What's new
Van's Air Force

Don't miss anything! Register now for full access to the definitive RV support community.

Open Source EFI

BruceMe

Well Known Member
My winter project is an open source EFI solution. It's based on the SpeedDuino project, but uses a more powerful variant; SPEEDUINO TEENSY 3.5 8X8.

A great friend CnC'd injector bosses compatible with my existing Mech-FI and it re-uses much of that system; feed-tubes, spider, electrical & mechanical fuel pumps. This system will remove the RS-style throttle-body, one entire mag/ignition and replaces it with the ElectroAir HAL crank sensor and replaces the bottom plugs with 12mm platnum. The open source EFI board has 4-port tuning and allows for in-aircraft fine mixture adjustments.

My failure mode is the remaining mag and a manual-value to dump fuel into the throttle-body, inspired by talking with Dave Anders.

I'll put together a site with all the details soon. What I can say is that the project is far less money than others (well under $1000), but there's a lot of R&D at this phase. I'm curious if this could be a community supported open-source engine fuel solution?

-Bruce
 
Intentions are everything...

I should follow-on with some disclaimers... the "EFI" wars in here are one crazy soap opera... my day job is fabulous, building and selling an EFI is totally out of the question for many reasons. I started researching this because I wasn't happy with any other solutions. If an open source solution does take-off, it will be up to the community or a vendor to fabricate the trickier parts (injector bosses and TB).

-Bruce
 
Curious...

"...I wasn't happy with any other solutions..."

This is NOT clickbait...

I am genuinely curious as to what you were not happy about with the currently available systems?

The major players in this field have been doing it successfully for many years...
 
I installed a megasquirt EFII in my Porsche after putting on a turbo. I can tell you that it is a lot of work coming up with a workable configuration and fuel map (granted, the turbo and variable fuel pressure regulator, made this more work for me). I can think of lots of times that I would pull over and adjust things over and over to get it right. Don't even get me started on mapping the initial start up fuel delivery volume (i.e. injector timing), compensated for temps to get it to start in the winter. The other gotcha here is the limited reliability of the O2 sensors in a leaded fuel environment. If you can't trust your AFR, you are doomed during initial fuel mapping efforts. I suppose auto fuel could be used initially.

I had considered this for my 6 and my 10. The inability to "pull over" and adjust made this a non-starter for me, though I did used the megasquirt for EI, which is WAY less trial and error adjustment. Pretty simple actually. If I had access to a dyno or test stand, I might have done the EFII, but I don't.

To me, as a guy who likes to make a lot of my own parts and systems, the real value of the OTS solutions is their base configurations and is worth the additional cost when you consider the consequence of getting it wrong and VERY limited, real world testing opportunities without those consequences.

Best of luck on the endeavor and keep us posted on your progress.

Larry
 
Last edited:
I have a lot of experience with EFI stuff, and could have "rolled my own". Pretty sure it would have worked.

But having to relearn painful lessons that others have, for one airplane, I thought the price of the two EAB EFI vendors was a bargain. Their individual cylinder fuel trim sealed the deal for me.

Sort of like an auto conversion. Do I want to do one? Yes, and I still might. But it, and a home grown EFI, are a project all their own.

Why not use a Megasquirt?
 
Until you get everything programmed, I'd think you'd want to have the existing system functional with the new system in parallel.............
 
My winter project is an open source EFI solution. It's based on the SpeedDuino project .......


I'll put together a site with all the details soon. What I can say is that the project is far less money than others (well under $1000), but there's a lot of R&D at this phase. I'm curious if this could be a community supported open-source engine fuel solution?

-Bruce

Thanks. Nice heads up on the SpeedDuino project. Looks good as a solid jumping off point for an a/c engine control.

Collaboration on things like hardware, packaging, wiring, connectors, configuration will will be especially helpful to make a reliable aircraft system out of some very good hobby electronics I would think; essential in fact, if it is to be other than a full time job.

On mapping, with the exception of reliability, an aircraft fuel system is really not very demanding, so I don't see fuel mapping as a big challenge. And one successful map is just a click away for another.

Keep us informed.

Ron
 
Bruce, go for it.

My impression is that the tubular magnetic pickup has not always been super reliable. For sure I didn't like the way they are fixated in the case, with a set screw driven into the side of the tube. Could be mistaken, but it's one reason I used ND pickups for my EDIS triggers. http://www.vansairforce.com/community/showpost.php?p=1060584&postcount=4

Sounds like you will try to regulate fuel rail pressure at something below the 25 PSI ballpark of a standard mechanical FI fuel pump. Eliminating the power draw of a full time electric pump is a moderate plus. Some will counter that the diaphragm pump brings its own issues to the game.

Can you get enough pulse width to flow the necessary fuel at the lower pressure?

I don't think dialing an EFI fuel map can be that big a deal, given it's done with buttons or a PC. Some of us old farts recall dialing in banks of carbs from scratch.
 
Last edited:
Big Reply

[snipped..]Why not use a Megasquirt?

My research started with the "squirts". But after running the tuning software on them, I was really disappointed. They are very detailed in complex, super/turbo and tweaking all kinds of subtle injector and ignition parameters, but lacked basic stuff I needed like injector tuning in the common (simple) models. My engineering sense said it would be a PITA. The Speeduino was much cleaner and simpler all around and had all the features I wanted in the standard distro.

Until you get everything programmed, I'd think you'd want to have the existing system functional with the new system in parallel.............

Kinda... So yes, the ingition is going to always be parallel (one slick + EFI's EI). But you can't run fuel systems in parallel... twice as much fuel is very bad. The existing mechanical fuel injection delivery system is largely unchanged (pumps, filters, fuel valve); but there is a second manual t-valve after the pumps (inspired by Dave Anders) that redirects fuel from the injector manifold (aka spider) to a straight throttle-body port as backup.... for _WHEN_ the EFI fails.

One aspect of this design I wanted was the ability to easily go back to the old mechanical fuel and slicks for two reasons... #1 - Failure is an option :\ #2 My long term plans include selling this some day I don't sell projects.

My impression is that the tubular magnetic pickup has not always been super reliable. For sure I didn't like the way they are fixated in the case, with a set screw driven into the side of the tube.

Thanks Dan for the support! I specifically chose the EA trigger because it looked relatively reliable compared to bolting an apparatus on the nose. So that's a bit disappointing. But I'll watch this.

Sounds like you will try to regulate fuel rail pressure at something below the 25 PSI ballpark of a standard mechanical FI fuel pump. Eliminating the power draw of a full time electric pump is a moderate plus. Some will counter that the diaphragm pump brings its own issues to the game.

Can you get enough pulse width to flow the necessary fuel at the lower pressure?

I chose the same injectors used on Harley's. They don't have a spray pattern and they have a high flow rate, which is adventagious. Yes, this is below the typical 40psi of many EFIs, but it is well within the spec for the flow rate. I could be wrong... proof will be in the running. If it's not enough... more pumps, more pressure!


This isn't my full time job, it's just something that appeals to my sense of efficiency and simplicity *of operation* and I will keep plugging away on it. The site I posted will have updates as I start installing components. There's lots more info there.
 
Thanks Dan for the support! I specifically chose the EA trigger because it looked relatively reliable compared to bolting an apparatus on the nose. So that's a bit disappointing. But I'll watch this.

I used hall effect sensors on my 10. I mounted them in an AL bar that bolts to the two forward most engine case bolts and pressed 11 magnets, set in brass sleeves, into the ring gear support holes for a 12-1 toothed wheel setup.

Larry
 
Last edited:
Ok, a question

For decades I've been under the impression that when you pulled the mixture for shutdown you actually cutoff the fuel to the carb (and FI would do the same). Thus I thought you could have parallel systems. Just pull mixture on original.


Is fuel cutoff or not when you pull the mixture?
 
Ok, a question

Is fuel cutoff or not when you pull the mixture?

It sure is... But my existing system is mechanical fuel injection, not carb. So I can't add another electronic injector. I am kinda doing what you're suggesting. I have a backup throttle body injector. This dumps fuel into the throttle body as a backup... So yeah I'm doing this.
 
I reviewed the Speeduino project. I like. Open source. Testing software, good support, cheap.:)

I am an Elektoniker, so your reference to a reostat connected throttle position sensor gave me some pause. You probably used the term out of convenience or familiarity, but there are good reasons to wire a TPS as a potentiometer and that is very likely your intention.

However, specing a linear pot for a TPS in that nasty throttle body environment might be asking for trouble. A linear pot with a slider seems to make sense, but how is it sealed? In the automotive world, the TPS, perfected to the limits of cost constraints, these buggers are still maybe the most troublesome of components under the hood.

Anyway, that got me thinking. By adding an additional pressure sensor, very reliable and cheap, it would be easy to compute throttle position by pressure depression and rpm, at least for our simple aircraft applications. The Teensy 3.5 MPU has gobs of head room to do it. The firmware is already off the beaten path oxy sensor wise, so some programming mods of fuel control path would have to be done anyway. And then if the TPS ain't there, it won't cause trouble.

BTW, the junk yards are full of throttle bodies with nice butterfly throttles and TPS's on well seal shafts that would seem to be almost ready to fly, and that is the path I will be following, so is the reason of your choice of a slider because of space restriction?

Ron
 
It sure is... But my existing system is mechanical fuel injection, not carb. So I can't add another electronic injector. I am kinda doing what you're suggesting. I have a backup throttle body injector. This dumps fuel into the throttle body as a backup... So yeah I'm doing this.

Bruce, let me throw a strategic thought in here.

Although we tend toward a culture of "backup everything", there is an argument which says additional components and complexity (an integrated backup system) can deteriorate overall reliability.

There is also a labor and cost requirement to perfect and test the backup system, which is time and money perhaps better spent perfecting the primary.

Personally, I'd put my efforts into redundant electronics, per standard practice with the kit EFI/EI systems.
 
Can you run speed-density and avoid a TPS altogether?

Probably, but it's just a POT resistor and I've built it into my slide throttle body. I think it's more responsive and it informs "accelerator-pump" style fuel boost.
 
[snip..]
specing a linear pot for a TPS in that nasty throttle body environment might be asking for trouble. A linear pot with a slider seems to make sense, but how is it sealed?

Same concern and it may not work :\ Nothing in my throttle linkage is rotary, so trying to tinker a very proven TPS into this was cumbersome. The linear slider seemed really obvious, but you're right, it's not sealed. So I built it into the throttle body slider and I'll externally encase it in RTV. This will keep the air sucking through it from the TB's vacuum, but oil and grease can still drop in from the slider. Honestly, this is an experiment. I'll run it for a while, dissassemble and if it's dirty... try another approach.

[snip..]
BTW, the junk yards are full of throttle bodies with nice butterfly throttles and TPS's on well seal shafts that would seem to be almost ready to fly, and that is the path I will be following, so is the reason of your choice of a slider because of space restriction?

Really really hard to find ones that worked (60mm, double flanged)... but I hear you. 99% of car throttle bodies have round air filter bores and I wanted both ends fanged, aircraft-style. If you have any recommendations... I'm open.

-Bruce
 
Curious what system you are using to manage the throttle body injectors fuel delivery/pulse.
 
Curious what system you are using to manage the throttle body injectors fuel delivery/pulse.

None... remember the TBI is the manual backup. There's a fuel-rail in the throttle body opening with small wholes. As one opens the slider valve wider, more wholes are exposed and more fuel mixes with the incoming air. It's simple. This backup fuel rail does not get any fuel when the fuel valve is off. It uses the EFI.

Throttle%2BBody.png


-Bruce
 
Last edited:
sorting it out

Re: open source EFI

I ran across another open source EFI thread that could have
possibilities for aircraft application. RUSEFI.COM.

It is more focused on making a unit that is in the direction of a
commercial product than the Speeduino, but shares some of its
heritage. And the software is open source as well.

The board development tool that RusEFI uses is KiCad. (google it) This
board development software (Electronic development Automation) is
free, which is amazing since much of this EDA software is in the
thousands or leased by the month. In any case, the tools are readily
available to do your thing if you are driven by electronic development.

However, the automotive hardware/firmware that is there seem to be
high quality and could be very close to what we need, so maybe, buy
the board from one of several contributor posting there or clone
something similar. Then play with tables and firmware. Refs to sources
for EFI enclosures and connectors are there that are of great interest
as well. It looks like it may be possible that a solid reliable open source
system could evolve without a career effort and on budget.

BTW, questions. Am I correct that a non-camshaft sensing EFI system
squirts once each rev? ( 2 squirts per intake cycle)
And, is there much advantage to cam input and timed injection?

Ron
 
Last edited:
I have a couple of Speeduino 0.4 boards I got a while back with the intention of building a throttle body injection system using two injectors up front at the throttle body. This was inspired by a friend who was testing a Surestart system a few years ago that started and idled well. I'm looking for suggestions on injector sizing for that sort of setup. My architecture would be two Speeduino boards, one running one injector at the throttle body, and the other injector ran by the 2nd Speeduino. Each board running two separate ignition modules. I would modify the code such that when only running on one system I would run a different map to handle the fuel and timing requirements when in that configuration.
 
I like the direction you outlined, Bob. I've been thinking along similar lines
of duplicate systems. It looks like a single injector in the 100 lbs/hr area
would be needed to get 180 -200 hp. Here is a link to a list of injector
specs. There could be several other lists out there, this one has some
good comments:
http://users.erols.com/srweiss/tableifc.htm

Ron
 
BTW, questions. Am I correct that a non-camshaft sensing EFI system
squirts once each rev? ( 2 squirts per intake cycle)
And, is there much advantage to cam input and timed injection?

Yup, and wasted spark similarly. I'm not trying to do fuel charge shaping or anything complicated. At sea-level full throttle it's going to be near 100% duty cycle.
 
I have a couple of Speeduino 0.4 boards I got a while back with the intention of building a throttle body injection system using two injectors up front at the throttle body. This was inspired by a friend who was testing a Surestart system a few years ago that started and idled well. I'm looking for suggestions on injector sizing for that sort of setup. My architecture would be two Speeduino boards, one running one injector at the throttle body, and the other injector ran by the 2nd Speeduino. Each board running two separate ignition modules. I would modify the code such that when only running on one system I would run a different map to handle the fuel and timing requirements when in that configuration.

I'm using the Weber / Magneti Marelli injectors (search for 'IWP-069'). 6gph x2, so 12 gph peak. That's a little light for an O-320 (very light for a 360). Good luck, sounds interesting too.

The link a couple posts above for the injector tables is _UNREAL_! I wish I had known that existed two months ago when I was trying to figure this out?! But it confirmed I picked ok.
 
Last edited:
"...I wasn't happy with any other solutions..."

I am genuinely curious as to what you were not happy about with the currently available systems?

I've been trying to answer this in my head since you wrote it. This isn't really about engine performance, quality or any "missing" features. It's really more about the experimental spirit and "hacking" the engine. I want to learn more and I would like to be part of an open sourced project solution to something that doesn't exist yet.
 
Re: open source EFI

I ran across another open source EFI thread that could have
possibilities for aircraft application. RUSEFI.COM.

It is more focused on making a unit that is in the direction of a
commercial product than the Speeduino, but shares some of its
heritage. And the software is open source as well.

The board development tool that RusEFI uses is KiCad. (google it) This
board development software (Electronic development Automation) is
free, which is amazing since much of this EDA software is in the
thousands or leased by the month. In any case, the tools are readily
available to do your thing if you are driven by electronic development.

However, the automotive hardware/firmware that is there seem to be
high quality and could be very close to what we need, so maybe, buy
the board from one of several contributor posting there or clone
something similar. Then play with tables and firmware. Refs to sources
for EFI enclosures and connectors are there that are of great interest
as well. It looks like it may be possible that a solid reliable open source
system could evolve without a career effort and on budget.

BTW, questions. Am I correct that a non-camshaft sensing EFI system
squirts once each rev? ( 2 squirts per intake cycle)
And, is there much advantage to cam input and timed injection?

Ron

Yes, crank based EFII squirts once per rev. 1 squirt per rev was used in the OEM world for 2-3 decades with very good success. There is no need for cam based injection in application like ours.

Larry
 
Back
Top