TBI Fuel Injection for VW Air-Cooled Engine 

Great tip, thank you!

I would suggest moving the injectors beyond the butterfly valves they may be prone to ice

Great point, thank you!

Yeap! More easy way to do thats holes.

Another suggestion is slower cutting speeds..i found those large hole saws work best at slower speeds..too fast and it skips over instead of digging in and its actually faster.. this is known in the lathe world as feeds and speeds..their is a sweet spot.and the larger the tool the slower you go as the diameter changes surface speeds.and drill presses tend to drill smaller holes and are set to go fast and that compounds it with a larger cutting tool..SO the faster you go the more wear on the tool as well... carbide can go faster but tool steel needs a slower range of speeds and the right feeds.

That's great to know! I do recall seeing the fuel fog at high RPM on the old carburetor, although it never caused too much trouble. I'm also running an air hose over to a separate airbox, so that should keep the fog (if any occurs) contained and away from ignition sources.

Great idea! Believe it or not, my newer ECU design actually controls spark timing via 4 HEI coils and a pair of optical pickups mounted inside the 009 distributor. This means I can actually add in vacuum advance via the MAP sensor signal if desired, and apply more or less any timing adjustments I want in software! With the latest ECU and programming, the buggy runs extremely smoothly now!

The homebrewn fuel injection doesnt have an accelerator pump. Thats why it stumbles - it has no way of handling transients, but only steady state. You need to look at mapDOT (Delta over time) or preferably tpsDOT to add in an "accelerator pump" function.

Thank you! Right now it's open-loop only, but my latest board has additional analog inputs that could be used with an O2 sensor for closed-loop control!

I believe as long as the total on-time of the fuel injectors matches the required fuel delivery, it should work. Since you are designing a synchronous system, the pulse width or pulse count should follow the MAP measurement. The more manifold pressure there is, the more on-time the injector should have during each rotation of the crank sensor.

Thank you! I got the injector caps on eBay here: www.ebay.com/itm/113711584424 I've been driving it every week for 30-50 miles, and it's been running extremely well! It easily does 75+ MPH on the highway and stays cool, even in the 100+ degree heat in Phoenix. I also passed emissions with it a month ago! I currently am using a newer version of the PCB that drives 4 individual HEI coils in wasted-spark configuration, and it runs brilliantly (better even than some modern EFI cars). The fuel injection control is identical, although the workload is split across 2 separate Arduinos to allow for more CPU time to run the 4 coils. I'll be making a video on the new board soon!

I believe 2 injectors should be adequate, as these are 195 gram/min high-flow injectors. 1 injector was almost enough for the 1200cc engine, although it would peg to 100% duty on a super hard pull at 4500 RPM. So, for the 1600 I believe 2 should be plenty. That's an awesome idea to run 2 separate throttle bodies! I think if I decide to use throttle position sensing in the future, I will probably use sensored single-bore throttle bodies, one per side as you say. Cheers!

@@DielectricVideos thanks for your reply! Looking forward to see the future of this. What you’ve achieved so far is very impressive - kudos to you for pioneering your own system almost from scratch and taking us along for the ride.

@@DielectricVideos hi one question what model o type the injectores and what is cae o vehicle?

Hi, what kind of info do you need. I had a Mexican bettle 1994 with the efi engine also a turbo efi engine running with megasquirt. Any questions let me know. To see if I can help.

It should help, as the more mixture you can get in at full load, the more power you can make. Also keep in mind that the manifold runners and heads will add restriction, so top-manifold pressure may not be exactly the same as cylinder pressure.

Great idea! I'll try that next time I have to weld something galvanized.

The A/F ratio problem with aircooled VW engine occurs due to the firing order. Manifold icing will still be an issue. Fuel mixture balance left to right bank will be an issue on this design. And as always, the exhaust system shown in video robs HP &Torque....

Some things to check: 1. What angle are your injectors entering the manifold? If the angle is too shallow (too close to perpendicular to the manifold), fuel can pool up as it hits the side of the manifold. 2. What fuel pressure are you using? If it's too low, the fuel might not be atomizing well. 3. Is your injector flow rate set correctly in the tuning software? Remember that the flow rate will be higher than specified if you are running a higher fuel pressure than the injectors are specified at. Also be sure your sensors are all connected and calibrated correctly. If it's a speed-density control system and the MAP sensor is left disconnected, it may run excessively rich due to the ECU thinking the intake is always at atmospheric pressure. To some extent, fuel pooling is an inevitable problem for TBI systems (which is why pretty much all modern cars use port injection or direct injection now). Even in my VW, it goes rich for a few seconds if I suddenly decrease throttle position after a long pull or cruise. But in general, it should be possible to get good mixture quality in most driving conditions with a TBI system.

I actually designed my own HEI system just recently, and it made a huge improvement in power. I'm hoping to do a video on it soon!

Waiting for more of your efi videos. 😍

I did some testing with one and may incorporate it in the future. It would provide a gradual adjustment to AFR target to compensate for drift in the injector performance, sensor performance, etc.

Great idea! If I ever build a 2nd version, I will definitely make it smoother. So far this one has performed very well, as I have put over 3100 miles on it!

The only value that needs to be changed when selecting your injector configuration is the injector maximum flow rate, "injector_MaxGramsPerMinute". To compute the correct value, simply add up all the flow rates of the various injectors (in grams per minute). Because all the injectors are simultaneously switched open and closed, the ECM only needs to know the total combined injector flow rate.

It makes perfect sense, I'll be uploading content with my development, I'll show my progress

It does, there is actually a bypass port on the butterfly throttle body that goes through to the vacuum takeoff

The ECU isn't actually aware of the crankshaft angle, it just counts pulses from the original breaker points to calculate RPM. In this system, fuel is dispensed at a fixed frequency, and is asynchronous to the engine rotation. Only the duty cycle of the fuel delivery is changed by the ECU, dependent on measured manifold pressure and RPM.

thank​ you

If you mean a MAF (mass air flow) sensor, then it is possible, but in my experience it isn't reliable. Long ago I tried a MAF-only EFI system, but the MAF sensor drifted considerably with vibration of the engine and bumps in the road. MAF sensors are better as long-term average sensors for correcting error in other measurements, but are not good for fast-response measurements. An alternative to a MAP sensor could be a throttle position sensor. The mass airflow could be approximated by multiplying the throttle position by the engine RPM, and then by applying a compensation factor to account for non-idealities in the throttle position. But you would want to carefully tune a lookup table or AFR table to give the correct mixture across a wide range of throttle positions and engine speeds, as the two are not likely to be perfectly linear to the mass airflow.

I designed a custom ECU based on an Arduino Nano. You can see the approach I took in another video: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-kj24iWmM_PU.html

The throttle body that mounts below the barrels has a vacuum port that connects downstream of the throttle valves. The small threaded elbow fitting connects to this, and the MAP sensor can be connected with a hose to this fitting. It's important to have the MAP sensor connected on the engine-side of the throttle body in order to accurately measure the air mass flowing into the engine.

lol

I'd recommend checking out the video on my open source engine control system: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-3S8LlHKaa_o.html

For the injector bungs, I used these steel ones from FRSport: frsport.com/products/atp-uncategorized-atp-uncategorized-atp-ful-misc2-st I used injector barb adapters from eBay with hardware store bolts to attach them to the bungs. The fit is not perfect, but as long as the adapters are securely mounted over the injectors, they shouldn't leak: www.ebay.com/itm/113711584424 For the injectors, I initially used some lower-capacity ones from Amazon, but switched to these high-flow 440 cc/min to ensure that I would never hit 100% duty cycle and lean out the mixture: www.ebay.com/itm/223283676528

@@DielectricVideos what do you estimate your duty cycle for e85 would be, do you think that CC would be enough, just curious

​@@Projects_and_Pistons For straight gasoline, I calculated that my current 1776cc engine running at redline RPM (4500 RPM) and WOT (atmospheric MAP) will consume about 429 g/min, or about 611 cc/min when running at a worst-case rich AFR of 11.0. At the design pressure of the fuel injectors (43 PSI), each one nominally flows 440 cc/min, so the overall duty cycle would be about 69%. E85 has a stoichiometric AFR of about 9.8, which is about 66% of the stoichiometric AFR of gasoline. So, for the same worst-case rich redline WOT condition in E85, you would theoretically need 1/(0.66) = 1.51 times the duty cycle as you would for gasoline, which would come out to an impossible 105% for these injectors. I actually run my fuel rail at about 58 PSI using this off-the-shelf regulator/filter ( smile.amazon.com/gp/product/B07W9H5TF9 ), not 43 PSI. So, in actuality I can theoretically get up to 593 cc/min out of each injector, so on E85 I could theoretically meet demand at about 77% duty cycle. At these flow rates, the fuel pump would likely be the greater bottleneck in maximum flow, rather than the injectors. I am almost certain the $25 fuel pump I'm using would not be able to keep up if I actually commanded this much fuel demand ( www.amazon.com/dp/B07FZ5YKXT ). Right now I typically run either 87 or 89 pump gas, rather than E85 (I probably would run E85 if it was easier to get in my area). I rarely run the engine to its absolute maxima (with the occasional exception during freeway driving). When I have occasionally pushed 4000+ RPM at WOT, it's run well enough for my liking. Really smooth on 89, and the slightest hint of knocking on 87. I haven't installed a lambda meter yet, so I can't say for certain if the engine is leaning out or not. No complaints about the system performance over the last 1800 miles, but as with any homemade system, YMMV.

@@DielectricVideos thanks for the detailed reply, I am planning on going efi so I can use e85 and reduce cylinder temps on long cruises

Ok I’ve looked everywhere and I can’t seem to find that throttle body you used, it looks used the only one I found like it is a holly throttle body for a 2 barrel. What did you end up using?

I'm running a homemade ECU that uses an Arduino, a slightly modified version of the one I covered in an earlier video: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-kj24iWmM_PU.html For the pump, I'm using this one: www.amazon.com/DWVO-Mustang-Thunderbird-Mercury-Continental/dp/B07FZ5YKXT And for the pressure regulator, I'm using this combined filter/regulator: www.amazon.com/gp/product/B07W9H5TF9

It still needs to determine engine speed from somewhere. It can use the signal from the distributor or other ignition system source to get this information. Crankshaft position is not required for it to operate, as it is an asynchronous injection system.

@Renato Passos It doesn't need the crankshaft signal, only points signal from the distributor. This signal can be obtained form existing spark ignition points, or it can be obtained using an optical or Hall effect sensor.

@Renato Passos You could use a MAF (Mass Air Flow) sensor as a true single-sensor EFI control input. They are prone to accuracy drift due to contamination, but are commonly used on modern engines for primary fuel control. If desired, you could try implementing a MAF-only system to see if you can get it to work well. Speed-density (RPM + MAP) systems are generally more robust due to the physical separation between the intake manifold and the MAP sensor, and also allow AFR (air fuel ratio) to be tuned over both engine RPM and manifold pressure. It would not be possible to measure mass air flow using Manifold Absolute Pressure (MAP) alone, however, as the same manifold pressure can exist at different engine speeds (and higher engine speeds will draw in more air, for the same manifold pressure). I suspect an engine could be made to run with only a MAP sensor, but it would be nowhere close to stoichiometric except at one particular RPM value.

Here's a great resource talking about both MAF and speed-density control strategies: www.mishimoto.com/engineering/2021/05/maf-vs-speed-density/

I'm driving them as simultaneous asynchronous injection under speed-density control. If you build a similar manifold, you can actually control it however you want, depending on how you configure your ECU.

@@DielectricVideos is that in your provided arduino code?

@@multimargejta123 Yes, it is. The currently available code operates the injectors at a fixed frequency, and the duty cycle is adjusted based on points-triggered RPM measurement and MAP sensor. The engine displacement, number of cylinders, and injector maximum flow rate are given as user-configurable constants in the code.

It's very similar. I am working on adding an AFR table to change the fuel mixture depending on load and RPM conditions.

@@DielectricVideos you have using throttle sensor code? Or can i get new code? Thanks

@@rofieasnan6085 I haven't written throttle position code yet; however, I just posted a new version of the code with AFR mapping and rate-of-change of MAP pressure, which emulates rate-of-change-of-throttle from a TPS. Here it is: github.com/JohnPattersonConsulting/DuneBuggyEFI/blob/main/MAP_EFI_V2.ino

Yes, they operate at the same time and are wired in parallel to the same MOSFET and 12V supply.

@@DielectricVideos which spec for this mosfet ? Thank you!

@@brunopmoraes I'm currently using an IRFP460 to drive the injectors. Using the same type of transistor to drive an HEI ignition coil as well!

It needs to be fuel-resistant, tolerant to high temperatures, and resistant to cracking in high vibration. Because of this, I felt that steel was a preferable choice to most of the common 3D-printed media.

🤣🤣🤣 Good work to the smart fella indeed

Maybe, depending on the injector. I originally had a pair of 273 cc/min injectors, and I could definitely get it to lean out at high RPM/high load. On the freeway, it would actually gain power when reducing from WOT to 50% throttle. I changed these out to 440 cc/min injectors and it seems to keep up all the way to WOT now. IMO it's much better to have extra headroom in the injector capacity than to risk the mixture going lean at high load. Aircooled engines will burn up really quickly if the mixture goes lean under power.

@@DielectricVideos i use a iwm52300(off a 1000cc engine) on my 1600cc, hold 140hp on a turbo setup too

@@MrIngorodrigues Very cool, looks like the iwm52300 is a real beast of an injector. Do you know the flow rate and rated fuel pressure on that one?

Sure, they are infamous here caled "dumb EFI" so they are cheap ones lol The specs i found are 29-36ml at 1bar 42-48ml at 1,5bar

By not good for you he means can kill you long term! Don’t do it!!!!!!!!