DIY-EFI

[WesRagle]

HI Guys,

I find myself without a project and have been trying to figure out what to do with myself. I really don't need another airplane so I've been thinking about improvements I could make to my Onex. After listening to the podcast one EIF for the VW, that sounds like a good candidate for an improvement project.

I've been following Peter's example and slowly gathering the bits needed. I've also been studying EFI in general and find myself wondering: If one designed an ECU from the ground up, focused like a laser on the existing fleet of VWs, what it look like? How would work?

A few of the many questions that come to mind:

1) Why aren't more people installing ECUs.
2) We have two ignition sources, why couldn't they be used to trigger injection.
3) If the two secondary ignition pulses were routed to the ECU why not provide true electronic ignition as well? Allowing the ECU to control spark would make for smooth ground operations while reducing the overall current draw by controlling dwell. If both ignitions are engaged (takeoff) the ignitions would operate in lock step.

On and on the questions go.

Now the big question. Is anyone else interested in this topic?

A good friend volunteered a Sonerai to be used as a test bed for whatever experimentation I wanted. I'm on the verge of "Taking the red pill" and seeing just how far this rabbit hole goes.

Wes

[lakespookie]

So a couple of responses,

Item 1. Electrical Dependence, Cost, No off the Shelf Components.
Item 2. your injector pulses are typically not in lock step with the ignition pulses, While you could do some fudging and make due using the ignition pulses as a cheap alternative to a crank positon sensor you would not want to do it with something that has fixed timing i dont 100% rermeber the timing layout of the VW in our configurations but i belive one of the timing systems is not fixed.
Item 3. To do Electronic Ignition you would want to make the spark triggers indipendent of any fixed timing system, of course you would still need to know crank position but a crank position sensor would work alot better and you could set up offset sensors for redundancy, That way you can cross check the crank position signal and be able to trigger the timing in lock step or as necessary for start up and other situations that would demand it for example you could have a knock sensor to retard timing protecting the engine or compensate for CHT readings to adjust timing and or fuel. The biggest challenge with all of this variability and EFI in general are the variable load fuel maps if you go that route and the related timing advance maps. In performance cars we typically use offsets of the base map to account for changes to the engine so that we dont have to tune at various engine loading and RPM ranges this makes life much easier as a tuner since the factory has gone to all the trouble of developing these tables it will get you close enought to the ball park that you wont hurt the engine. And then the real fine tuning is done at WOT to eek out the maximum amount of power from the engine. When i used to do this regularly on highly modified vehicles we would manipulate startup maps to facilitate agressive cams and high compression engines to assist with starts and idle. Some of the newer tools and some race applications were i had access to modifing valve timing and duration were alot more involved of course with off road vehicles you have the benefit of track testing and much greater telemetry than on a street car, Although i did play around with adding some rudimentary telemetry to some personal vehicles.

[WesRagle]

Hi Jaun (I think),

Thanks for taking time to respond.

I'm in the middle of nowhere looking for plumbing leaks left by the sub zero freeze we had a couple of weeks back and dont have a keyboard so I have to be brief. Assuming success:

1) The only missing component is the intakke logs. Electrical dependent probanly yes depending on backup. Cost,, TBD.

2) SDS uses batch injection so no cam sensor needed. I would propose "semi sequential", one squirt per rev per cylinder for even distribution per cylinder for smooth idle. Easy with wasted spark.
And "cheap" (inexpensive) is a primary design goal.

Spark triggers are always tied to a fixed reference. Phase shift and pulse width is a software function. Cam sensor is only needed for full sequential. I would adjust the secondary ign to TDC and start on that. Then once running switch to the other trigger and place the ignition in a timing trough for smooth reliable idle.

As far as knock sensor, O2 sensor, closed loop, ... I dont see that as practical. If EFI is going to ne accepted it has to be simple, reliable, affordable, and repeatable. Basically a bolt and go solution.

And yes, the VE, spark maps, acceleration enrichment, ... would have to be developed but once done for any standard installation, it's done. It is no good to deliver a pile of parts and tell the end user to "tune" the system. That is no way to improve safety/repeatability and enjoyment of a VW conversion. The only tuning would be fuel ballancing the cylinders.

I see no problem with reversion to batch fire based on primary ignition if the secondary ign fails. Again phase can vary. If both ignitions fail you're going to land anyway.

OK, I've pecked on this thing all I can. I'll be back to civilization late tomorrow.

I do appreciate any and all feedback. Please find an unfixable hole in the concept and save me a lot of time :-)

Wes

[BRS]

Wes,
I've installed the SDS EFII on three aircraft. First engine was a rotax 912 that had been converted to a 914 (turbo). It ran like a kitten after getting rid of those two bing carbs. The second one was a Sparrowhawk gyroplane which had a subaru frankenmoteur (2.5 block with 2.2 heads). It ran really well and started easy. My third and current aircraft is a Glasair Sportsman with a Lycoming IO-360. Again, I love the way it runs. Also going to the SDS eliminated any problems with vapor lock after a short stop to fuel up.

I would definitely do it again and on a sonex. Perhaps I will after I get this sonex finished and perhaps next year I'll be looking for another project.

So I mentioned how much I like their system, oh and Ross and Berry are very friendly and helpful on the phone, but there are trade offs. The biggest down side is the all electrical system. On the gyroplane (subaru) I had only a single computer and one plug per cylinder with just the battery and an alternator with a one-wire mod. The battery was a small 3 lb LIPOfe. No redundancy but then the gyroplane can land in like 20' and I didn't over fly densely populated areas. The Sportsman (IO-360) is on the other end of the spectrum. It has two alternators, two buss', two batteries, two ECU's, two fuel pumps. It's built to go places with precious cargo (family) on board.

Current draw seems to be about 11 amps. The fuel pump and injectors draw a lot of current. It does use a crank sensor (two actually) which I like very much. This is done with a few small magnets installed in the flywheel (starter ring gear). The timing is variable and is tuned each 200 rpm (something like that). I can go lean of peak at the push of a button. It's rather amazing. Push the LOP button and the fuel flow drops 26% (in my case on the IO-360) and the timing is bumped up a few degrees. I usually go from 11 gph to 8 gph and much lower CHT's in an instant.

The most difficult part of the installation is doing the initial tune. Really it does not take that long once you understand what you are doing but this is where damage to the engine can be made. Figuring out what timing maps to use is a difficult part then the mixture is tuned based upon an AFR gauge and O2 sensor in the exhaust. The knock sensors are not used in the planes as the propellers seem to mess with it too much. Also Closed loop is not often used in the planes either.

It's probably overkill for a sonex but it would be a fun project and that little VW would hum.

Good luck with what ever you do.

[GordonTurner]

There have been at least a few aerovees run from Ross” SDS system. In addition to the sonextalk podcast, which had a lot of good information, there was a thread here many years ago that was full of details from another multiple SDS aircraft converter. I would start with those threads. A few takeaways for me:
1. The crank sensor needs to be mechanically and electrically bulletproof
2. The intake logs can be machined for injectors amd this has been successful.
3. Obviously the electrical system needs to be well thought out and carefully constructed. According to Ross and everybody else I have talked to this is the area that causes most efi engine failures, the crank sensor is the second place.

You are in a good position for making this conversion as your plane is tested and proven. I visited Ross at his business somewhere west of Calgary and talked about converting my Corvair engine to efi. His advice was wait until you prove the airframe and are comfortable with everything before making big new relatively untested changes. So another year or two.

Good luck, hope you go for it. Gordon

[WesRagle]

HI Guys,

OK, back in town for supplies. Only three leaks. I got off pretty easy.



Hi Brock,

So I mentioned how much I like their system, oh and Ross and Berry are very friendly and helpful on the phone.

Everyone who has used their system says it's great. SDS obviously knows what they are doing and have been doing it successfully for a long time. My comment on "tuning" wasn't meant to be a snipe. It was just a thought bubbling to the surface.

As far as power consumption, that is always an issue with VW conversions. 11 amps seems about right.

For the VW cruising with one fuel pump:

Fuel Pump 4 Amps
Injectors (4 at 50% duty) 2 Amps
Sec Ign 5 Amps
----------------------------------------
There is your 11 Amps

You don't have to agree with this, but this is the problem/question as I see it:

How do you correct the poor induction/fuel distribution problem with VW conversions without destroying the VWs most desirable salient feature, affordability?

The only answer I can come up with is a more affordability ECU. EFI is an interesting project and developing an ECU is closely aligned with my professional experience. So, I can't help myself. I'm going to give it a shot.

What I envision is a speed density system as the primary ECU. Backup will be an independent alpha-s ECU, laid out on the same board. The backup ECU will be TBI driving a single injector and using the primary ignition to time the injection. That will keep current draw down to about 5 amps in an emergency. Communication will be via CAN bus with protocol closely patterned after ARINC 825.

Initial development will focus on the primary ECU. The ECU will "live" forward of the firewall. All components must be rated for 125 Deg. C. operation. The STM32L476 MPU will be used. I've used the Cortex M4 core for over 10 years and I'm comfortable with it. The STM32L476 has all the needed peripherals and the development system and evaluation boards are readily available.

So here I go. I seldom brought my work home so I have to get some tools. Where can I buy a good Electronic Tech. I suck at bread boarding :-)

Wes

[BRS]

You can do it with one injector but then that will not allow fine tuning. This is something I really like in the SDS. I can adjust each injector independently so that each cylinder peaks at the same time. But as you hinted, that kind of sophistication might be overkill for a sonex & VW engine.

[WesRagle]

Hi Brock,

Sorry, I wasn't clear. You're absolutely right. To solve the "problem" you have to have an injector for each cylinder. The primary ECU will have four injectors. I'm just a timid guy and I have to have redundancy. Using TBI as a backup in the event of a primary ECU failure, rather than a fully redundant ECU, is just a cost cutting measure. It also reduces the installation effort and the power requirements while running on backup. It was a requirement while selecting the MCU that there be sufficient timers that an independent timer be available for each of the four injector to allow for fuel balancing.

Most of the timers have four "input capture/output compare" channels. As it turns out, I think I will need only three pulsed inputs for the primary ECU. Two inputs for secondary ignition and one for primary ignition. All four channels of a 32 bit timer will be implemented so that if (sometime later) a cam position sensor is added sequential FI could be implemented.

BTW, at what power level do you hit the LOP button? Does the ECU ensure you are below say ... 75% power?

Thanks,

Wes

[WesRagle]

Hi Guys,

From humble beginnings ....

The supplies are starting to arrive.

The evaluation board for the STM32L476:



There will be two of these processors, one for the primary and one for the secondary ECU. The first $20 of the BOM.

A neat idea from Ada Fruit to quickly hook up for evaluation:



Just peal off a jumper and plug it in.

And the test bed:



One mag and Dyna S for secondary.

Just a little explanation about this project. I'm not quite as delusional as I might seem. I have no visions of grandeur. There is no money to be made here, it's a hobby. This project will be of roughly the same scale as building an airplane. Like building an airplane it's conceptually simple. All you have to do is read a few sensors, squirt some fuel, and throw a spark. But, as they say, the devil is in the details. So, for a season, my hobby money will go to Digikey and Mouser instead of Sonex LLC and ACS.

Worst case: I have an interesting/engaging project to work on. For at least a while it keeps me off the streets and out of the bars. But ultimately, the concept is flawed, and I give up.

Best case: The project is successful and we drag the VW, kicking and screaming, into the 21st century.

I've been doing this type of work for a long time. Yes, I owned a TRS-80 model 1 with a tape cassette for program storage. I also owned a Commodore 64 with a floppy drive that ran so hot it would melt the data right off the disk :-)

I'm perfectly capable of implementing the software. The learning curve will be what not how. Example: How big of a primer pulse is needed on startup.

So, why am I bothering you guys with this? Simple, I don't want to feel alone. I can't talk to my wife about it. Withing 20 seconds she will have an involuntary yawn reflex and her eyes will literally glaze over.

You see this type of initiative come and go on chat groups. But, I've never seen one come to a positive conclusion. Enthusiasm fades at the first sign of trouble. I hope this time it will be different.

If you're interested, please contribute. If not, ignore me. Periodically I'll mention upcoming needs. If you want to contribute some grey matter or material support, please do.

The first roadblock will be the intake logs. It is company policy at Sonex LLC not to sell individual conversion components and I understand their position. My backup is to use GPAS logs. I've talked to a machinist and he can tig some material onto the logs, face it, and machine it for injectors. We'll see. Ultimately for this project to be successful a permanent solution to this problem will have to be found.

If anyone has some Sonex intake logs you're not using I would be happy to buy them. Or, anyone who owns an AeroVee could call up Sonex and tell them the dog ate your intakes and you need a replacement set :-)

Anyway, I think this will be a fun ride.

Wes

Edit: I got an inexpensive O-scope. Four channels @ 50 Mhz. Plenty good for this effort. And I can post screen shots to facilitate discussion and clarify questions :-)

[NWade]

Cool stuff, Wes!

Have you looked at the open source automotive projects along these lines? It’s been a few years since I was involved in racing, but the “Megajolt Lite Jr” and “Megasquirt” projects were pretty solid efforts at microcontroller-based fuel injection and ignition, with variable timing and all kinds of tunability.

—Noel