Develope a AFR-meter

[chille]

Hello!

I just found the FreeEMS project, and I must say I already love it. Me, and a couple of friends, have thinked about designing a open source ECU. But it would be stupid to reinvent the wheel when i can contribute to the open source wheel.

About a year ago i started designing a wideband lambda controller compatible with LSU 4.2. Unfortunately the project died a couple if weeks later. But now, when I have found FreeEMS, I wondering if it would be a good idea to pick up the project again. The idea was to design a small (like 30x30mm) PCB with SMD components. The controller should ONLY be able to control a wideband oxygen sensor. The output should be both digital and analog. The digital is perfect to connect to other equipment, such as FreeEMS. The analog is perfect for those who don't run FreeEMS, but want to be able to use a data logger or dashboard meter/display.

Of course i would make it open source and use a licence compatible with FreeEMS, so it would be possible to integrade this controller with FreeEMS in the future. Or maybe just run it stand alone and hook it up to FreeEMS via the digital output.

Another question is if it is a good idea to use a separate MCU (i prefer AVR), or if it is better to use the FreeEMS CPU.

Separate MCU
+ The product can be used stand alone from FreeEMS
+ It easily to hook it up to FreeEMS via the digital output
+ The user can easily add more sensors
+ The user don't need to pay for control electronics for a sensor he's not using
+ We can at a later time integrate the software and hardware with FreeEMS (if i avoid assembly of course)
- It's a waste to use an external MCU if the main MCU in FreeEMS have capacity to do this
- We will, for example, need a separate voltage regulator on the PCB

FreeEMS MCU
+ We already have a MCU, voltage regulator, etc, it's cheaper to use this
- We can not run it stand alone

So what do you say? Is this a good idea? I Really want to contribute to FreeEMS. The easiest way for me to contribute is to develop something that can me made externally from FreeEMS, and integrated at a later time.

[shameem]

It is a very good idea to develop an open source, cheap and dependable wideband controller. Existing controllers are proprietary and costly. JAW is kinda cheap - but it's closed source and lacks active support.

[jharvey]

I guess I'd try to have my cake and eat it too.

Perhaps you could make an Arduino shield in KICAD that can also be used by the freeEMS stack.

This shield could perhaps be used as a reference PCB for an add-on card to the freeEMS stack, or perhaps it could have a pinout that would match the TA pinout. I'm not sure what the size of the circuit would be.

Can you sketch up a schematic so we can get an idea of what parts it might include and how big it might need to be?

[Fred]

Chille,

Welcome along, glad you like it.

I hate to admit it, but things are moving VERY slowly firmware wise at the moment, partly because I got a job, but mainly because my HDD died :-(

You are most welcome and indeed encouraged to do any FOSS engine related project here, esp non-ems projects like a wideband, datalogger, display unit, etc. I feel another EMS could dilute the resource pool for this one though.

I strongly think that the project should be stand alone and use the more common CPU, ie, the AVR that you mention. The FreeEMS CPU will be stretched to the limit to do what we want to do with it and having the option to build it separately is DEFINITELY a big plus IMO. VEMS has two wbo2 channels built in, and some people like that, but I feel strongly that the components should all be modular. IE, the EMS does only that. A trans controller ONLY the trans. Ditto wideband. Ditto displays. etc.

Once you see my post I'm going to move this to another section as it's not strictly about FreeEMS even though it could be integrated in the future.

I like the digital and analog idea (as will Abe) and think either CAN or SPI are goers for this. I've been thinking long and hard about comms stuff and will have to add this to my list of things to consider.

Thanks for signing up and sharing your ideas :-)

Fred.

[chille]

jharvey: I don't know anything about Kicad, I use Cadsoft Eagle. I put together a schematic + PCB containing the largest components. It's far from complete, but it gives an idea about how the final product will be, and what components needed.

Fred: How slow? Are we going to see an test-engine running with FreeEMS within 6 months? I'm building a Nissan 200SX S13 and had plans to use VEMS. But if FreeEMS could be stable enought to run a turbocharged engine, it would of course be the first choice.

Actually I'm working on a modular datalogger. CAN is too expensive, so I'm going to try RS485 instead. This will probably be a commercial product , but who knows, maybe I change my mind.

ESP/electronic limited slip differential, using the ABS, have also been on my project-i-want-to-do-list.




The board is 30x30mm and contains voltage regulator, MCU, opamps for current sense and buffering and a logical controllable transistor for controlling the heater.

[Fred]

I'd still like to see an engine up in 6 months, but not promises at all.

Dev has been pushed back a full month because of the issues :-(

Sean is still working on stuff in parallel though, and Jared on hardware, so the project is still moving forward, just not as fast as I would like.

The number one rule in diy efi is NEVER wait for a product. Always use the best available choice for your app at the time you need it. Also, waiting till the last minute to buy is a good move typically as if stuff does come out/change/appear you get it.

I want to run my turbo 2.0 on it in 1 year, so I'm working towards that with an intermediate goal of basic running well before then.

Time will tell.

----------------------------------

If you are planning PWM on the heater, be sure you have the right CPU type as some ADCs don't like noise...

In any case, transistor, or fet? The latter has much less potential to heat up.

Fred.

[Fred]

BTW, don't feel pressured to FOSS your datalogger. Commercial stuff is fine, just not the focus of the site. It's fine when it's obvious that it's commercial anyway. In other words, do as you please, but do post pics etc.

Fred.

[jharvey]

I'd agree, commercial isn't a problem. Just as long as you don't claim to be open source, then selectively pick people or groups to ban from it. That's not open. I of course would encourage for things to be as open as possible, but there are reasonable limits.

Those parts are quite small and could easily fit in many packages. You could probably decrease the footprint quite a bit by moving those resistors around a bit.

I haven't looked at eagle for several years. Does it include 3D these days? I seem to recall that PADS, Protel, and a couple others allow to export a 3D compatible file that can be imported into a 3D package, but didn't include 3D natively. I'm curious because KICAD is the first I've seen to include that natively.

Also have you seen the current layout for freeEMS 1.0? I'm looking for people to review it, looking for things like current draw, voltage isolation, appropriate net connections, ect. Anything that might be a problem when it's built.

[Delta]

I think like jharvey has said an Arduino shield might be the go here. Techedge's latest diy WBO2 is based on an Atmega168 - which is whats on the Arduino deciemilla (and I have a feeling they have used arduino themselves for the prototyping). I've got a few Arduino clones lying around here as I'm working on a gearbox controller based on one. Another micro to consider - and the reason why the arduinos are now lying around at my place is the Atmega644p, its compatible to the arduino environment (easy programming) but it comes as a 40pin DIP - I2C is not shared with ADC, PWM outputs are not shared with SPI etc etc, it also has 2 serial outputs, which would come in handy - one for logging, one for a gauge/display.

[toalan]

I am almost 100% positive that the TE unit does not use ardunio. TE came out with their first WBs using AVR well before ardunio, so moving to ardunio for their new products would be a step backwards IMHO.