old ecu and do a proper dead time test at varying volts as I should have done in the first place.
I kind of doubt this will solve my issue even though there is a slight voltage variance I saw in the logs 14.5 was a few days ago and today when testing it was 14.1.... does that work out to 1 AFR point leaner at cruise? who knows ... will find out after injector tests :D
By jackscr
#51327
OK so measured dead time at various voltages and set the correction curve .
Retuned idle a little bit and it worked great all the way to stable running temps. SUper... NOT ... Shut the engine off...left it to sit until the CLT got to about 60C and when I restarted it barely ran and almost died immediately ,gave it some pedal and managed to keep it running until the WB warmed up to display AFR and it was lean of course... but this time it is breaking up allot more and I can hear it missing either from the lean mix or the ignition.
So replaced the spark plugs (even though it runs fine on the other ecu). still the same...
When I measured the dead time I checked all the injectors and they flow and work perfectly.. so it's not that...
What could cause this kind of behaviour ?
I attached the log from when it was running fine and fully warmed up and a log from the restart and running like s***
got a new mega 2560... maybe I should swap it...IDK I'm getting desperate here...

Thanks for the help so far guys I really appreciate it ! ( I would have completely missed that V correction :)
Attachments
(80.97 KiB) Downloaded 141 times
(828.37 KiB) Downloaded 144 times
(252.99 KiB) Downloaded 140 times

I think possible reasons for that would be: 1) loo[…]

It looks like you have a fuel supply issue. readin[…]

Will this have an updated version about this featu[…]

Perhaps some different points and pictures. Instr[…]

Still can't find what you're looking for?