Design an ECU project? | Page 3 | FerrariChat

Design an ECU project?

Discussion in 'Technical Q&A' started by mk e, Jun 24, 2009.

This site may earn a commission from merchant affiliate links, including eBay, Amazon, Skimlinks, and others.

  1. mk e

    mk e F1 World Champ

    Oct 31, 2003
    12,899
    The twilight zone
    Full Name:
    The Butcher
  2. mk e

    mk e F1 World Champ

    Oct 31, 2003
    12,899
    The twilight zone
    Full Name:
    The Butcher
    Ha! I got 1! My neighbor is a software engineer ans signed up :)


    The engine will have
    12 54mm TBs
    12 60lb high impedance shower injectors
    12 18lb (probably) high impedance port injectors
    36-1 crank trigger with a hall effect sensor
    Cam position sensor, hall effect
    MAP sensor…..I’m going to try to make this read well
    TPS
    Air temp sensor
    Coolant temp sensor


    The 18 lb injectors will run to about 30% throttle then will step down to say a constant 50% duty cycle and the 60 lb injectors will do the rest. I'll work on some what happens when layouts...I've got company for the weekend noew though.

    Mark
     
  3. fatbillybob

    fatbillybob Two Time F1 World Champ
    Consultant Owner

    Aug 10, 2002
    26,294
    socal
    Why not save yourself time and do a mega squirt on the low end or motec on the high end?
     
  4. mk e

    mk e F1 World Champ

    Oct 31, 2003
    12,899
    The twilight zone
    Full Name:
    The Butcher
    #54 mk e, Jun 25, 2009
    Last edited: Jun 25, 2009
    I would need 24 megasquirts and still not have propper control of the injector due to inadiaquate maps or 2 motecs with 2 ignition expanders for around $12k and still not have traction control because the 2 ecus don't talk to each other making any attemp unstable. Bad solutions imo.


    You guys really need to read more than just the top post on the thread.....this is about the 6th or 12th time we've covered this topic in a day and a half :(
     
  5. ricrain

    ricrain Karting

    Nov 1, 2003
    213
    Dallas Area
    Full Name:
    Ric
    I'm jumping in a little late, and I haven't had time to read the other posts. I have designed an engine controller before under contract. I can't say who or what other than to say it was for a V-8 ICE running street gas.

    My initial comments are: 24 fuel outputs seems excessive. I run my TR on a Motec M48 with only 6 fuel outputs. Any ECU using modern Power MOSFETs for the fuel driver will easily drive 2, 3 and maybe even 4 high impedance type injectors per driver. The REAL problem is not the driver's ability to sink current (a modern TO-220 Power MOSFET can drive 80-100 AMPS... EACH), but the fact that there are not enough ground pins on just about every after market ECU I've ever looked at (For example, the M48 has ONE system ground pin).

    I understand the need for staging injectors, but even with 24 injectors, you should be fine with 6 (or 12) injector drivers. True Sequential injection for all 24 injectors will net a small fuel savings and a gain a tiny bit of power, but IMHO is not worth the god-awful complexity.

    Also, you could run variable fuel pressure to compensate for fuel injector size and run only one set (12 injectors). You can get about 180% of rated flow from a Bosch injector (40 PSI ramping to about 80 PSI).

    Six ignition drivers can be used in a waste-spark configuration. I would only use a unit that uses IGBT ignition drivers (for durability). IIRC, the diyefi/megasquirt system uses these. The coil over plug devices being used on motorcycles are way cool (it's what I have on my 308 motor).

    MAF systems are not as restrictive as you might think, as long as the bodies are sized correctly. MAP/TPS (mass air flow) works, but the tuning tends to drift for reasons I've never fully understood (I think the VE of the engine drops as it wears the rings).

    I only use wide range O2 sensors for maximal tuning and power plus peace of mind ("it's running lean... HOW LEAN?"). The VW units with the integrated calibration resistors are relatively cheap and quite effective. One O2 sensor on each bank would be sufficient. I think the diyefi/megasquirt only supports one, but I could be wrong.

    I might be interested in cranking out another system. The CPU I'm most familiar with in the embedded world is the SAB C166 series (80C167, etc.). It's also the CPU that I have all the high-dollar tools for. I have a working prototype of a CPU core now that has a 64MHz CPU, 512K of RAM, 256K flash (more than enough) and supports RS-232, USB, RTC, CAN and compact flash memory (16GB of datalogging... probably exceeds the life of the engine at 8 channels of 20 samples/sec... 27,000 hours... LOL). It's about the size of a credit card and rated for Automotive temp range operation (theoretically could be installed in the engine bay, but I wouldn't...)

    I have a paper design for the driver section that has 8 ignition drivers, 8 injectors drivers (with current sensing), MAP, CTS, IAT and two inductive position sensors, along with 10 uncommitted A/D channels and four Push-Pull drivers.

    FYI. As mighty as the reputation is for Motec, the M48 and M8 (and even 400 and 800) have some design problems. Not to mention their GUI tuning software feels ancient.

    One poster suggested that he might one day replace the CPU in a 360. Well, good luck with that one, especially on an F1. There are several controllers in a 360 that all have to talk to each other (via CAN) in order for the car to even start. 348, yea. 355 6-speed, yea. 360... not so much. :)
     
  6. mk e

    mk e F1 World Champ

    Oct 31, 2003
    12,899
    The twilight zone
    Full Name:
    The Butcher
    Just the person we need on this project :)


    This is the point where all the current ecus take the wrong path. When you ask the mechanical guys sequential injector operation is not really optional, it's required. Yes the engine will run without it but it won't run as well as it should nor will it make the hp it should. I don't think there have been any OEM systems there are not sequential in 10-15 years becasue it's required to make the engine run propperly. I've run a couple engines both way and talked to several other who have and the answer is always the same.....sequential injection mkaes a very noticable difference.


    I've been thinking about this as a fall back plan. It's not as good as 2 injctors becasue it can't solve the injector placement problem, but it would help an injector small enough to idle right supply enough fuel on top. On a boosted engine doing this is as simple as a ratio regulator.....but on an NA engine????? A pressure sensor and control it with the ECU is all I came up with?? The other issue is that spray pattern of the injector goes to hell when you deviate too far from the designed pressure, you get the flow but you lose control of the spray pattern and mixing (and hp) suffer. I think it's a good fall back position.


    Yes, 6 ingnition channels is enough. I like the motorcycle coils and would love to have the ecu drive them directly, but honestly I would install distributors if it got me the fuel control I want. The fuel control in much more important.


    Anything with the word restrictive in it is probably bad.


    i think you're right

    I had another thought. Since a driver can drive 2 injectors, would it be possible to wire the high and low injectors to one driver, but put the high injectors thought a solid state relay board and use one to the programable outputs to switch them on at the desired low point? That would let a haltech of motec run the injectors the way I want them.
     
  7. ricrain

    ricrain Karting

    Nov 1, 2003
    213
    Dallas Area
    Full Name:
    Ric
    I think production cars run it for EPA more than anything else. I've ridden in not one, but two 800+ HP vehicles that used batch ignition and batch injection. One was a TR (John Carmack's) that made something like 1013 HP (I clocked 0-60 with my casio watch at 3.2... for a 4000 lb car, that's impressive). If I'm not mistaken, the TR was run totally off an M48.

    I'm not doubting your experience or conclusions, but it would interesting to see actual HP/Torque gains for pure-sequential vs. what I'm doing on my TR (2 squirts per rev).


    Well, the Veyron, all current Ferraris and all current Lamobs runs MAF... and there's some impressive HP/Litre numbers in there.

    There's an idea... Yes. Using an AUX output, you could "enable" a small external high current MOSFET that would effectively "enable" the larger injectors. The transition would be problematic though, as you'd be running, let's say, 70% duty on the small injectors at the point of "cut-over", then your large injectors would be momentarily running at 70% if the fuel map and the AUX transition didn't occur at exactly the same spot. I'd have to look through the M48 settings to see if the transition point could precisely be matched to and AUX output transition. If so, the temporary richness would only be for one or two revs, I'd guess.
     
  8. mk e

    mk e F1 World Champ

    Oct 31, 2003
    12,899
    The twilight zone
    Full Name:
    The Butcher
    #58 mk e, Jun 27, 2009
    Last edited: Jun 27, 2009
    No question its' driven by the EPA.......but the goal to control mixture and that is what works. Ideal and off idle running are always noticably smoother with full sequential operation and for a street car that's important, at least to me. The dyno runners I've seen where on NA engines and looking at injector duty cycle and showed hp dropping when duty cycle expeeded 60-65% on a sequential engine and I don't seehow you could run duty cycles that short and have it run at idle without sequential.

    The realy interesting thing is the standard numberrs on mixture go out the window when the mixture is well controled and mixing working....the peak hp was at 14.5:1.

    The other things is I've going to have a lot of cam and massive ports....I'm reall concerned about the low end performance and want to do everything possible to help it.


    No doubt, but still 10-20% under what I'd like to make per liter.....I don't have those pecky EPA rules to worry about ;)


    I figured set fuel map load points at say 50 and 50.2 then set the injectors on at 50.1 or something like that. it wouldn't be perfect but most likely unnoticable. Then I go with 1 motec m800 and have everthing I was looking for.....maybe
     
  9. Mowgli

    Mowgli Formula Junior

    Feb 28, 2009
    435
    Bristol, CT
    Have you looked into the HKS F-con Pro? Top Secret used two of them to run their V12 swapped Toyota Supra. It has 32x32 maps but can only run up to eight injectors. So maybe that won't be enough?
     
  10. mk e

    mk e F1 World Champ

    Oct 31, 2003
    12,899
    The twilight zone
    Full Name:
    The Butcher
    I haven't looked at that one. There are lot of 8 injector driver units out there.....which are at least 4 too few :(


    I'm really liking the high injectors on relays idea.......
     
  11. pad

    pad Formula 3

    Sep 30, 2004
    1,418
    Tequesta, FL
    Full Name:
    Paul Delatush
    Mark,

    If all else fails, here's a fall back possibility. I was was talking to the Electromotive techs the other day on another matter and mentioned your requirments (actually, I emailed them you list of must haves and options). Without going into details, they said (2) of their Tec3R will give you everything on your list except traction control. For that, there are several products on the market that will interface with the Tec3Rs.

    Now back to our regularly scheduled thread........
     
  12. CraigFL

    CraigFL Formula Junior

    Jan 17, 2001
    954
    Panama City, FL
    Full Name:
    Craig
    I jumped in late too but here's what I know from my experience. Hardware is the key. You have to have "hardened' hardware that will last in an automotive environment. If you have to develop it yourself, it will take a lot of experience and testing and... you'll never be sure it is reliable until you sell 500 units and then it will be a nightmare if you have to replace it under warranty. Only testing and actual usage will prove it out!. If you choose to use to use someone elses hardware, you will be stuck with what is not the latest technology(not the fastest) or not as flexible and it will be provided in their size and configuration. My guess is 1K man hours for using existing hardware-- interfacing while it would be upwards of 6K hours for designing your own-- not including testing.

    Whatever you choose, you will need to document it, simulate it and test it thouroughly. Because of the things it could do, you need to be sure it's safe --- no WOT by accident if it gets stuck in the wrong software loop. The software alone is a large somewhat complicated development project that will take the knowledge of software people as well as the engine/mechanical people - both theorists as well as the mechanics. Without the hardware, my guess is 7K to 10K man hours.

    Of course you can bypass a lot of this if you just want a working model on a test car that would never be sold to anyone...
     
  13. samsaprunoff

    samsaprunoff F1 Rookie
    Silver Subscribed

    Jun 8, 2004
    4,160
    Edmonton, AB Canada
    Full Name:
    Sam Saprunoff
    #63 samsaprunoff, Jun 29, 2009
    Last edited: Jun 29, 2009
    Good day Craig,

    In response to some of your points:

    What about temperature and vibration chambers? There used all the time for system testing is some really extreme environments (sub-sea, subterranean, etc) and certainly validate one's design).

    There has been no talk of commercialization,etc and so how or if warranty is/will be offered is somewhat premature at this point. I believe, and I could be wrong on this, is that this thread is a basis of discussion and possible direction to create a new programmable controller for those who have special needs/requirements not currently met with with on the shelf systems.

    Absolutely!

    You would be surprised what some ambitious people can do... The time factor here would be dependent upon the experience of the people doing the work... For example, I typically spin very sophisticated hardware and firmware designs in 3-6 months... from concept to complete production. So, the time lines you presented could very well be extreme outside numbers.

    Agreed! Testing is the key!

    Again, you would be surprised at what people can do and how fast they can do it!

    Cheers,

    Sam
     
  14. mk e

    mk e F1 World Champ

    Oct 31, 2003
    12,899
    The twilight zone
    Full Name:
    The Butcher
    I'll have to contact them...Wil had that same idea.
     
  15. pad

    pad Formula 3

    Sep 30, 2004
    1,418
    Tequesta, FL
    Full Name:
    Paul Delatush
  16. Pizzaman Chris

    Pizzaman Chris F1 Rookie

    Mar 13, 2005
    3,919
    New Hampshire
    Full Name:
    Pizzaman Chris
    Mark,

    I just read this thread and got really dizzy from it. :eek:

    Good luck on it. I think you got alot of smart folks to help you with it.

    I know i will need one too, but i'm hoping the Haltech E11V2 will work for me.
     
  17. glasser1

    glasser1 Formula Junior

    Sep 2, 2006
    510
    Oregon
    More thoughts:

    In addition to selecting a CPU and appropriate development system no one has mentioned the following:

    1) Design and build a separate hardware interface board. The OTS (off the shelf) development boards can send out and monitor logic levels and, for those with on-board A/D converters, standard dynamic voltage ranges of 0-5v, etc. So you will need a separate board that comes after the dev board that does signal conditioning for all the sensors and contains required amplifiers to drive outputs for EFI, etc. I am sure some will say the slick solution is to build both from scratch on one board, but that adds too many problems for a one-off prototype effort. Go with a standard dev board that people can easily use to develop and bench-test firmware while others are building another board to interface the final I/O signals to a real motor.

    2) Software to run on a laptop that talks to the ECU firmware and allows you to monitor sensors, log data, and do real-time mapping while the engine is running. After you have spec'd what you want I would have someone write it in either Visual Basic or Visual C++.

    Now the ECU firmware will also need a component to manage USB communication for #2 above. Also, I would at least put in the hooks for communicating with a standard OBDII port so you can log error codes in the future. These are all things to keep in mind when deciding what CPU to use.
     
  18. mk e

    mk e F1 World Champ

    Oct 31, 2003
    12,899
    The twilight zone
    Full Name:
    The Butcher
    Sorry I'm slow to respond this week....I've had company and projects.

    Reasons like these make me think the only real possible way to go on a custom ecu is a version of the diyefi set-up.....there is too much to do otherwise it seems like.


    Before we go any farther on the custom path I think I'm going to contact the motec poeple and see it they think the high injectors on relays would work or it they can make the code changes to make it work. They have all the other features I want except the 24 injectors......it would be by far the easiest option.
     
  19. pad

    pad Formula 3

    Sep 30, 2004
    1,418
    Tequesta, FL
    Full Name:
    Paul Delatush
    “Before we go any farther on the custom path I think I'm going to contact the motec people and see it they think the high injectors on relays would work or it they can make the code changes to make it work. They have all the other features I want except the 24 injectors......it would be by far the easiest option.”

    Good choice
     
  20. glasser1

    glasser1 Formula Junior

    Sep 2, 2006
    510
    Oregon
    When 360 ECUs start to go and the options for replacement are few and very expensive then this will become an option. Some folks consider this kind of a challenge fun. CANBUS analyzers are cheap. Even more fun would be to build an F1 controller as well.
     
  21. glasser1

    glasser1 Formula Junior

    Sep 2, 2006
    510
    Oregon
    Another possible option someone just referred me to are these folks...
    http://www.fuelairspark.com/
     
  22. glasser1

    glasser1 Formula Junior

    Sep 2, 2006
    510
    Oregon
    #72 glasser1, Jul 1, 2009
    Last edited: Jul 1, 2009
    How effective can this be without being able to throttle back the air? Is leaning up the mixture by just dropping fuel not going to cause excess heat as well? I'm not familiar with how a "soft rev limiter" works, but is that really what is done - lean up mixtures cylinder by cylinder?
     
  23. mk e

    mk e F1 World Champ

    Oct 31, 2003
    12,899
    The twilight zone
    Full Name:
    The Butcher
    You need fuel, air, and spark to fire a cylinder and if you are missing any one of them the cylinder will not fire. What a hard rev limiter does is stop either the spark of the fuel on all cylinders at the preset rpm....just about sending you through the windshield if you happen to be in 1st gear or unsettling the chassis in any gear if you're in a turn. Pre catalytic converter the spark was turned off dumping the unburned fuel out the exhaust, but that would overheat the cat so now they stop the fuel. Not just a little causing a lean fire condition and the heat that goes with it, but completely causing a no-fire condition.

    A soft rev limiter is basically a proportional control method where the closer the engine gets to the hard limit, the more the power is cut. This usually happens over about 200 rpm so at 200 before the hard limit you start killing cylinders at so time interval and the closer you get to the hard stop the shorter the time becomes. Basically it’s a PWM control scheme where interrupting fuel or air interrupts the torque just like interrupting current interrupts the torque in an electric motor. I’m pretty sure from the should the f1 guys were doing it this way as it uses less fuel and keeps the engine cooler.

    The other way to do traction control it is to retard the timing which reduces the engines torque output but it does it by reducing the engines thermal efficiency causing all the lost power to be turned into heat. This is an analog solution like putting a variable resistor in line with an electric motor, you get speed control but you get a lot of heat too. This method is smoother and is probably the only practical answer on a 4 cylinder engine but on an 8 or more cylinder I seriously doubt there would be any noticeable difference in smoothness.
     
  24. hotrod406

    hotrod406 Formula Junior

    Sep 18, 2007
    540
    Grand rapids area,MI
    Full Name:
    Tim
    Newer cars have electronic throttle bodies so they can limit the air along with the fuel to avoid a lean condition.

    Dropping cylinders is not going to be smooth on any engine. I've gone up against an MSD rev limiter on a V8 and it is definitely not smooth. Retarding the timing is going to work the best imho. That's how every traction control system before electronic throttle bodies works. Some use the ABS too at low speeds but that's not applicable here.
     
  25. mk e

    mk e F1 World Champ

    Oct 31, 2003
    12,899
    The twilight zone
    Full Name:
    The Butcher
    The problem with using a servo controled TB is it just cannot respond fast enough because it has to physically move something with mass and inertia vs electronics that move at nearly the speed of light. OEMs are doing it now but I think it's always in addition to something else.

    Cutting fuel does not give a lean condition on a sequentially injected engine, the injection pulse never occurs as there is no fuel at all going into the cylinder so it could not possibly fire. It could and would cause a lean condition on an engine that is anything less than fully sequentially injected....on a non-sequential set up ignition cut really should be used instead of fuel cut.

    Are you talking about a soft of hard rev limited on an MSD box?....I think they are all hard limiters aren't they(?) and I agree a hard limiter would not be acceptable for traction control in any way.

    My last 2 VWs use the abs system.....and going up an icey hill were the system is in a prolonger use will just plain fry the brakes so I have to remember to switch it off when I start up a hill. This type of system is probably the most accurate but is no good for perfomance applications were there is enough power to spin the tires at will or often.

    The motec ecu uses timing retard....but I'm pretty certain from the sound the F1 cars were making when they had traction control it was a cylinder kill system. When you do it this way the engine stays cooler and the fuel stays in the tank instead of being turned into waste heat.....but you need a high cylinder count for it to work propperly.
     

Share This Page