Car Electronics, PCM / ECU / EMS

[TwinTurboLs1]

So I am curious as to what it would take to make a circuit that would translate analog data from cars output sensors like Oxygen, Mass Airflow, Crank shaft position, Engine coolant temp into digital data and get it into a linux box?

I have found this board here

**broken link removed**

After calling the company, it will accept data at a high enough resolution to hand a motor spinning 10k, the money for this project is not really a issue.


Is there anyone that can point me to information on converting analog signals to digital signals? Some of the signals are variable voltage, some are variable resistance.

 

[jrz126]

Are you looking to make your own dyno?
Also what ECM are you running? you may be able to get what you need from the interface (OBD II) It should already have converted it to digital for you. No sense reinventing the wheel... I've experimented with this but I have a 95 Monte Carlo, so Im stuck with the intermediate OBD 1.5.

As for signal conversions, you'll need some Analog to Digital converters.

Nice username...I cant wait until I'm a rich engineer and I can own some of those motors.

 

[DirtyLude]

The easiest (build your own) would use a microcontroller with ADC to gather the data and output it where you want; rs232, LCD display...

There are plenty of off the shelf generic data aqcuisition or datalogging boxes out there that you can adapt to your purpose, though.

like Jrz126 said, if you can get it out of the ECU with an OBD II interface, that would be the easiest.

 

[bmcculla]

**broken link removed**

This board has 16 bit Analog to digital converters. The board in your pervious post only has digital IO.

If money really isn't a concern take a look at Labview. www.ni.com. Its a super powerful graphical programing environment. National instruments has tons of A to D boards. The measurement computing board is designed to work in Labview also. Labview allows you to build really complex aquisition and control applications really quickly and easily.

 

[TwinTurboLs1]

No I am not trying to read data from the stock PCM, or make a dyno.

I want to make my own standalone engine managment system that replaces the stock PCM 100% and has the ability to read twin MAFs, have intigrated controllers to fire additional fuel pumps, or nitrous, etc.

 

[TwinTurboLs1]

http://www.measurementcomputing.com/cbicatalog/cbiproduct_new.asp?dept_id=353&pf_id=1538&mscssid=9CF2RSKFCTN08M0M9BECELV48GUKAC94

This board has 16 bit Analog to digital converters. The board in your pervious post only has digital IO.

If money really isn't a concern take a look at Labview. www.ni.com. Its a super powerful graphical programing environment. National instruments has tons of A to D boards. The measurement computing board is designed to work in Labview also. Labview allows you to build really complex aquisition and control applications really quickly and easily.

I would be writing the code in C++ on Linux, this is a very real time critical application as it would need to sample around 50,000+ times per second when the car is at its redline, and faster if it got over revved (And of course, the code would pull timing, spark, etc to limit revs and damage)

That NI one looks pretty good, it carries about a $1300 price tag, and it looks like it does what I need, but the MMC one also looks like it can do what I need.


The MMC one says it can work as a oscope, which means that it (out of the box) is good enough to read any sensor in my car and let the software decode it.

 

[DirtyLude]

No I am not trying to read data from the stock PCM, or make a dyno.

I want to make my own standalone engine managment system that replaces the stock PCM 100% and has the ability to read twin MAFs, have intigrated controllers to fire additional fuel pumps, or nitrous, etc.

Engine management does not require that powerful a system to run. Running it off a full Linux system is way overkill. The hardware and circuits for interfacing to all the sensors and driving injectors and spark is 90% of the ECU. Not to mention that with a PC based system you have boot up time and extra hardware for power. If that PC loses power for an instant, your car is dead on the track/highway until the thing reboots everything back up again.

If you want a starting point, there are EFI projects out there using microcontrollers that can give you a better idea what it would take to make your engine management with the requirements you are looking for.

It's interesting as a project, but for the cost of a $1300.00 data card, plus Linux system, plus the cost of the driver hardware you'll need to put together, you can buy an off the shelf ECU that will do everything you want and be more practical.

 

[bmcculla]

If you want a really slick solution National Instruments sells a system they call CompactRIO. Its an FPGA based system with built in A/D and D/A. NI has literature about a company that used this system to prototype a motorcycle ECU.

I agree with DirtyLude - if you are going to use this ECU off the bench a completely embedded solution is the way to go. High level OS' arn't designed to do real time computations. They arn't very detirministic so ensuring latency is really tough (i.e its hard to make sure things happen at the right time) if possible at all. A slower embedded solution would probably end up with better latency.

For your solution i'd recommend Silicon Labs (SiLabs.com) 8051 controllers. They are really fast for an 8 bit controller - up to 100MHz and have really fast A/D and D/A all built in. For ~$150 you can get a complete dev kit. If you need even more processing power Atmel has a line of highly integrated ARM processors up to 66MHz. For even more number crunching power Freescale semi has a hybrid Controller/DSP chip 56F8XXX that runs at up to 120Mhz and is capable of really fast math (one 16 bit multiply and add per clock). They have a cheap Dev kit and a well priced C compiler.

 

[TwinTurboLs1]

If you want a really slick solution National Instruments sells a system they call CompactRIO. Its an FPGA based system with built in A/D and D/A. NI has literature about a company that used this system to prototype a motorcycle ECU.

I agree with DirtyLude - if you are going to use this ECU off the bench a completely embedded solution is the way to go. High level OS' arn't designed to do real time computations. They arn't very detirministic so ensuring latency is really tough (i.e its hard to make sure things happen at the right time) if possible at all. A slower embedded solution would probably end up with better latency.

For your solution i'd recommend Silicon Labs (SiLabs.com) 8051 controllers. They are really fast for an 8 bit controller - up to 100MHz and have really fast A/D and D/A all built in. For ~$150 you can get a complete dev kit. If you need even more processing power Atmel has a line of highly integrated ARM processors up to 66MHz. For even more number crunching power Freescale semi has a hybrid Controller/DSP chip 56F8XXX that runs at up to 120Mhz and is capable of really fast math (one 16 bit multiply and add per clock). They have a cheap Dev kit and a well priced C compiler.

It sounds like I have lots of options - my biggest problem is I do not know a lot about electronics, I am a programmer, which is why I was looking for a card that could read all the sensors for me and let me drive outputs, I know I will have to design some stuff (Or pay someone to do it) like a low impidence injector driver setup, and something to fire the coil packs (I have 1 coil pack per cylinder, 8 cylinders).

And there are no off the shelf engine managment systems that I can find that support twin mass airflow sensors.

 

[DirtyLude]

And there are no off the shelf engine managment systems that I can find that support twin mass airflow sensors.

Twin ECU's support twin MAF sensors.

I'm assuming this is a twin turbo 8 cylinder, based on your nik. It's common to run two ECU's in this case, treating the engine as 2 4 cylinders. You can run 2 fairly simple ECU's and have lots of spare inputs/outputs.

 

[Juglenaut]

Lab View can and will with proper VI set up you can I/O read-write ports and lines.

Most Cars made use a 5 volt reference on all sensors. Some issues with NI card is Duty Cycle, lines and ports sometimes have to be conditioned, CPU speed is a factor as well.

Our class is studing Lab View, and it seems to work very well. Right now we setup VI's to control Stepper motors, and also read the position of the stepper to the same VI.

I also have been working on a injector project.

 

[taurus3]

hi! does any one know about making a simplest car security. If so please share your ideas.

 

[DirtyLude]

hi! does any one know about making a simplest car security. If so please share your ideas.

This is an old thread that doesn't have anything to do with car security. Make a new thread with your question and be specific as to what you want to achieve, or what you want help on.