PDA

View Full Version : Megasquirt is running!


Cloud
July 15th, 2009, 08:58 AM
So I finally got the time to put into all of this. I will say it hasn't been easy. My car for one reason or another is notoriously hard to start on e85 to begin with so getting an acceptable low end tune has been a nightmare but it will idle (poorly) without stalling and drives just fine. There were about a million little things I had to go through to get it working properly but everything seems to be alright now and ready to go through the tuning process.

The closed loop idle control is awesome, there is no way my poorly running car would work without it. I had to do a LOT of work tuning the PID controls on it and the step sizes etc... since I am not running the stock isc in the stock configuration, but rather as a bipolar instead of unpolar stepper motor.

One thing I am most interested in is the timing map and getting that worked out for E85. I'd like to hear from you guys about how you handle E85 cold starts, for instance how much extra timing you run at startup, and how much additional fuel you add, etc... If someone has a really solid running timing map I would love to see it.

Still need a solution for a knock sensor signal amplifier... that will come later with boost.

Cloud
July 26th, 2009, 05:11 PM
Well, definitely making progress with it. Took a look at the base timing and it was off 15 degrees :rolleyes: That helped. Getting a stable idle is still a bit difficult but it is getting quite a bit better. Warmup is the hard part. Found out that tps signal noise was triggering accel enrichment, that was pretty annoying for tuning. All in all it seems like a VERY tunable platform, you just have to know what every one of the million parameters is and you have that much more control.

I do find it odd that the car idles best at 15-17 degrees of timing though considering stock was about 8...

drcustom
July 27th, 2009, 09:39 AM
I do find it odd that the car idles best at 15-17 degrees of timing though considering stock was about 8...

When you're setting timing you have to ground a reference that holds the timing at base. Because of that stock timing may be actually be something closer to what you're seeing!

I heard about MegaSquirt at a junkyard in Hawaii of all places a few years ago...haven't seen anyone local get it running yet but looks very cool and definitely tons of potential.

Did you buy a kit or build from scratch?

Cloud
July 27th, 2009, 10:41 AM
When you're setting timing you have to ground a reference that holds the timing at base. Because of that stock timing may be actually be something closer to what you're seeing!

I heard about MegaSquirt at a junkyard in Hawaii of all places a few years ago...haven't seen anyone local get it running yet but looks very cool and definitely tons of potential.

Did you buy a kit or build from scratch?

Well, stock timing while grounded is 5 deg, without grounding is around 8 or so, at least in my experience. At the idle speed i'm running it was around 10ish from the factory.

It is a very cool standalone. You *can* do things with coil on plug though it can eat up other features, and sequential injection still isn't really there, it's still mostly batch fire. So I assume there is some room for better idle quality from that in future iterations. The control over everything is just enormous though, you really have to know what you are doing with most of the values, especially the PID control, which can be applied to idle speed, o2 sensor feedback, boost control, anything that you want closed loop. It's my first real foray into a standalone setup.

And kit/scratch... yes. You still have to build it by scratch even if you buy a kit ;) I do wonder how many people in colorado are running MS setups though, it seems like I am in a VAST minority.

dsm_gsx97
July 27th, 2009, 11:16 AM
Anyone running a stand alone on a DSM is in the minority with ECMLink/DSMLink out there. :)

Congrats on getting things going. Running a stand alone for the first time isn't easy and figuring out how things work together is even harder. I know because I had to go through it too. I'm glad you got the car running and I am excited on where you're headed with it. Good stuff.

matthewdesigns
July 27th, 2009, 11:27 AM
I do wonder how many people in colorado are running MS setups though, it seems like I am in a VAST minority.

IIRC a customer of Jack's uses it on his 1g. You are the only other person I know of. Glad to know you've got it up and running!

prophecymiller
July 27th, 2009, 08:57 PM
Yes, good to hear it's working! Can you tune each cylinder with this assuming you have the proper testing?

Cloud
July 28th, 2009, 09:08 AM
Currently there isn't really enough processing to do feedback and control to individual cylinders. You can do coil on plug but that's about it. There should be a sequencer add-on at a later date you can use that will allow you to do all that but I haven't seen too much on when that is supposed to be available. Currently you can do 2 injector signals, and 2 o2 sensor signals.

Cloud
July 28th, 2009, 08:31 PM
Some news, reported injector deadtimes on the internet are decidedly false imo. Either that or MS reads a different conceptual value than what the internet uses as a standard. But, by switching back and forth between 1 injection per cycle, and 2 separate injections for the same cycle, you can tune deadtime. Ideally they will have the same AFR if your deadtime is right, so you can switch and play with deadtimes till they come together. This made my idle nearly perfect. This is aparently the key to idling larger injectors well, dead time has to be spot on.

Cloud
July 29th, 2009, 08:54 AM
Ok, figured out the deadtime thing. Reported deadtimes for dsmlink and such are deadtimes ADDED to stock deadtime. So, for instance my pte 1000cc injectors are listed at about .45ms, but the stock deadtime appears to be .765ms. Making total deadtime about 1.215ms. This is pretty close to the 1.15 I have been using. Good to know!

Cloud
August 3rd, 2009, 09:35 AM
Well, the quest continues.

Idle has not appreciably improved despite my best efforts. I may opt to go with a high idle and not use closed loop control for a while just to make it drivable.

Beyond that I have developed a couple of issues. One is scary. The rpm signal drops out at about 5500-6000 at WOT and backfires/goes lean.

Second is hot starts. I tried to start the car after shutting it off, IAT sensor was reading over 150 degrees and the car was not happy. It flat out refused to start despite changing fueling and other parameters.

Does anyone have a preference on where they put the IAT? I am thinking before the throttle plate in the end tank of my air/water intercooler might be ok to prevent heat soak but I wonder how accurate it is as well since it doesn't account for manifold heating... sigh!

D Walker
August 5th, 2009, 06:38 PM
Well, the quest continues.

Idle has not appreciably improved despite my best efforts. I may opt to go with a high idle and not use closed loop control for a while just to make it drivable.

Beyond that I have developed a couple of issues. One is scary. The rpm signal drops out at about 5500-6000 at WOT and backfires/goes lean.

Second is hot starts. I tried to start the car after shutting it off, IAT sensor was reading over 150 degrees and the car was not happy. It flat out refused to start despite changing fueling and other parameters.

Does anyone have a preference on where they put the IAT? I am thinking before the throttle plate in the end tank of my air/water intercooler might be ok to prevent heat soak but I wonder how accurate it is as well since it doesn't account for manifold heating... sigh!


I can help you with some of that if you can get it by the shop.

I always put the IAT in the IC pipe immediately before the TB. I find that if you put it in very thick material, esp aluminum, such as an intake manifold etc. it will heatsoak much quicker.

Feel free to give me a call if you like.

Cloud
August 5th, 2009, 08:54 PM
Well, I gave up on closed loop idle control and went to a manual coolant based step count and it works a lot better. It still bogs down when powersteering or fans kick on and it still has a higher idle otherwise but it doesn't stall and it works a lot better for driving at least.

The IAT... yeah I don't have any piping before the throttle body, straight coupler from air/water IC to the throttle body. I think I can tap the end tank of the IC though, it's thick as hell. I may do that, my IAT's are pretty counter intuitive at times and it would be nice to see if it improves the stability of the A/F's that I am seeing.

All in all the car runs great now, the startup hasn't failed me once now that the idle has been switched over and tuned a little better. The car runs very strong with rock solid a/f ratios at 17psi and I turn the boost up a bit more every time I drive ;)

Still working on a knock sensor solution...

mobythevan
August 10th, 2009, 08:51 AM
Sounds like you are making good progress. I run megasquirt on my 1G and it has worked very well for me for over two years. We also put it on my friend's 1G, but he has had a few problems. We are both running MSI_extra_hires code and both running E85. Our cars are very mild though. I am running the high compression (like 12:1) and running 18psi. He is running stock engine with 18 psi I think. We both run with just a fast idle valve for cold starts (on/off type). We have learned a lot of little things about tuning with megasquirt. I have run the computer on several of my cars going back 5 years and I have always liked it. I have helped other people install it and they have had lots of problems, so who knows.

JackM
August 11th, 2009, 08:45 PM
Hey, I have a couple of stupid questions. What is the largest size MAP sensor you can run with this? Also, when you hit the limiter is it like antilag or like the DSMlink launch limiter? How do you guys set up your limiter in this? Example: ignition is at 7k and fuel at 7.1k in my GEMS for the best NLTS at WOT.

My GEMS is starting to suck now with its 3 bar max MAP. I was hoping the MS could use a 5 bar...

Jack

Cloud
August 12th, 2009, 09:41 AM
I run a 4 bar map sensor. You can run whatever the hell you want as far as I know, you just need to know the sensor calibration data to input into the software so it knows what voltage is what.

I haven't spent a lot of time with the rev limiter stuff yet but it appears to have 'hard' and 'soft' limits that are similar to what you are talking about for the ignition cut and fuel cut points.

There are some things that MS has a ton of controllability for and others that it just is limited by hardware. Like sequential injection, you can run sequential injection but you just have to add more transistors and crap to it that it doesn't have provisions for in its current form. Overall I like it. It doesn't have the power or infinite configurability of a full race standalone like a haltech but it costs like $500 including bells and whistles. Good compromise IMO.

JackM
August 12th, 2009, 10:07 AM
Okay thanks! I was thinking of switching over to the EVO flash stuff on my car but all of the bugs for speed density haven’t been sorted out yet and I do not want to go back to a MAF. Looks like MS is the only other cheap solution for now.

Oh, one more question. How's the MS at detecting real knock? I know AEM sucks at it as many stand alones are pretty worthless when it comes to knock.

Jack

Cloud
August 12th, 2009, 12:34 PM
I have yet to find a solution for knock. If I do anything it's going to be with the stock knock sensor, though what I use for an amplifier/conditioner is wayyyyy up in the air at the moment still. What have you been using for a knock setup?

JackM
August 12th, 2009, 01:18 PM
Nothing. The GEMS is supposed to be really good at detecting it but it showed nothing when I blew the spark plug apart when my injector got plugged up from the E85.

Jack

mobythevan
August 12th, 2009, 02:03 PM
I have not been able to try any of the knock detect stuff with megasquirt yet. I have an amplifier/filter unit for my 5.3 LSx engine. It was supposedly set up for the bore and stroke of the 5.3 and should give megasquirt a useable signal for knock detect.

Cloud
August 12th, 2009, 02:31 PM
What did you use for the amplifier moby? A gm unit like in the megamanual or the knocksense thing that boris sells or what?

mobythevan
August 13th, 2009, 01:23 PM
It is the knocksense unit that Boris sells. I don't know if it will work or not.

Cloud
August 14th, 2009, 10:25 AM
I don't agree with using the same knock sensor for each application. That's why I don't really trust the knocksense unit, and I haven't heard a lot of good testimonials from it either. Even if boris is using a flat response knock sensor, a sensor with a tuned first natural frequency to the bore size is much more ideal. Our stock knock sensors are perfect for our cars. If the knock sense unit can be used with the stock knock sensor I wouldn't be too against the idea of trying one out.