Installing the FAST EZ-EFI 2.0 Multi-Port Retro Kit in a 1987 Pontiac Fiero V6 (Page 3/8)
lou_dias FEB 23, 12:45 PM
Without dyno # feedback, you can say you want 13.5 A/F ratio and have it auto tune but that doesn't mean your making the most power since you have to adjust your timing as well. When you adjust your timing, your a/f ratio adjusts as well. You don't know which direction to go in until you have dyno #'s.
It's a feedback loop that doesn't yet exist because no ECM is aware of actual power output.
If you just want to cruise, it's overkill, but just fine.

[This message has been edited by lou_dias (edited 02-23-2023).]

82-T/A [At Work] FEB 23, 02:12 PM

quote
Originally posted by lou_dias:

Without dyno # feedback, you can say you want 13.5 A/F ratio and have it auto tune but that doesn't mean your making the most power since you have to adjust your timing as well. When you adjust your timing, your a/f ratio adjusts as well. You don't know which direction to go in until you have dyno #'s.
It's a feedback loop that doesn't yet exist because no ECM is aware of actual power output.
If you just want to cruise, it's overkill, but just fine.





You know your stuff Lou, I've been talking with their engineers and they've specifically stated that the idea has merit, but would require an ability to identify "changes" in performance (why a dyno has value). We would need to measure that in some way, which could be done realistically with something that measures improvement or increase in G-forces and momentum to augment this. But it is true what you're saying, there is a ~2-4% variance on getting more power out of it above and beyond optimizing fuel ratios.

Goal for me with this was to have a solid-running / driveable car. But now I'm interested in seeing where I can make this go with their engineering guys. They're interested, so the conversation has started.
claude dalpe FEB 23, 08:16 PM
Hi, 82-T/A [At Work]

I was one of the first to successfully complete a 3.5 LX9 swap with a bug free 7730 ECM about 13 years ago, But it took me at least 1 1/2 years to successfully (I was at the 30th of the fiero in Indianapolis with this engine)

I worked with the things Moates was selling at the time (emulator and chip burner) You burn a chip and you will try it on the road (very long process).

Anyway Moates is closed now

All those at the time that I contacted for this swap to have a tune already done for my 3.5 LX9 they had no final one and all wanted me to send them scan data to correct the tune so i decided to do it myself

In short, all this to tell you that it is very long and requires mechanical knowledge that I have (mechanic) to do engine tuning
Of course today there is the Flash memory on board with a 7730 it's more fast

So I really advise you to do with the Fast Ez-efi kit that will give you a correct setting for the use you want to make with your fiero

Don't give up and keep us posted
Raydar FEB 23, 08:41 PM

quote
Originally posted by 82-T/A [At Work]:
...
My 3.1 is every so slightly hotter than what a stock 3.1 would be. I have .040 overbore pistons, and also decked the heads and intake so the compression is a little higher as well. I believe I'm getting around 9.2:1 compression (if I remember correctly). I don't suppose that makes too much of a difference, but I'm using 17# Accel (pintle-style) injectors. So I think the 17# are probably decent for what I'm doing, especially if I get rid of EGR and cold start injector.

I'm going to have to ask you about the springs when the time comes for that. Also, a problem I have is that I can't / shouldn't even start that engine with the EZ-EFI since if I remember correctly, to properly break in the cam I'll need to bring it to 2,000 rpms for a couple of minutes (it's been a long time since I looked at how all of that works..



I'm thinking those injectors will be fine. Wish I would have had those, as my 19s seemed to be just a bit large.

Your comment about springs just reminded me...
Be careful about lifters. Most people I've heard say to use the GM lifters (assuming they are still available). Several people on this forum had other "high performance" lifters to fail, and wipe out cam lobes in the process. And yeah... the GM lifters are more expensive.

Regarding the startup / break in. When I first started my 3.4, my headers glowed quite brightly. I tweaked the "tang" on the TPS to tell the engine that the throttle was open a bit more than it was. They still glowed, but not like they did at first.
My take on it is that if it's not overheating, and not turning the headers a bright red or white, just let it rip. It's close enough for the break-in to not damage anything. (Fiero headers glow a bit under normal conditions, anyway.)
Others may disagree. That's fine. I'm listening. Just my personal opinion.

9.2 should be fine. The 272 likes high compression. When I first got my 3.4, it pinged pretty badly with the stock cam, even on mid-grade. After I installed the 272, it ran fine with 12 degrees advance, on regular gas.
pmbrunelle FEB 23, 08:43 PM

quote
Originally posted by 82-T/A [At Work]:
Are you talking about the EZ-EFI 2.0? Because that is not at all what it does. You set the displacement, number of cyls, the type of timing controller (internal, or external), and that's literally all that's required. It has some basic fuel maps which are determined by cyls/displacement and then it self-learns and builds it's own fuel tables off of that. You can adjust the additional settings if you want, but it's unnecessary unless you want something specific.

I'm not sure why you guys are so adamant about me buying a Megasquirt. I'm happy to buy one and throw it on the shelf if that'll pacify the "I could have saved $500 bucks" argument? The money isn't important to me, reliability and simplicity is. My time is extremely valuable. I spent the first half of my adult life as a programmer, and don't want to have to sit there with a laptop fiddling with the settings of some old-ass GM controller... going back and forth, up and down the street, trying to dial in my fuel maps.

A truly intuitive system should not require that ... that's old technology. The EZ-EFI system continuously learns, essentially using machine learning to continually build those fuel tables, not a decision-tree algorithm. That's what I want.



Yeah, I was talking about the EZ-EFI 2.0.

It reads user-defined lookup tables, which are then mostly likely read by some human-written procedural program which then toggles the outputs as needed.

Here are some excerpts from the instruction manual:

Ignition curve, user-defined:


AFR table, user-defined:


Except for the VE table, nothing else in this ECU is self-tuning.

********************************************************************************

I do not think that you should favour one ECU over another.

I am adamant however that the EZ-EFI 2.0 is just like the other ECUs in concept; a procedural program following user-defined lookup tables.

As such, for the engine to run well, someone will need to tune the car doing trial/error trying different settings and seeing what works best. Just like the other ECUs.

The idea of install, push a few buttons, and drive, is IMO a well-engineered marketing dream. One can fall for the marketing, or not...

********************************************************************************

As mentioned, you can't teach a machine to learn anything if it's blind and without feedback. Perhaps eventually automotive ECUs can be self-tuning with AI, but not with the sensor complement of your powerplant.
sleek fiero FEB 23, 10:59 PM
Hi 82 T/A [at work] . Just a word about your cam. Yes that is a great grind. I have worked in the industry at an automotive machine shop reconditioning engines for years and here is my advice for your cam break in. Get yourself a tube of engine assembly lube or better yet a tube of Crane cam lube which is a molybdenum grease and use an ample amount on the cam and on the bottom of the new lifters. Also get a pan full of lucas High Zinc 10-30 for older cars and submerge your lifters pushrod end up. Take a pushrod and insert in end of lifter and push down multiple times until the lifter fills and you can't push it down anymore. This will prefill the lifters so they will not rattle on startup. Like wise use the lucas high zinc oil for engine breakin and keep using it for the life of your engine. Don't worry to much about your startup rpm as tou probably wont' have much control over idle speed at first but with the proper lube on your cam it should not gall on startup. Set your lifters to 0 clearance where with your fingers you can turn the pushrod freely . Then tighten 1/4 turn and your lifter will be set for street use. Some racers will not give the final 1/4 turn for a track car. You should check with Crane whether you can use your stock springs or get a matching performance set. Hope this info helps you and good luck.
MarkS FEB 24, 12:24 PM

quote
Originally posted by Raydar:

Your comment about springs just reminded me...
Be careful about lifters. Most people I've heard say to use the GM lifters (assuming they are still available). Several people on this forum had other "high performance" lifters to fail, and wipe out cam lobes in the process. And yeah... the GM lifters are more expensive.




Agree, GM- the modern hydraulic flat tappet lifters are a constant source of controversy in the hot rod world, installation notes often say remove the inner valve springs before break in, figure this is their escape hatch if you don't & they fail. FWIW only, on my last cam swap in the 400, which was only a year or so ago, I used a set of NOS TRW lifters date coded to the early 80's, quality stuff back then. Worked out great WITHOUT removing the inner springs from the Edelbrock heads and the TRW instructions had nothing about removing valve springs. I believe VL94's were the TRW lifters from that time frame. "HyLift Johnson" are a decent modern lifter from my experience & others.

[This message has been edited by MarkS (edited 02-24-2023).]

La fiera FEB 24, 08:33 PM

quote
Originally posted by 82-T/A [At Work]:


Hahah... perhaps. Incidentally, that's what I do for a living. I literally run a team of AI researchers. As the chief scientist of the company I work for says, "AI cannot reason," which is what I think you mean. But I think it's important to understand what AI can and cannot do well. Number crunching, and normalization is something that AI actually does really, really well. To be honest though, the EZ-EFI 2.0 isn't new. It's been around for almost a decade if I'm not mistaken. I think it came out in 2013, and there have been a couple of updates. It uses rudimentary machine learning, so not exactly the sophistication of a GAN or an NLP... but to be fair, the type of processing that's required in a fuel injection system is extremely simple.

If we're using AI terms... both systems use some form of machine learning logic. The difference is whether or not the learning model is done through supervised (what you're doing) or unsupervised (what I'm doing) learning. The inputs are the same... TPS, MAP, IAC, O2, etc. You go through a lot of effort to "dial in" these specifics (supervised learning)... but you're also merely using the input from these sensors to make a decision. This is something that machine learning can easily do on it's own.

Truth be told, I'd be willing to bet that all of these systems that are out there are so old that they're using very basic math algorithms. If someone put in even the slightest bit of effort into redesigning one of these... that "issue" would be solved tomorrow. But there aren't a lot of car guys who are also mathematicians, who are also computer programmers, who are also AI engineers. Never the less, it's not a hard problem... and there's no reason to go through the effort of repeatedly dynoing and "dialing in" a car's computer, when it simply should be able to optimize the efficiency on its own through basic math.

My kit comes in the mail today... so I'll post pictures when I get it. But I still need to tow my Fiero from the storage unit first.


BIG question I have though... if I'm going to be installing a new cam in my engine... I want to run it with the existing computer that's in there, right? I need to be able to run the car at 2,000 RPMs from the get-go (or whatever the break-in procedure is for a flat-tappet cam). It's been like a decade since I've really turned a wrench, so I've forgotten almost everything. If I put the FAST system on immediately, I'm concerned I may wipe out my cam.


EDIT: You encouraged me to reach out to Edelbrock and Holley to see if I can get some of the guys on my team to work with either of them on a new, more advanced system. I just asked to be able to buy the unit at cost when it comes out. I'll let you know what comes of it.




Well, you gave it away that you are an AI programmer in one of your comments in one of your earlier post that's why I mentioned AI. Even though you didn't say it straight on, psychology 101 teaches you to read between the lines and I took psychology right after high school because I was bored. I'm surprised you asked about flat tapped break-in when you have Artificial Intelligence that can do everything for you. Now you can clearly understand what I mean by AI missing the human touch. But I'm always about progressing and learning. Critics make you better on whatever you want to achieve. If it wasn't for critics I would have never reached the goals I've reached with the Iron heads. So don't take this in a negative way, take it as a way to improve your skills, just like I did.
What I'm trying to say is that whatever you do, give it 150% and learn along the way, critics are there to give you insight, not to hinder you. i wish you the best on your quest!
But I have a challenge for your AI and it's programmer.
ericjon262 FEB 25, 11:47 AM
I'm very skeptical of "self learning" EFI systems, and "plug and play" systems that make big claims.

I haven't dug into this system in particular, but I would highly recommend checking with dyno tuners in the area and seeing if any of them are willing to mess with it. I wouldn't be surprised at all if the self learning functions left a ton of performance and drivability on the table.

The other thing you should do right now, before you take the car apart, is attempt to set a benchmark for the performance of the car as it sit right now. be it a dyno, dragstrip pass, 0-60 time, 20-80 time, whatever, do something to give a origin datapoint prior to the changes, that way after, you have a basis for comparison when you install the new system, and can verify you're at least meeting the original performance levels you had before you started. I would also try and record as many of the atmospheric conditions at the time you set these benchmarks, and try and do your after tests with as many of the variables similar that you can.

I've warmed up quite a bit to megasquirt over the past few years, it's much more refined than I had realized in the past, but it also still doesn't offer OEM levels of detail, knock detection/preventions is a decent example of this. On the other hand, with MS, the program running the car, is known by the tuning software, and datalogging can provide massive amounts of insight as to why things are happening,

here's a screenshot of the fuel calculations summary from my MS3, in my case, I don't have a ton of the available features turned on, so alot of the multipliers are just 1, but if I were to have a lean spike, I could use this to see if the lean spike was a commanded trim to the fueling, and which parameter drove this command, or if it wasn't, and I have some kind of other mechanical problem causing that.




I do think a dragstrip could be extremely valuable for making adjustments to the tune of a vehicle, as you remove many variables, that said, the majority of the drag strip tuning will be for WOT, unless you want to baby it down a few passes. it also offers limited transient tuning, this is where datalogs of street driving come in handy,

all that said, to me, it sounds like a ton of work to swap an EFI system, and not want to try and optimize it by tuning it, or have it optimized by someone. I see cars all the time at car shows that guys dumped a ton of money into, and have piles of great parts on it, but they run like junk and fowl plugs because they haven't taken the time to jet the carb, and adjust the ignition timing, or are making adjustments without any metrics for comparison, and then thinking that because it's running it's good. it's really sad to me, because in this day and age, anyone can download a GPS 0-60 app for their phone, and very easily see how their car's performance has changes mod for mod, for nothing more than the cost of the fuel to run the engine, and a little time.

------------------
"I am not what you so glibly call to be a civilized man. I have broken with society for reasons which I alone am able to appreciate. I am therefore not subject to it's stupid laws, and I ask you to never allude to them in my presence again."

I invited Lou Dias to trash me in my own thread, he refused. sorry. if he trashes your thread going after me. I tried.

maryjane FEB 25, 01:36 PM

quote
Goal for me with this was to have a solid-running / driveable car. But now I'm interested in seeing where I can make this go with their engineering guys. They're interested, so the conversation has started.



Mission creep....