Design an ECU project?

In my mechanical engineering program, I custom designed an engine control system for the UC San Diego Formula SAE team race car. It was based on the Electomotive TEC-II ECU and coil pack for a 4 cylinder. I looked on their website and they have an application for a 12 cylinder, coils, etc. The software was easy to program, they provide base-line fuel and timing curves to at least get the engine started. It is based on a speed density closed looped management system and uses GM sensors and fuel injectors based on a specific resistance. You can have batch fire or full sequential fuel injection (for a smoother idle and throttle response) and just uses 1 O2 sensor.

Check out their website: http://www.electromotive-inc.com/products/tec3.html

Hope that helps save you some time.

 

Jon,
I've looked at these. To handle my 24 injectors I would need at least 2 of them then need to buy a TC sytem somewhere else and try to get everything playing nice together. It's a difficult path, but possible.

What I've decided to do is use motec stuff. The best option is to I wait for the new motec M190 to be released which should happen in about a year and will handle my 24 injectors in 1 ecu.....I'm trying to get on the beta test list but so far haven't had much luck.

The other option is to canbus link 2 M800s so they will share sensors and once set up I can do most of the tuning as if it were just 1 ecu. It's a little tricky to set up, but the newer software versions do allow it and it has been done and tested.

 

The traction control solution is simple-- buy a RaceLogic box (or two, if necessary, probably wouldn't be). You can configure them to either do a progressive soft fuel cut (getting more aggressive with more slip), or ignition cut. The fuel cut option is very "factory-like" and is safe for use with cats. The ignition cut is handy if you're into fireworks.

Currently the biggest box they offer controls 8 injectors/coils, which with 24 injectors might not be quite enough, but you can easily run two units. Control over 16 will be plenty, and you can map the system to use whatever cut methodology you like. You also get launch control.

Implementing TC at the ECU level is a royal PITA, and it's generally no where near as good as RaceLogic's solution, and I include the MoTeC flavor in that. Secondary throttle plates and DBW throttle control is crap and far too invasive. You want to be cutting injector/coil pulses for TC.

As for the ECU itself, I'd just use an EFI X5 and call it a day. If you reallyreallyreallyreally had to have true 24 channel sequential, run two of 'em. There's NOTHING on the market that in any way compares to the quality and reliability of the EFI boxes, IMO.

 

I didn’t realize there was a more expensive solution than motec….but it looks like you found one for me that would run about double and take me from 1 or maybe 2 boxes up to 4. I’m sure it’s good stuff, but it’s a lot of stuff, a lot of set up and a lot of money.

I will most likely end up with the coming M190 motec and I’m trying very hard to get time worked out so I can be on their development project list and get one for free

 

How's that? The X5 does 12 injector sequential and 10 coils. If you absolutely need 24 channel sequential fuel, two should be plenty. Unless you're talking about running two CDIs as well, but you'd need to do that with motec as well.

Oh well, be a cool project to watch, nonetheless.

 

I will have 24 injectors, so 2 ecus and then 2 TC boxes is what I was thinking vs 1 motec M190.

 

The twin RaceLogic suggestion was assuming pulling spark/fuel on eight cylinders/injectors wouldn't be enough to provide meaningful TC. Chances are extremely good you'd need only one, and for what it's worth, the EFI boxes also do a similar TC function.

I just like RaceLogic for TC as it's such an amazingly good, well designed system, with some very clever software to set things up just how you want. I've never seen its equal, and I've played with TC on a lot of big power cars, as virtually every factory/OEM TC implementation is a "save idiot driver" system rather than a "go faster in the wet" system.

Besides, you could easily pick up two RL boxes for the cost of the motec "advanced" key for use with TC.

 

It seems like it would be a bad idea to only have some of the cylinders connected to the TC and motec only charges $375 for the option last I checked and it's the same option I'd need for the hi/low injector control anyway so the TC is free.

Have you looked at the motec software recently? 3 years ago the only option was ignition retard, but now they have fuel cut, ignition cut, or a comination of either and retard. It seems pretty good but I haven't tried running it.

 

Hm, last I checked the motec TC was part of the $3,000 "advanced" option group, and couldn't be purchased without it. That's interesting to note if they've changed that recently.

I've used the RL 8-cylinder systems on V10 and V12 engines, pulling 8 injector events is plenty for useful TC. Though, if motec's pricing has become considerably more sane of late, that's probably as good an option as any.

 

You know.....as much as I should I haven't forgotten this custom ecu thing just yet.

My current thinking is the DIYEFI open source project has some synergy and might make the custom thing make sense.

I started kicking it around a bit over there:

http://www.diyefi.org/forum/viewtopic.php?f=41&t=734

I need to learn C++ I think.

 

Here is what I'm thinking so far:
I've got a hardware designer full in and starting
I've got a guy who has a working open sorce user interface in to make whatever changes are required to make it work on this ecu

Firmware people or I need to learn how to do it.


General requirements:
1) Easy setup. This should be a 1 board solution if possible, with commercially produced boards available and no extra anything required but DIY board construction should be possible.
2) Finished product appearance when installed.
3) The smallest foot print possible…should compare to commercial option (I’ll get a number for this)
4) And Oooooh it needs to be fast with way more input/ouput capability than anyone will ever actually find a need for (what should this spec actually say???....something we can put a number to)
5) Water tight case
6) Surface mount components preferred
7) CPC or connector(s)

Software requirements:
8) Compatible with FreeEMS tuner
9) Traction and launch control possible
10) 32x32 tuning maps, with user defined points
11) At least 8x8 trim maps on each fuel/ignition output…16x16 would be better

Output requirements
12) 24 fuel High impedance injector driver
13) 12 low high impedance injector drivers and
14) 12 ignition outputs capable of driving coils directly (this may want a separate board, but must be in the main case.
15) IAC Stepper motor output
16) At least 8 digital outputs
17) At least 8 PWM out puts

Input requirements
18) Standard inputs for TSP, MAP, coolant temp air temp
19) 2 Hall and 2 reluctor triggers
20) At least 4 O2 sensor inputs
21) 12 inputs for ETG would be nice
22) 4 Hall inputs for wheel speed
23) At least 8 digital inputs
24) At least 8 analog inputs

 

We are back to the big question,

Who knows how to right code and can help? This is an open source/open design project so you will be helping SAVE THE WORLD from poor performing or over prices ECUs.

I've got a line on a couple guys that may be interest that I will follow up with but the more the better.

The second question is what changes are needed on the Design inputs list? This is intended as God's gift to the performance car world, the ECU with the power to meet all needs. So what's wrong or missing?

 

Safety. By your posts, I guess you aren't an experienced programmer in the embedded systems field. If I was you, I'd really keep my hands off the coding. A engine ECU is a highly safety critical element, and if you're taking the car out on the road, it is imperative that your code is perfect. Perfect as in "perfect", to a much higher standard than any Windows program. At the moment, I'm workign as an intern in a robotics company in the software department. 20% of my colleagues are only working on ensuring the control software's safety standards - and a robot is on a MUCH lower SIL (Safety Integrity Level) than a car...
Imagine you messed up somewhere (which you inevitably will), your software decides to fully open the throttle while you drive through a corner, you hit a SUV with a soccer mom and 8 kids - they don't need to die for you to get sued into oblivion.

 

I fully intend to keep my hands off the code and more importantly the ECU away from the thottle and braking systems making anyhting like situation you describe impossible. The only think at risk will be the engine itself....which is still not somehting I want to see damaged, but it is a money issue, not a safety issue.

I notice from your post you are most likely exactly the type of person I need and I'm guessing you have friends. Are you in? Can you get your friends in?

 

I use to write ALL my robotic apps in C++ ... not so much lately with some of the controllers I've been playing with though. I'll give it some thought since it means "saving the world" LOL.


cheers

 

Mark,

All the C++ programming I've done had libraries of functions that were included and sometimes specific to the hardware. I briefly looked at what you're buying/developing. Can you point me to the C++ libraries that are done and exist for the hardware that works and tell me anything else you might know about integrating these libraries with this other hardware you're looking at?

cheers,

Sean

 

Sean,
You were on my "people who need a PM list figuring you must at least know someone if not yourself who could help.
I can't answer your questions buy lets get you registered on the DIYEFI forum and get your questons answered with propper answers. It took them a long time to process my registration...they turned off the auto thing to try and stop spambots.

Go here
http://diyefi.org/

and try to register and I will alert them that they need to make this happen right now.

Here's the directions

If you are looking to sign up, please email me at fred (dot) cooke (at) gmail (dot) com and include your desired username and temporary password. I'll try to get it done ASAP and notify you when the account is ready. Thank you for your patience, this was necessary given the volume of spammers at the moment.

 

Thank you Sir, you will not regret youyr decision...ok maybe you will a little, but you are helping save the wrold, it's worth it!


Now, who else is in? It's SAVING THE WORLD we are talking about!

 

I sent them an email to register.

cheers

 

And I posted asking them to look for the email.


Is there anything you see missing or wrong with the design inputs? I want to lock this down in the next couple days so the hardware guy can run, he's chomping at the bit to build something way cool.

 

dude ... you're asking me? I don't care if a machine has 10 or 100 (or I mean 8 or 1024) inputs/outputs. We should be able to just duplicate the code if all of the pieces already work in some fashion if we have a processor that's capable of carrying out the tasks. That's simplifying but my take on the problem. You're not trying something with drivers or hardware that's not already been done right? Just combining a bunch.

I am gonna regret this ... but making your head studs was too easy mate .

cheers

edit: I generically lumped all your devices into "inputs/outputs" ... point being these should be proven circuits ... otherwise you're asking for trouble ... or f'd as I like to say .

 

What I'm asking is realy if there is somehting way cool the ecu should be able to do that you think I left off? The hardware guy will put the drivers on the board I ask for....but I have to ask and differnet thing want different drivers

 

There are now 3 programmers in.

Come on guys, you don't want to be left out, it's time to jump in!

 

Guys, it time to work so if you've signed up or ar thinking about it or willing to give advise you are needed here:

http://www.diyefi.org/forum/viewtopic.php?f=10&t=736&p=8550

Here is what I think is the (pretty) complete design inputs list:

General requirements:
1) Easy setup. This should be a 1 board solution if possible, with commercially produced boards available and no extra anything required but DIY board construction should be possible.
2) Finished product appearance when installed.
3) The smallest foot print possible…should compare to commercial option , under 160mm x 120mmx60mm preferred
4) Automotive grade components
5) Water tight case
6) Surface mount components preferred
7) CPC or connector(s)
8) On board memory for logging….say enough for 1 minute.

Software requirements:
9) Compatible with FreeEMS tuner
10) Traction and launch control
11) 32x32 tuning maps, user defined points prefered
12) 8x8 trim maps on each fuel/ignition output
13) 4D Fuel and ignition tables (RPM, TPS, MAP)
14) Closed loop O2 or Lambda control
15) Idle speed control
16) Boost control
17) Nitrous Control
18) Data Logging
19) Sensor diagnostic limits
20) Hi/Low and staged injection

Output requirements
21) 12 fuel High impedance injector drivers, may also be used as auxiliary outputs
22) 12 low impedance injector drivers (should also drive high impedance injectors)
23) 12 ignition outputs capable of driving coils directly (this may want a separate board, but must be in the main case.
24) At least 8 digital outputs
25) At least 8 PWM/variable voltage out puts (tach, IAC, ect)
26) 5V (sensors)
27) 8V (sensors)
28) 12V

Input requirements
29) 2 reluctor (triggers)
30) 8 digital inputs, 0/1v (hall trigger, wheel speed, ect)
31) At least 4 O2 sensor inputs, should accept narrow or wide band
32) 12 Thermocouple inputs (EGT temp, ect)
33) 8 digital inputs 0/12V (gear selection, nitrous switch, ect)
34) 8 analog inputs (TPS, MAP, ect)
35) 4 Temp inputs (air temp, coolant temp, ect)
36) 12V DC (approx 8V-16V)
37) 2 knock detectors

Communication Requirements
38) CAN
39) RS232

 

Good day Mke,

DIY construction ( Item 1 above) and using surface mount ( Item 6 above) can be mutually exclusive requirements. This would be totally dependent upon the size and pitch of the surface mount technology (SMT) components used, the assembler's skill level, and the equipment he has (MetCal solding iron, Hot Air reflow, inspection scope, etc). Consequently, a reasonable amount of effort will need to be used in the component selection...No 0402 or 0201 package sizes for sure... even 0603 are a bit challenging without the right equipment... I would recommend nothing smaller than 0805 packages for discretes. For the IC's I think you should be fine except for the microcontroller (MCU)/processor. Some of the lead pitches can be fairly small (less than 20 mil) and I would recommend that no BGA (ball grid array) packages be used (BGAs can be manually mounted with hot air reflow, but I would not recommend it for reliability... BGAs really should be stencil'd and placed in a reflow oven with the correct temperature profile).

On the communication side, I would also include USB as well. A lot of laptops these days do not have RS-232 ports, but have USB ports... USB implementation can be fairly easy (some processors have on-board USB interfaces, etc) and so its engineering burden can be minimal. If it was up to me, I would bail on RS-232 and replace it for RS-485 as it is more of a robust electrical interface and can be run in very noisy environments, long distances and multi-dropped networks.

Cheers,

Sam