Using 4 ECU's Simultaneously On 1 Engine

[ACCLUDE91]

Before flaming me and saying I'm an idiot, please read this whole thing and keep in mind that I am not an idiot and am very capable of making this setup work! I know this will not yield major gains, but that's not my purpose at this point!



I want to try something which I don't think has been done on our cars before. I have an idea which would require each cylinder to be controlled individually. I know some of you are thinking it's pointless or that it makes no sense, but you just gotta trust me that it does. At first, I want to try it on an unmodified motor with unmodified ECU's. After the kinks are worked out, I have much bigger and better plans for it (which may require 8 ECU's to be used) but I need to start somewhere

I want some valid input or even some help with this setup. I know cp[mike] is up for the challenge because I already talked to him in detail about this so maybe he can help explain somethings that I might leave out without giving away the bigger picture.


Anyways, here's the general idea:


OBJECTIVE:
To get accurate air/fuel ratio on all 4 cylinders independently using one ECU per cylinder.


PHYSICAL CHANGES & REQUIREMENTS:
4 O2 sensors (one per cylinder)
4 OBD1 ECU's with 4 sets of ECU plugs
lots of wiring


SOFTWARE CHANGES & REQUIREMENTS:
none so far, but plenty to come later


GENERAL DESCRIPTION:
The ECU's would all be wired up together to get the SAME required sensor input like tach, VSS, TPS, MAP, IAT so that they are all on the same page with that. However, the ECU's will independently provide the ground for the fuel injectors, and will get O2 input independently. For example, ECU #1 will ONLY control the fuel injector for cyl#1, ECU #2 will ONLY control f.i.#2 and so on. Only one ECU will be incharge of the ignition timing to keep it simple. Also, only one ECU will be connected to the car harness while the other 3 will only have the necessary connections for them opterate correctly. Each ECU will have it's power and ground wires seperate for simplicity and safety,


ECU TRICKING:
In order for it to work and run at least as good as factory or better without kicking on the CEL or running in limp mode, each ECU needs to be tricked into thinking that it's running the engine by itself. This is going to require some pretty extensive use of resistors and diodes and possibly some potentiometers to be able to tweak the resistance of a given circuit and to keep circuits isolated. For example, ECU#1 would only be controlled fuel injector#1 (pin A1), but it needs to see a valid signal for injectors 2,3 & 4 so pin A2, A3 & A5 would each connect to a 13ohm resistor with a +12V source so it "thinks" it's got the injectors hooked up to it. Same thing with the other 3 ECU's. Still working out the details for tricking the 3 ECU's into thinking they're controlling the ignition timing, but it would also need to be done to pins A21 & A22 for each of them.


TOTAL COST:
4 O2's @ $60ea - $240
4 PT3 ECU's @ $10-20ea - $80
3 ECU Harnesses from junkyard - $25
Various semiconductors - $50
GRAND TOTAL - $395

or if someone wants to donate any of these items, it would be appreciated!!!


I have not started on this project yet and it may be a little while until I do. This is stage1 of a project that's going to have to be built in about 8 or 9 different stages. Eventually, it will be used on a one-of-a-kind engine that I'm not disclosing details about at this time.


Please give me your valid input and offer any kind of advise or help that you can. Maybe you can help prevent some issues that are not expected. I know there are cars with multiple ECU's (usually running multiple fuels) but I have not seen a setup like this yet and especially not a factory motor with factory ECU's. Looking for your input. Diagrams will be coming after more details are worked out.

[hemlockz]

wow... um, WOW. props, but what you been smokin to get all this inspiration? I would guess that the biggest design flaw is... I suspect you will loose more power than your setup would gain by having an O2 sensor in each primary of the header. It would slightly interfere with the flow.

you might be interested, the new crome has fuel trim per cylinder, but this is to control differences in injectors. Actually if you are shooting for perfect AFRs in each cylinder, the best way to do it would be to get your injectors balanced. After that maybe use ITBs. Then you'll have a pretty closely matched fuel and air for each cylinder. I say ITBs because each intake runner has different flow characteristics and would probably mess with your air ratio even if you got the fuel nailed.

[phatdoughnut]

If you are serious about it and i dont know what hit you over the head to get this idea(because you are a pretty damn smart guy.) You will need wideband o2 sensors, 4 individual o2 sensors will not allow you to tune better than 1 wide band tunning the whole engine.

plus you can get a pretty damn good tune with a wideband if you have someone who knows what they are doing.

[cp[mike]]

If you are serious about it and i dont know what hit you over the head for this(because you are a pretty damn smart guy)

its a long story, this is just part 1 of something larger...



pat, IMO youre going to need to compile a list of all the sensors an ECU reads, cross out which ones can be eliminated by chipping (a p06 for example, no KS, IAB, VTEC, etc), then the remaining list needs to be broken down into 4 categories...
1. unnecessary outputs (IAB)
2. necessary outputs (injector)
3. necessary inputs (ones youll actually use)
3a. static inputs (ones where a single resistor/diode can spoof it)
3b. dynamic inputs (sensors that vary their reading as conditions change)
4. unnecessary inputs (can be spoofed or left disconnected)

then youll need to find specs and information about the sensors youll actually be using (2 + 3), and decide the best way to spoof or use each one. once each sensor is taken care of, the ECUs should basically "think" they are working alone, with their own set of sensors... then fuel tables can be adjusted independently.

youre going to have 4 CEL's, you realize that, right?


i have a feeling the ECUs may not like sharing sensors, like the TPS... which one is going to be allowed to provide power? or are you going to tie all of them together with diodes before getting to the TPS, then share the output between all 4? youd have to tie the grounds together, which may freak the ECU's out... youre going to have some weird bootleg workarounds, imagine 4 TPS's all tied and moving together with a single center shaft.

[Bird333]

I would try to just split off all inputs to all 4 ecu's and only have necessary outputs coming from each ecu. You may want to look into creating your own custom jumper harness for this. Good luck!!

[hemlockz]

Also you know you can't program PT3s so your cost will be more like $100x4 for chipped P28 or P72 family ECUs...

LOL cp[mike] 4 CELs thats like the the deepest layer of hell I could imagine!

[Accord R33]

I think the biggest question is not how..but WHY?? what is this going to do for you that you cant do with one ECU?

If your going to go that far..you would really need individual intake manifolds (not just ITBS). with individual MAP sensors for it to really be effective.

this is definitly not worth it in my opinion

[h22sparkle]

Ontop of that that would prolly cause the car to have a personality disorder to not run right at all. Not to mention shit would be harder to diagnose wowza the horror.

[starchland]

makes alot of sense, i remember researching engine controller/o2 shit somewhere and one had memory for 8 sensors. I mean im sure they do this on f1 engines and shit, cause that way they can tune each cylinder for max potency, and also each cylinder wall/head/hone/etc will be a little different.

[cp[mike]]

If your going to go that far..you would really need individual intake manifolds (not just ITBS). with individual MAP sensors for it to really be effective.

this is definitly not worth it in my opinion

ultimate goal is 4 MAF's, but that will come later. concept and prototype are the goal at this point.

[immeraufdemhund]

i think that chipping the ecu and maybe looking into just makeing your own ecu would be better than having 4 complete ecu's.

not only that if you have even a 4mhz processor figuring all this stuff out that means it can handel ~4million operations a second. granted if you split it up into individual sensors the number will be significantly less. but even if it was say 100,000 times a second that is still pleanty fast to keep up with your motor using only 1 ecu. even at 8k rpm's that's only 133 rps. the ecu split up gives you 25,000 rps max...and that's if it's at 100,000. Imagine if it was faster.

heheh 25,000 rps is 1.5million rpms...that's hella fast.

[ACCLUDE91]

Thanks for the input so far guys!

Yes, I know I need widebands to tune, but for now I just want to get the damn thing to run!

Chipped ECU's would be the second step and I can mod them myself so cost isn't too bad. It may be needed to do what I want for stage1 or stage 2 of this project, but chipped ECU's are not going to do me any good in the later stages.

Eventually, there will be one computer running the full thing, but I need to do it this way for now. I know it sounds kinda crazy or pointless, but you just gotta trust me that in the end, it makes sense.

The ultimate goal is to design and build a computer from scratch to run the unique engine I want to build. The problem is that I have a one of a kind engine design that has not been made by anyone yet and I'm trying to work on getting an ECU setup to run it FOR DIRT CHEAP because there is no ECU in existance that can do it yet. I'm not a master engineer or anything so I gotta do it the ghetto way for now. I won't be able to use Crome or Uber or any of that good stuff on the custom engine because it is nothing like any Honda engine AT ALL!!! Shit, It's not like ANY engine at all! For now, I'm just trying a little test to see what kind of results I could get using 4 seperate ECU's because my custom engine will require at least that much for now! If I was rich, sure, I'd just hire a team of engineers and have them make me a comuter for my needs, but I'm not so I need to do it myself and for as cheap as possible. This is the cheapest, but still effective method I can think of. You'd understand if you knew how radical the engine is that I want to run. Lets just say that it has 1/4 of the internal moving parts that a traditional 4-stroke motor, but has 3-4 times the amount of sensors and devices to control. For now, I just want to try it on a factory, common motor like the F22's to get a good idea of what needs to be done for the next step.

[wed3k]

all in know, is that im willing to donate my pt3 (thats the a4 ecu right?)

[That1guyN8]

I would be willing to donate mine also once I swap it out.

[immeraufdemhund]

all in know, is that im willing to donate my pt3 (thats the a4 ecu right?)

and a1. i have an extra pt3 around as well. And a wire harness that you could bum off me for shipping....sho far i have an entire extra cb7 with no engine tranny if you want it.


as for the engine it sounds kinda similar to what i wanna build my self. It's a make shift version of the bourke motor design is what i want. This way it's more simple like a 2 stroke, and less crap to go wrong. I dont mind have sensors out the butt, yet it seems to me the more crap on a car usually means it runs fine untill....the dreaded tune up date. Then it's an arm and a leg to fix it. If you dont mind you could pm me the design of the motor. I wouldn't mind gandering at it. I have like 3 extra honda motors laying around and 1 vw bug motor....anywho this should sorta be in a pm, so i'm gonna wrap it up.