DIYEFI.org Forum

I'm here because I believe in freedom of hardware and software for DIY ECU. I was on the MS forum and I wanted to share with the group an industrial PLC that we cheaper than a MS disassembled kit.

WOW I have never seen a topic deleted so fast in my life. so here I am to share what I have found with you.


The Company Comfile Technology build a series of industrial controllers that can be programed in Ladder Logic or PIC Basic simultaneously if needed. All the units have several 0-5v / 0-10v analog input/output, PWM output, digital input/output and high speed digital counter input. Some of the produces even have a built in 320X240 touch screen.

The Unit I have in mind has the following processor speed and memory.
Microprocessor Atmega128 @ 18.432Mhz
Program Memory 80KB
Program Speed~36,000 instructions/sec
EEProm 4k

Its not as fast as the MSII but at $130 for a complete unit (Board and processor) I can live with the slower speed.
Board: http://cubloc.com/product/01_03cubase32m.php
Processor:http://cubloc.com/product/01_01cb280.php

There are also cheaper integrated units with relay/board/processor combo here: http://cubloc.com/product/03_02.php

Other option that could be used are as follows

Relay Board 8ch:http://cubloc.com/product/01_03relay8.php
Relay Board 4ch:http://cubloc.com/product/01_03relay4.php
Blue/white tone display:http://cubloc.com/product/02_02.php
expensive fancy color display:http://cubloc.com/product/02_05.php

What I believe is with that its allows us not to have to worry about hardware development but now can give our full attention to software.


There is one minor problem with comfile unit, all the of them runs on 24V but this could be solved with a dc-dc boost strap

What I am trying to do is under stand the the equation and process from efi332 so that I can write a program in pic basic which I will have to learn also.

what will really help me is if some one could provide me with a flow diagram showing me how the fuel equation are used step by step.
If not ill just have to continue figuring it out my self.

Nice, and thanks for posting. Also welcome along. I'm sure you're post won't be deleted here. I got banned for similar activities. Seems they don't like it when someone points out a flaw in their system. They self heat the thermistors with something like 30mW to 40mW in certain (normal operation) conditions while the sensors can typically only handle 20mW max. Not good for long term reliability, but good enough for many applications. I got banned just after I inquired about how to fix such an issue when this over heating is a problem, as my sensor didn't have the thermal mass that many other sensors do have.

It appears these parts have the potential of being formed into an EMS, do you know anyone that's done this yet?

I seem to recall there's a fellow that has made an Atmega128 EMS, perhaps that code can be changed to work on this platform. Also I think SECU is Atmega based.

jharvey wrote:I'm sure you're post won't be deleted here.

It sure won't, but it will be moved to the right sub-forum.

We welcome other ideas and designs.

As for fuel equations, you'll want to work them out for yourself because it very much depends on your var sizes and desired configurability etc.

Fred.

There we go it was freeECU noted in this thread. However, it was ARM based not Atmega based, so it can't be compiled and used in that above noted PLC device.

viewtopic.php?f=4&t=874

The PLC stuff has too much excess to be used for an ECU. PLCs are hella slow, no doubt this will be able to compete against plain vanilla PLC controllers on the market, but for automotive work you need more than 36k instructions per second. The AVR itself is capable of 1 instruction per clock cycle (based on Atmels marketing literature) and AVR can run upto 20 Mhz. A first order approximation 36,000/20,000,000 = 0.18%, that is you lose 98.2% of the speed of AVR when you run interpreted basic. My gut feeling is that you can create an ECU on an 8 bit micro running ~ 1 million instructions per second, but you have to be a bloody good programmer to squeeze every last drop of performance from your code. With 36k instruction per second you gotta be the cyber Jesus to pull off a miracle of that magnitude.

Edit: Suppose 10k RPM is the worst case RPM, that equals 166hz. Suppose you want to respond within 1 degree of engine revolution. 36000/(166*360) = 0.62, means you can only execute 0.62 instructions per degree of engine revolution @ 10k RPM.

toalan wrote:My gut feeling is that you can create an ECU on an 8 bit micro running ~ 1 million instructions per second, but you have to be a bloody good programmer to squeeze every last drop of performance from your code. With 36k instruction per second you gotta be the cyber Jesus to pull off a miracle of that magnitude.

ROFL, Alan, at the least, you gave me a good laugh! :-)

Edit: Suppose 10k RPM is the worst case RPM, that equals 166hz. Suppose you want to respond within 1 degree of engine revolution. 36000/(166*360) = 0.62, means you can only execute 0.62 instructions per degree of engine revolution @ 10k RPM.

This is funny too. The CPU is still scratching its head while the engine has done another lap, rofl.

Fred.