Megasquirt EFI conversion

[dpstark2]

I got the throttle fixed and now have an extra 8% of throttle. So I guess I was getting about 92% WOT before. That 8% makes a difference! Cold starts are getting dialed in, as is the accelerator pump function. It is take WAY less fuel on tip in than before to handle the lean spikes.

Having done this in stages, I would generally recommend going straight to MPFI at this point. TBI was a good test bed for me, and CHEAP, but it caused way more headaches that I thought it would. I would be ahead if I had gone straight to MPFI at this point.

Consider this VE table from TBI:


And then this one now:


Sorry. I know the angles are different, but notice the MPFI table is much smoother and lacks the weird peaks. Something was clearly not right with the previous setup. I wish I had the tools and knowledge to know what it was.

 

[SNKEBIT]

Crap, I might have to get one for the Road Runner. Been looking at them a lot lately.

 

[Average Joe]

I should have done MPFI on my set up.
I thought I could save money doing the FAST Sportsman throttle body injection and not have to pull the manifold off to have it drilled out for injectors.
I ended changing the injectors in the throttle body to 1000cc so really the only money I saved if at all was the cost of the machining.
I believe if I would have done the Megasquirt software and machining on the manifold it probably would have cost the same.

 

[jamie]

Here is one of my cars I installed Megasquirt on a few years ago.I built the ECU and many other of the components.Was a great learning tool if you want to know everything about EFI.Here is the pulling truck I built and use a Megasquirt ECU.I use Tunerstudio....well you can see that I guess.

https://www.youtube.com/watch?v=AcDVKLL3RIM

https://www.youtube.com/watch?v=YVFQuOmErzg

 

[dpstark2]

Megasquirt is incredibly configurable and flexible. And I think that's part of the problem with it... The best tool you can have if you decide to give it a shot is a friend that already has a car using it. Knowing what I know now, I would suggest doing multi port straight out of the box. Batch fired is fine. Use an MSD box or HEI conversion to supply the tach signal to the computer. I was never able to get points or a magnetic pickup to trigger the computer correctly. Standard sensors and a wide band O2 and you should be in business.

Oh yeah- pick a single grounding spot on your engine block or head. It doesn't seem to matter how well connected grounds are, it takes one ground location to make things work.

 

[dpstark2]

Cold starts and idle advance

The Megasquirt has the following tables that can be adjusted for cold start:

1. Initial IAC position vs. temperature. This sets the IAC position during cranking based on coolant temperature. Generally more open for colder temperatures. I have mine set to fully open (250 steps) for anything below about 50F. It tapers down to about 50 steps for a warm start (160F).

2. Cold advance. Sets a global spark advance based on coolant temp. This was set to basically 0 degrees from about 50F up, and had up to 6 degrees advance for very cold temps (-40F).

3. After start enrichment. Adds a percentage of fuel only directly after start (going from "crank" to "run") for a period of time. The percentage and time are taken from coolant temperature. Might add 20% to fueling for 5-10s in most situations.

4. Warm-up enrichment. Adds a percentage of fuel based on coolant temperature. This generally starts high (up to 100% additional, which is a 200% multiplier) and tapers down to 100% (no enrichment) at 160F.

Numbers 1 and 4 are most like how a choke operates- cracks open the throttle, which then reduces as things warm up, and closes the choke which adds fuel.

The system also operates in a closed loop mode for idle control, meaning it has an idle RPM target based on coolant temp. I have this set currently to 1100 rpm when coldest (50F or lower) and 750 RPM hot. The computer adjusts the IAC open or closed to raise or lower RPM.

On a dead cold start, the computer has to fully open the IAC to achieve even a 1000 rpm idle (neutral). The IAC stays full open while the car warms up to about 120F. At this point, the target RPM and current RPM start to match (around 1000rpm). It then closes the IAC gradually until operating temperature is reached (currently 160F). At this point, the IAC is set to 180 steps, which is basically 60% open. Even at 180F, the IAC is still about half open (130 steps).

Now, after driving and fully warming the car, the IAC will nearly fully close (10-20 steps) to maintain a neutral idle of ~800 rpm. Once this achieved, if I use a test mode to set the IAC position manually, fully open will result in more than 1400 RPM, and fully closed just less than 800 RPM. Put in gear, the IAC will open to about 50-60 steps to maintain 750 RPM. As best I can tell, this means my throttle plate stop is in a good position. More could let it idle higher cold, but would result in a big drop from Neutral to Drive.

This gives me plenty of IAC adjustment to maintain idle under most conditions. What puzzles me is why the initial cold start can't achieve even a steady 1000 RPM idle. Hot, the IAC has more than enough "room" to adjust.

I did some thinking and decided to try tweaking the "cold advance" feature in the MS3. It was enabled by default, but didn't add any advance until temps were below like 30F. So I just thought, idle is set to 10-12 degrees, and the motor can tolerate 16-17. I set my advance to 1 degree at about 140F, 2 at 100F, 3 at 70F, 4 at 50F, etc. up to 6 below about 0F.

Firing off the car, it added about 4 degrees advance, taking idle from around 12 to 16 degrees. Idle was initially all the way up to about 1200 rpm, and came down to the 1100 target with the IAC at about 230 (vs. 250 full open). As it warmed up, it eventually settled down to about 150 steps with 0 advance at warm idle (750-ish in Neutral at 180F).

This combination gives me the range of IAC adjustment I need dead cold to hot. I'm still confused why this is necessary, as a carburetor has no cold advance and it works fine. I'm also confused why dead cold, the IAC can adjust idle from 800-1000 rpm, but hot it adjusts from about 800-1400 rpm. Carbs do behave similarly- the same fast idle cam notch will cause a much higher idle when hot than cold. So maybe carbs just have a lot more "opening" available through the fast idle cam and throttle plates.

An interesting aside: I had to retune the warm up enrichment for this as it warmed up. What was previously 13.5:1 at start changed to like 12.2:1. I had to reduce enrichment across the entire table. So apparently changing timing a few degrees at idle has a big effect on AFR? When cold? Maybe it just had a bad burn to begin with and the advance allows more time to burn the charge when cold.

 

[dpstark2]

The benefits of data

I was still having some issues with cold starts. In particular, at the ambient temperature around here of 45-55F, the car would start but cycle back and forth between the desired idle speed (1200 rpm) and nearly stalling (600 rpm). I got a few data logs of the cold starts and determined that the AFR was going lean as it nearly stalled. Clearly there was not enough fuel for the conditions. Why? There are so many variables it's hard to narrow things down. But with a few logs to look at, I noticed the problem wasn't related to the after start enrichment, it was temperature dependent, and the motor recovered when it entered a much richer part of the fuel table.

So, doing some quick math, it seemed I needed about 20% more fuel to be added during warm up in this temperature range. I added 20% to the cold start enrichment table around 50F. The next time I got out to start it and log, it fired right up and went to a steady 1200 rpm idle. It quickly came down to the target 1100 rpm, and warmed up correctly without stumbling, cycling, or any other issues.

My 20% addition was a bit too much. On the second round of trying this (you only really get one chance per day), I reduced the fuel a bit and got the AFR where it should be. There is still work to be done here- this was all done stationary in the garage or driveway. Driving it while warming up resulted in a much richer condition than I expected. Likely the table needs to come down another few percent.

Overall, I'm pretty happy with it now. I did some driving this weekend and it's pretty good. The fuel table is shaping up nicely. One other thing I've noticed is that the effects of the table change quite a bit between when the motor has just warmed up and is in fact truly warm. In the future, I will not do any tuning until the motor has reached it's true operating temperature- about 195F. The difference between that and just reaching 180F is a few percent in the table, which does make a difference, especially while warming up and operating in open-loop mode.

 

[dpstark2]

Full sequential

About a week ago, I made an cam lobe on the lathe and replaced the reluctor wheel in a magnetic distributor with that cam lobe. I used a points breaker plate to make switch that opens and closes roughly half of each distributor rotation. I phased the cam lobe so that the points were open about 1-2 cylinders before TDC #1 and closed 180 degrees opposite that. The point of all this was to convert my distributor (which is currently only used to send spark to each cylinder- the MS3 controls an MSD box which generates spark) to a cam sensor, allowing full sequential operation.

Well that didn't work. The MS3 fell back to semi-sequential operation immediately, but more importantly, the motor seemed to run somewhat poorly. And it wouldn't rev at all. So I broke out the tools I have...

The "composite logger" available with the TunerStudio software shows you the fuel and ignition triggers:



The top line is the cam sensor pulse. I was expecting a square wave (more or less), so this was confusing. It's possible in my configuration I should only see the pulse, but it should be even and should occur every two crank rotations. The second line shows the crank wheel teeth. There is a slight gap where there is a missing tooth every rotation that you can see. The cam pulse should occur just before this missing tooth, which it does... most of the time. So with that, I can at least guess that I have a noise issue on the sensor.

I put a scope on the sensor itself with the motor running:



And look at that! You can see the sparks transferring to the cam sensor. Well, that's the end of that approach. Without going to coil on plug, this simple switch version won't work. This is very similar to what I saw when I tried to use the factory electronic distributor to run the MS3 tach input- spark energy got everywhere and it just didn't work. The duty cycle is a bit off, but that's just a function of the points gap and could be fixed.

The only option I can think of is to convert to a Hall sensor like I used on the crank wheel and drastically reconfigure the distributor. However, I don't think that's worth the effort. Semi-sequential vs. full sequential is a minor improvement, so at a minimum I'll be going to individual coils first. Once that's done, the distributor can be converted without worrying about spark energy, so this approach may work.

The best news of the weekend is a nice clean cold start at 35F. I'm pretty happy about that. I let it idle down to about 1k rpm, then drove away. No sputter, no coughing, no hesitation.

 

[kb67mopar]

Its been a while since I read this thread. Sounds like you are becoming a tuning wizard. Lots of good information and lots of good experience!

 

[snakeoil24]

dpstark2 I am enjoying this thread a great deal. After reading what you have done I think i am ready to give this a shot on my son's car first. since He has a GM product and most modern day fuel injections systems are based on the work of Gm research, I think it will be a good chance to get read to do the hemi next year..

 

[dpstark2]

Good luck with the conversion. It's a tougher road than I expected, but I'm finally happy with the results. The only changes I'd make to the path I've taken would be to skip the trouble of modifying the fuel tank for an in-tank pump, and I'd go straight to port injection.

 

[dpstark2]

Interesting data point- I'm running a fairly mild 440. 224/230 duration, almost .5" lift, and 10:1 compression. On a carb, the best fuel economy I ever got on the freeway was around 10 mpg. I got 11 mpg once cruising at 55 mph. To be fair, I never really tried to tune the carb.

The TunerStudio software allows you to see fuel delivery in gallons per hour. I don't have a speed sensor hooked up, but I have a speedometer. At 65 mph cruising, I'm using about 3.5-4 gallons per hour. That's around 15-18 mpg. And it's not even fully leaned out yet. That's only around 15:1 AFR and it should tolerate 15.5:1.

This makes me rethink my plans for an overdrive. Aside from the noise, I'd be pretty happy if I could cruise my current setup at 15 mpg...

 

[dpstark2]

I finally got the time and weather necessary to do some highway cruising and tune parts of the map that I couldn't reach around town. You really need some steep up/down hill stuff to reach the high RPM low and high load parts. The map is very stable through about 4200 RPM now. I haven't run it beyond that much because I want to be safe about it and not destroy my motor.

Two points: first, it the system was pretty fat at highway cruise RPMs (3100-3400 RPM low load). I ended up removing as much as 10% from the map at low load. High load was pretty consistent already due to previous tuning. In the end, the motor was able to tolerate better than 15:1 AFR at cruise speeds, which results in somewhere around 18-20 mpg (on level ground, maintaining speed). Interestingly enough, efficiency is higher at 75 mph than 65 mph. You can see this in the map, but the fuel consumption gauge confirms it. For my car, this is the difference between being solidly "on the cam" and in the manifold efficiency range, and not. This can all likely be tuned slightly better with some tweaks to timing, too.

Side note: That makes me rethink my eventual plans for an overdrive. I have a non-lockup 518 I was going to use, but now I'm not so sure. Without the lock-up, I'm not sure I'd see any benefit. And it would cost around $2k to do the job, I'm not sure it's worth the money.

Second, cold starts are mostly ironed out. The "after start enrichment" isn't quite there, but it now consistently starts in 1-2 revolutions in all conditions. The only hitch is a slight stumble in some conditions when the after start enrichment phases out. The only way I know to tune that is trial and error, so I'm going to make a point of slowly increasing the after start enrichment time until the issue is gone.

At some point, I kind of felt EFI wasn't worth the trouble. I've changed my mind. The caveat is that the learning curve is steep and patience is very necessary.