Lost with Megasquirt

Wierd post/rant :

I got my car running on Megasquirt this summer. It was my first time doing such a project. The install/hardware part went well. The tuning aspect is now driving me nuts. The car now runs semi-good... Meaning I can drive it, but under certain situations it will act up. For example, one day it will run right and feel crisp. Park the car and come back 1 hour later, it will idle lower and the throttle response will be worse than it was. It will misfire at idle, etc. But nothing was changed in the ECU during that time, and the outside temperature is the same. Then I can park it another hour and come back, everything is normal. I'm at a point now that I have so many questions, I see so many parameters that can be tweaked that I don't even know where to start. I look at logs and it's more puzzling than anything. I know the MS extra forums are pretty good, but you need to ask specific questions. Sometimes, tuning to solve one problem might cause another one. I read a lot of documentation and forum posts, but I feel like it's either too vague (I see things like "simple: just tune the fuel map and after tune AE") or too detailed... I feel like I'm 85% there but I don't know how to get to the last 15%. Anybody has a good resource for this? Any instructions or general guidelines? Something like "you want to achieve this and have a log which looks like that. X value is generally too high and Y is too low". Even better, any one on this board in the province of Quebec who could lend a hand (or a brain)? Thanks GRM.

Matt Cramer shows up on this forum sometimes. His book is really good, it's called Performance Fuel Injection Systems or something like that.  

My first impression- your issues sound related to idle enrichments, which are temperature dependent. Hot restarts are notoriously challenging. Learn how PID controllers work and start changing one variable at a time. The last 5% of tuning takes 90% of the time.   Don't use autotune in the idle cells of the fuel map, those cells need to be adjusted manually. 

ShinnyGroove (Forum Supporter) said:Learn how PID controllers work and start changing one variable at a time. The last 5% of tuning takes 90% of the time. 

Yes, I need to Learn about the PID Controllers and the last 5% is totally true.

Greg

Can you post some good and bad datalogs along with your tune?

"Inconsistent" can be a bad ground issue somewhere.

"Affected by heat" can be your IAT placement.  Are you running factory everything, having only changed the ECU, or is this some cobbled together mess like I would do?

Without a log its hard to tell, but good when parked and not when restarted shortly after is likely heatsoak in the intake temp sensor. You can flatten mat related air density past normal air temps (100f) to combat this. 
 

I also remote tune/teach and have 20 yrs ms experience. 

What kind of car is it?  How heavily modified is the engine?  Turbo?  IRTBs?  Big injectors?  E85?

In addition to the heat-related stuff mentioned above, if you're using large injectors then getting the dead time numbers correct is vital.

You guys are awesome. I'll tell you more. The car is a 1992 Mercedes 190E which had CIS injection originally. So all the original hardware is removed. I think I did a good job with the "integration" part. It looks OEM-ish:

Bone-stock 2.3 8V inline four apart from a 2.25 exhaust and header. Ford Thunderbird EV1 30# injectors. Saab 9-3 fuel rail. Walbro 190 lph pump. MS2 running batch fire injection and speed-density. Ignition is with a 36-1 wheel and Ford VR sensor. 4 LS D585 coils in wasted spark mode. Stock Bosch idle valve driven with the MS. The ebay electric fan is also driven by the MS. The stock ECU does not control anything anymore.

IAT heatsoak is a possibility. The throttle body is mounted vertically on the intake plenum. I put a piece of steel tubing on top of the TB with the filter on top. The IAT is the classic brass GM type. All that metal probably gets hot. Maybe if I made that tube out of aluminum and used a plastic sensor, it would help? Anybody know a source for a plastic 3/8 NPT sensor? Do I really need to mount the sensor in the intake stream? Could I mount it upside down in a plastic or phenolic "bung" in the air filter housing? See, that's four of my 1000 questions 

I didn't know Matt Cramer wrote a book. I'll definetly find it. It should help a lot.

I'll write more about the "inconsistent" part tomorrow. I'm too tired now. Thanks guys.

Gm makes an open element sensor that uses an open plastic cage. They are about $22 down here not sure if they are easily to get locally. They are normally used on boosted setups but I use them on everything as the quick response avoids this issue. 

I've had a Megasquirt in my MSM for 7 years and the hot idle still isn't quite right.  Heat soak from when it sits for a while after being driven seems to mess up the normally working tune.  To compensate, the idle is setup fairly rich until it's up to normal temperature.  At least it still doesn't stall occasionally.  It's pretty frustrating but I just haven't had time to revisit the tune.  One thing that has helped quite a lot on the TR6 is that it doesn't have a PWM idle valve.  That one thing has made it so much easier to get a working tune.

If I was closer I'd love to help, but I'm still not such an expert anyway.  I can certainly empathize.  Hopefully Matt will see this and offer assistance.

There are super fast IATs available that make the open element IAT look slow.  One of the turbo guys on rx7club has been experimenting with them.  (And they come from a GRM advertiser)

I have found that a lot of the weird tuning inconsistency issues I have had over the years were due to poor injector characterization.  If you don't have your deadtimes spot on, the fueling will skew at different temperatures, since the IAT calculation applies only to the "real injector pulse" (full pulsewidth minus deadtime).  That is why I sucked it up, bit the bullet, and bought some ID injectors for my Megasquirted car.  I didn't just buy them because I knew they weren't junk like my Injector Connection injectors were (1200cc injectors: one flowed 850cc, one flowed 750cc) but also because they come with precise injector data.  That data is worth MANY hairs on your head when it comes to tuning.

One reason hot idle is challenging is that the injectors themselves can heat soak and the dead time values actually drift a bit, depending on the injectors. Usually this calms down once cool fuel from the tank has run through them for a few minutes. This issue can be worse with returnless fuel systems. 

Pete. (l33t FS) said:

There are super fast IATs available that make the open element IAT look slow.  One of the turbo guys on rx7club has been experimenting with them.  (And they come from a GRM advertiser)

 

I have found that a lot of the weird tuning inconsistency issues I have had over the years were due to poor injector characterization.  If you don't have your deadtimes spot on, the fueling will skew at different temperatures, since the IAT calculation applies only to the "real injector pulse" (full pulsewidth minus deadtime).  That is why I sucked it up, bit the bullet, and bought some ID injectors for my Megasquirted car.  I didn't just buy them because I knew they weren't junk like my Injector Connection injectors were (1200cc injectors: one flowed 850cc, one flowed 750cc) but also because they come with precise injector data.  That data is worth MANY hairs on your head when it comes to tuning.

I'd repeat both Pete and ShinnyGoove's posts, but am only literate to do one...  Since Pete is a little more detailed...

Hot Fuel Handling is a pretty common problem that is dealt with- if you have two logs that are identical in terms of inputs, and the outputs are so different that it's a real problem, then it's likely that it's a fuel system problem- where the injectors and rail get so hot that it can vaporize the fuel.  In addition to what they have already posted- make sure that under hot conditions you are getting enough fuel pressure- which will help in eliminating bubbles in the fuel.

I know tuning can be confusing- but the key is to get what you are asking for.  And if you are not, take data to understand who is moving the ship to the wrong path.  

I agree with Pete on the deadtimes, at least with large injectors. On a 30# it really won't matter as long as you leave the regular batt v correction in place and are .9-1ms deadtime. You are in a pretty linear region and far up enough it can easily be tuned around if it's a little off. 

See if you can get data logs of a good idle and a bad idle under similar circumstances. You can overlay these in MegaLogViewer and see what's changed between the two - perhaps the MAT sensor reading is higher in one than another, or it may be something entirely different.

Paul_VR6 (Forum Supporter) said:

I agree with Pete on the deadtimes, at least with large injectors. On a 30# it really won't matter as long as you leave the regular batt v correction in place and are .9-1ms deadtime. You are in a pretty linear region and far up enough it can easily be tuned around if it's a little off. 

Yes, the shorter the length of the idle pulse the more the dead time is an issue.  Practically speaking this is most significant on engines that make a lot of power with single injector -- usually that's either high boost turbo ones or rotaries.

That said, EV1 injectors are old and dead times were not as well controlled in 90s as they are now.  What's your target power level for the car?  Google says the OEM horsepower was 130 and that only needs about 17-20 lb/hour, so 30 is a lot of headroom.

The best thing I ever did for my MegaSquirted car was invest in a set of flow-matched and tested EV14 injectors.  In a couple of days of fiddling around my idle was more stable than OEM.

In reply to Paul_VR6 (Forum Supporter) :

....admittedly, of all of the cars I have worked with, not counting OE-like tune touchups, the smallest injectors were 60lb-hr.  What you say makes total sense because with smaller injectors the deadtime isn't half the injector pulse...

Regarding OP, I remember that my Impulse Turbo had a cooling fan for the injectors and fuel rail that would blow for a certain period of time after a hot shutoff.  I think some Audi 5000s had a similar device as well.  I wonder, if you are dealing with a fuel injector heatsoak issue, if something like that may help.

If you let it burble and fart after a hot restart, does it eventually settle down, or does it continue to run like poo til the next key cycle? 

Any hot start situation always makes me think of boiling fuel, whether its a quadrajet or a turbo Ferrari.  Do you have residual pressure holding in the lines when its hot?

In reply to Paul_VR6 (Forum Supporter) :

I'm not sure if it's a good or a bad thing for me to discover this, especially seeing that you're actually pretty local. 

I'd start with logs to make sure there isn't a sensor value going nuts. Once you've spotted what's different, you can deal with it. If nothing shows up in the logs, then you start looking at the real world outside the computer :)

That's the nice thing about diagnosing an ECU. You can talk to it, unlike a carb. It's like the difference between being a doctor and a vet :)

This may not help as much as you'd like, but you have to approach this as if you're test engineer.  Because you are.  Log everything, and by log I mean keep a notebook (real or virtual) of every change you make and the corresponding test log from driving the car.  Change one thing at a time.  Eliminate as many variables as possible, and then gradually add them back in.  Save every tune so you can go back in time if it seems to be gradually getting worse. 

I have a few minor issues with my MS'd 190, but it starts every single time, and runs and idles nicely 98% of the time.  That other 2% is a little maddening, and it may not be the tune at all, but something else that's broken on a 30-yr old car with a selection of parts from probably about 10 different mfr.  

There is no high HP goal with the car. I just wanted to "update" it from CIS and learn how to tune. Plus I needed a way to drive the ignition with the manual swap (no magnet on the flywheel, so not compatible with the stock ECU). It'll stay at 130 HP or so. I had the injectors laying around, unused from a previous turbo VW project. I had them ultrasonic cleaned and tested by a shop and used them. The test report says they all flow the same at 330 cc/min.  I figured it would be better to have 4 identical injectors instead of chasing issues that might have been caused by different flow values between all 4 injectors.

Just tonight, I did a bit of research on dead time and injector characteristics. My injectors are Bosch 0280150756 (T-Bird and Cougar 3.8 S/C, some Mustangs and Syclone/Typhoon).

According to this table, which is supposed to be a book figure, they flow 303 cc/min:

http://www.injectorcleaning.co.uk/flow.htm

According to this other source, which is measured test data, they flow 321 cc min:

https://www.motormanfuelinjection.com/Fuel_Injector_Flow_Rates.html

And according to the test data when I had them cleaned, they flow 330 cc min... I guess I'll use this number, I just found out I had 310 in Tunerstudio.

Then I tried to find the specified dead time for this injector P/N but couldn't find it. Seems like Bosch made a few similar models though. From 2 sources, I found out that the dead time should be around 0.46-0.4 ms at 13.2 V, not 0.9 ms!

https://www.ms4x.net/index.php?title=Fuel_Injector_Deadtimes#Bosch_0280150945 (I used the data for the 0280150945 that I interpolated to 13.2V)

https://www.phearable.net/information/tech-area/injector-dead-times.html (I used the "Mustang turbo" data, seems similar)

Also according to those sources, the battery voltage compensation should be 0.1 ms/v. I plotted the values in a graph and we are in a linear region:

At least with this I should be a little closer. I will try that. I will also install a fuel pressure gauge and see what happens in the rail during a hot restart. Funny thing is that the car had originally had a lot of hot start problems with CIS. There is a stock fuel cooler using an A/C refrigerant line coiled around the fuel line. Since my A/C was broken, I had to remove that. Now, at least it can do a hot start, but the idle will be lower for a little while. If it's 35C outside, I have to feather the throttle a bit.

Now one question about fuel mixture and catalytic converters: I wanted to play nice so I installed a cat in the exhaust. I know rich mixture can damage the cat. I set up my cruise mixture at 14.7 so the cat works.  But if I idle at 13.7:1, will that damage the cat? I can't really lean it out, otherwise the idle isn't smooth. 

Some of you talked about a plastic IAT sensor. Do you have a P/N? I will revisit my IAT installation. I guess under ideal conditions, the IAT during a hot restart will be about the same as when the engine was shut off minutes before? If it's higher, the sensor is heatsoaked?

I took two logs last night. A bad one, followed by one power cycle and it gave me a "good" log. Can't figure out why for now, but I will study the 2 logs carefully. Drove the car to work this morning and came back in the pouring rain. The car behaved beautifully.

One last thing. I remember PID tuning from when I was in engineering school, mechatronics class. I was pulling my hair off my head! Tough course... Digging through my memories to know what each of the 3 parameters do, I tried to set up closed-loop idle last Sunday. After a bit of fiddling, it worked! The car held a steady idle at 950 RPM. Thinking I had tamed the beast, I went to fuel it up. When I started it again, it would stall every time I pressed the clutch! I turned it off. Do I need to put set the "Closed duty" at the minimum duty I know the engine won't stall or lowest idle I want to get? The engine stalls if the valve is fully closed, around 35%. Or do I need to set up the throttle body so the engine idles with the valve fully closed at that 35%?

Thanks!!

There is probably a switch on the clutch and one on the trans that indicates neutral. Given the way your car is behaving, I think your ECU is missing the trans signal. You put the clutch in and it thinks it's in closed throttle decal (ie, fuel cut) instead of idle. Just a guess. 

Rich doesn't kill cats the way that raw fuel (misfires) do. That said, you shouldn't have trouble getting it to idle at stoich. Idle can be a pain, keep at it. I think you can use ignition timing to cheat. 

Rich doesn't kill cats, rich makes the cats "go to sleep", which is a good chunk of why the OEMs do power enrichment.  The rest is keeping the exhaust valves happy.

This is how some modern turbocars run stoich even under high boost - they have better cooling in the exhaust area and good materials in the valves, and modern converters are a lot more tolerant of heat.  I am sure that the temperature drop across the turbine helps too.

I like to set up throttle position so that it will do a no-load hot idle with the IAC completely shut.  This allows a kind of fail-safe, and also gives you more range for idle control when the engine is cold, or the A/C and electric fan is running, or - my favorite - on a cold start when the A/C is on because it is like 140 degrees in the cabin and you need you some refrigeration happening ASAP.

Those injectors have real deadtime in the .9-1ms range. Listed values are usually just the solenoid on/off or off/on time and not the same thing. Surprised the batt correction isn't closer to .2ms but see if it works. I usually tweak that by unplugging the alt and letting the batt drain and tweak this so afr stays steady as the voltage drops.

On the IAC/closed loop what kind of valve? There are a few different ones with very different behavior, the big Bosch ones even have two styles that act different. Some are full open at 0 and close at 30 and back full open at 100, so min duty needs to be 30. It can be you are in a weird spot in the valve motion and the closed loop gets confused.