Gauging Interest: Open-Source/-Hardware Engine Management.

[Mykeys Toy]

I need not give my opinion you fellas are clearly beyond me.. I love the idea.  Keeping it under $100 would be cool but If you can make it do half of what you want and more later I would pay twice that without a qualm.  Subaru especially older ones are so left out it isn't funny there is support fro darn near everything else why not us.. It has to start somewhere. 

 

Mike

[Twitch de la Brat]

Actually AMD as Intel is a modified x86 processor, but anyway, I was thinking with the main CPU being an x64 and then the subsidaries being the low level processors you speak of, you could remap the driving computer on-the-fly between changing songs

And frustrations are a thing of the past for me with Subarus, I've gone straight to threatening the car with a multimeter and a handgun

And the flash based memory storage system would handle being drawn on no problem as you mentioned

 

Twitch

[NorthWet]

CISC processors don't like the heat.  Ever see that YouTube vid of the AMD processor powered-up without a heatsink?

 

Mykeys Toy:  Your opinion is far more useful then my bloviating on obscure tech stuff.  I may know a fair amount about the inner workings, but it doesn't mean jack if that knowledge does not lead to something useful.  You know what is useful to you.

 

IMHO, one of the biggest problems with tech is that the self-proclaimed "gurus" want to be priests relied on by the great unwashed masses.  Tech is just everyday mechanics that are put together in a clever, useful way.  It should be shown no more awe then would a toaster, and should be played with like Lego blocks.  Figure out what would be useful, and lets find the blocks to put together.

[NorthWet]

Twitch, I did some looking at the Raspberry Pi, and my impressions (worth maybe 2 cents ) is rather mixed.  It seems a very interesting and powerful unit with good price/performance.  It is intriguing to me for use as a computer, but It seems less well-suited for a process-control role.

 

My very first impression was that it is marketed as a "boutique" item:  Cutesy branding (names and logo), branded accessories (e.g. a backpack that looks like a raspberry), and limited add-ons (meh... newish product).  It's targeted demographic seems to be people that want a cheap Linux box with great graphics.  It gave me the impression of trying to have the cache/exclusiveness/coolness that Apple wanted to surround their iMac.  I am more hyper-rational than hip, and the marketing left me feeling out-of-place.

 

Now, some concrete issues.  Currently, the hardware is tightly controlled, with one source for all units.  The Foundation that produces it has exclusive deals with (IIRC) 2 distributors in the world, from which resellers can buy units to sell to us. The UK firm Broadcom is the producer of the SoC (System on Chip) that the RP uses, and they seem to tightly control availability and information on their chip.  I have yet to find any way to purchase this chip, which is not surprising since the RPF's FAQ says:

  "But I demand the documentation for the chip. Give it to me!

To get the full SoC documentation you would need to sign an NDA withBroadcom, who make the chip and sell it to us. But you would also need to provide a business model and estimate of how many chips you are going to sell."

(source: http://www.raspberrypi.org/faqs )

 

This kind of stuff disinclines me to think of the Raspberry Pi as a jelly-bean unit.  I think it might make a great user interface for an ECU, but since it's video is HDMI-only, I don't know what display devices would be practical.  It is still a new product line, and hopefully it will grow over time.

 

Thanks for getting me thinking about the Raspberry Pi.  It looks like a great value for stuff that I could do.

 

Cheers!

Edited June 24, 2013 by NorthWet

[Twitch de la Brat]

Ah, I didn't realize how locked down the platform really is.

And the HDMI out for visual isn't that big of a deal, thankfully, as smaller and smaller displays are supporting the HDMI standard.

As for the "niche" thing, the manufacturer wasn't aiming for that, it just grew around the product, so they went with it. Gotta follow the money trail. Anyway, they were aiming for a small, ARM based SOC which was easily accessible for most anyone to make a computer out of.

I'm pretty sure the creators were part of the open computing initiative.

 

Anyway, you're right, Pi is meant more for media and entertainment than menial controller tasks.

Arduinos are looking better and better...

 

Twitch

[ivans imports]

I whould be interested and have  test pilot cars being able to addjust timing or add extra sensers howbout being able to control a 4eat with the same ecu as engine ? thiss is a wall for me. Also boost control extra feul control and nos control rev limiters and over preshure limiters ect

[NorthWet]

Twitch, make sure that you make your own assessment of anything from me that doesn't involve a lot of think-time.  I am an iterative thinker rather than linear thinker:  I turn around and around a thought trying to look at it from all angles, and it can take some time to reach a truth.  It makes me very good at problem solving, but crap on snap judgements.  Good to hear about the HDMI displays.  Much of the MCU world seems to favor simpler LCD displays (with a lot of them being touch screens ).

 

Ivan, my aim is to make a gateway to stuff like you mentioned, rather than just reinvent the Subaru ECU.  If I/we/someone gets the platform in place, a lot of the enhancements should be near-trivial.  The ignition/injection function can become our daemon, doing the bidding that we want.  Lots of possibilities once a platform exists.  And so much of the electronics needed to do it are available in mass-market ICs.

 

And yes, if we have a good insight into how the TCU interacts with the 4EAT, we could build a module or unit that will handle that.  The processing power is available, as are the solenoid drivers and monitors.  Perhaps the best way to approach this is to make a module that audits the TCU's inputs and responses during actual driving.

 

Cheers! 

[Twitch de la Brat]

I think what we need to do first is build a piggyback system that records/mimics the signals from the various sensors and senders.

That way we can strap it to a functioning car and see what all signals we're supposed to be getting rather than making up our own.

 

Twitch

[NorthWet]

Yeah... why reinvent the wheel if you can clone it?  One of the possibilities is to feed synthetic sensor data to a bench mounted ECU and record its responses.  A lot of information could be gathered without (directly) burning any fuel. 

 

I have another thought along the same line, but I need to do some more research first.

 

I am satisfied with my proto-design for the crank/cam sensor interface"breakout board".  It uses through-hole resistor and capacitors rather than cheaper, smaller SMDs, so it could made smaller than its current 30mmX30mm size.  The Maxim IC will be a pain to breadboard for testing, as it comes in a QSOP-16 package, which I haven't been able to find any existing carrier boards for it, I think that the fine spacing between the leads won't work well for a home-etched board, and I would prefer to validate the design before I have PC boards made for it.  I may end up using an X-acto knife to cut the gaps between leads.

[Twitch de la Brat]

Hmmmm...

Bread boards are so rudimentary though.

Its too bad somebody doesn't have access to a solder dip tank...

 

Twitch

[ivans imports]

could it just control the stock trans tcu ? 

[NorthWet]

Twitch, the breadboard is just to make sure that I have the basic concept functional.  Since I haven't etched my own boards in decades, and the cycle time for PCB Service companies is a couple weeks (mostly shipping transit), I see no other practical option.  Once a design is basically sound, professional boards would be made (at a couple of USD/board).

 

Ivan, I am not certain what you have in mind regarding controlling the stock controller.  My guess is that you want to "spoof" it, to give it synthesized inputs to control its output; if so, then this might be a little more difficult then doing a straight replacement...  I don't know.  The controllers can simulate any reality, so almost anything that doesn't violate physical laws is possible. "Practical" may be a little different.  For a straight replacement, the main thing that I would be concerned about is not so much what the TCU does but how it does.  The duty solenoids are PWM-controlled, so we should know the pulse-rate and any other oddness from the solenoid-drivers.  My understanding is that the "POWER" resistor affects the solenoid signal(s), somebody saying that it controls an overlay-signal on the main signal.  I don't know, and it would be something to find out.  It should all be doable, just need to gather some data.

Edited June 25, 2013 by NorthWet

[NorthWet]

Anybody interested in checking-out/learning the nuts-and-bolts side of Open Hardware/Software might want to check out a few sites...which are likely to lead you to other sites.  To me, it is refreshing to see the free exchange of ideas and products, without the usual proprietary nonsense and pretending that it is all magic.

 

Some suggestions:

adafruit.com

elecfreaks.com

sainsmart.com

sparkfun.com

 

For checking-out component  (ICs, connectors, etc) specs/prices/availability:

digikey.com

mouser.com

 

The above are not endorsements, merely good starting points.  SparkFun has what appears to be a nice tutorial on the basics of using MCUs... might be worth a look:

https://www.sparkfun.com/tutorials/57

 

Cheers!

 

Edit:  Even if you aren't so interested in the nuts-and-bolts, the first set of sites can give you some idea what is already available and/or possible.  Lots of it is geared towards playing with robotics, but it is still applicable to vehicles.

Edited June 25, 2013 by NorthWet

[ivans imports]

See right now only have 1rst / 2nd and reverse maualy controled by switchs no tcu forces me to run max line preshure all the time and cant shift on its own

[NorthWet]

I am sure that we can cobble-up something better than 2 switches and no TCU.

[ivans imports]

have been fighting the tcu for years was always defaulting to 3rd half track couldent figure out why thick i overheated the temp senser or somthing but it made it unreliabble to run tcu at least now its garented to work. One switch on the shifter in 1rst out 2nd               3rd 4rth might be scary

[NorthWet]

Although a slight distraction from this project, I came across something that might eventually benefit it:  One of the areas of interest in the "Open" universe is making 3D printers. (More specifically, ones that can replicate many of its own parts.)  One offshoot is machines that do CNC milling, and CNC PC board etching.  I watched a Youtube vid of a machine made by a teenager, and was simply amazed.

 

This could make producing prototype boards at home much easier, and without noxious chemicals.

[heartless]

I cant really add much to the conversation, other than to say I think this is a really neet idea, and I would love to do some testing - as a "general public" type of tester, of course...I have zero programming skills and only a rudimentary understanding of programming. but testing for/as "general population user" - that I could be good at! LOL

 

One thing I did not see mentioned in the sensor discussion, however, was any reference to the EGR system? it would be nice for those going from EGR to non-EGR (like me) to be able to "tune" that out - or is this even possible?

 

Ok, I am going to go sit back in the corner and watch with interest now...

[ivans imports]

Just need to make more power and more power and more power up agaist a engine mangement wall

[NorthWet]

Ivans Imports, not much power to be had on this path, except minor ones from efficiency or perhaps more by removing rpm limitations.  As far as TCU issues, let me know what issues you are having and what you would hope to see:  There might be something quick-and-dirty (lacking the street refinement of the stock TCU) that could be put together so that you could play with it.

 

As far as progress on the main topic, I have wasted a bunch of time on a power supply/management module.  This may sound simplistic, but there are some useful things that can be integrated into management.  (e.g. - a separate, super clean power source for sensors.) 

 

After days of frustration trying to lay-out the board, I discovered that the chip I was using was superceded by what looks to be a simpler, cheaper, better-optioned chip... looks much more elegant... and still only US$3.  It provides a clean 5V (or 3.3v) at a couple amps, lets devices know when the voltage stabilizes on power-up, and provides CAN and LIN communication.  The chip is an automotive grade IC, intended to manage power in car electronics.

 

Hopefully, I will have the circuit planned and the board routed tomorrow.

[ivans imports]

so this will drive a realy borad that drives curcits or the chip controls realys that are part of same board

[NorthWet]

Ivan, which are you referring to, TCU comment or the power/supervisor chip?

 

For the TCU, using a jelly-bean MCU board connected to a secondary board ("shield" in Arduino terms) that has the solenoid drivers ( electronic rather than mechanical relays).  There may already be a "shield"/board that already has what would be needed.

 

Regarding the power/supervisor chip, it will have the function of supplying clean power to the other modules/mini-boards, and allow reliable standardized communication to the system of modules. (BTW, the concept I have is of these modules being around 2"(5cm) on a side, and the user plugging the needed modules together sort of like LEGO bricks.)

[NorthWet]

Ivans:  Basic TCU function seems practical. The FSM maps out most of the controls and actions.  The devil will lie in getting it as smooth as the original, as this will take some playing around to get the right firmness and quickness of shifts.  Best bet seems to be to use a (software/file) configuration for lazy shifts and then refine from there.  Looks like it needs 6 solenoid drivers and about a dozen light/indicator drivers.  If you want to lose the shift-select lever, then that will be a little more work to get the electro-mechanical actuators sorted.

[bluto5]

I will be building megasquirts version of what you have just described (megashift) sometime in the near future. id like to try a subaru auto as i have heard them described when built for high performance.

 

for whatever faults ms may have it has a large user community to sort out and devolope new software and hardware solutions to new ideas and problems that crop up.

 

sort of like this thread x100. with a x1000 archive of knowlegde and problem solving.

 

i suppose from your backround and view it may seem like a closed enviroment with weak hardware but from mine it is wonderfully open with hardware and capabilitie that are beyond my(and 98% of the population) ability to tune or make full use of.

 

almost everything that has been put forth so far as a desirable option is already ms capable and in the low price range you are aiming for.

 

if your lifes goal is to design and scratch build an ecu that will run an engine than you are on the right path. if you want to build a car to suit your idea of driving pleasure(that you will actually drive) then you are going the wrong way.

 

Sam

[NorthWet]

So, I thought that maybe I was out of touch with MS pricing, since I felt that there was a significant difference in costs between the 2 approaches.  I may be looking at the wrong places (first place I looked was DIYtune), but nothing came in for under $300.  That may be in the same price range to many, but to me there is a significant difference between $300-$600 and my target of $50-$100.  (If I wanted to go a single-chip solution for the SPFI community, it could be done for less than $20.)

 

I understand that many feel that the MS community would better suit them, and that is fine, but for many years I have been hearing about this person or that person planning on building an MS unit;  I have thought of doing it myself.  But in all of these years, I have personally known only one person who actually did it, and maybe 2 or 3 more that say that they have.  There seems to be a lot of talk but not much success.  For me, the thought of paying $250 (at the time) for a box of electronics components, still having to sort out wiring, and not having any base maps was kind of off-putting.  And, back then, the capabilities were whatever the designer said that they were, so tough luck if you wanted something different... like the ability to compensate for altitude changes without having to turn off your engine and restart it.

 

I am sure that MS units have improved, and offer more flexibility.  But the price is still rather high for spend-thrift owners of 30-year old econoboxes.

 

If you want to look at a community supporting itself, look at robotics.  Or perhaps better, the robotics people that are making their own 3D printers... printers that can print many of the "soft" parts needed to make more 3D printers.  The people that have successfully built their own 3D printers is probably 10 times greater than those who have built or bought MS units.

 

The point?  Proprietary process control, like MS, will always be a niche market, and it is likely to become a smaller niche market once the tech-savvy kids see how easy it is to take their robotics-community hardware and software wiki and make their own process control units.  And, when they are not driving their vehicles, they can unplug the ECU-specifics from the MCU and use it to run robotics races, or print something they need, or take it hiking as a GPS/radio/locator-beacon.

 

MS is fine.   I hope it works out well for you.

 

Too me, MS was a lot like Apple:  Tightly controlled, overly expensive appliance, losing market share as the wild-wild-West PC market became dominant.

 

For me, if this doesn't play out well, I can still make a CNC mill, and then a 3D printer.

 

Cheers to all, regardless of their view of my views!