Megasquirt on a K75. (DIY fuel injection system)

[wmax351]

Bringing this post back from the dead. The Formula SAE competition in lincoln, ne this year has inspired me. As well as the 33mpg i get while cruising at midwest superslab speeds to get there.

I am thinking i will use VEMS, which is similar to megasquirt, but a more solid system, that is based on a similar architecture to the mpg guages i am making, facilitating my programming it. Also has a built in Wideband O2 sensor controller.

Some of the profit from the mpg guages will go to this.

[mystic red]

33mpg? What do you get at say, 60 to 70? Should be almost 50 with a 75.

[frankenduck]

Something is amiss if you're only getting 33 mpg.

[wmax351]

Some depends on the altitude plug. With it in at 5000 feet, get 42 or so at 70.

I may try bumping the timing up some more.

Although, i am still interested in doing this as a fun project. Will vastly improve throttle response as well. Rather than having the jetronic catch up to the throttle position, the throttle position (combined with manifold vacuum) will determine the fueling, calibrated by the wideband o2 sensor.

Some of my thoughts are that the bikes contemporary, the V65 sabre, gets better mileage with an 1100 cc, 150 horsepower, carburated monster.

[mystic red]

Max, I get a solid 48 with my 1100 as long as I keep the RPMs under 5 or 6 K. Only time it dropped into the high 30's was when I was running on 3 cylinders.

[wmax351]

Well, I pulled the trigger. Ordered a Megasquirt 1 kit. I will use the Airflow sensor for accurate fueling. Tunerstudio MS has a very good autotune feature. So I will be riding around with a laptop strapped to the back of the bike.

Basically, the initial setup will be Megasquirt 1, with a wideband O2 sensor controller. Stock AFM, actual throttle position sensor (rather than a switch).

I will either tune using Alpha N (throttle position and RPM) or MAF (air flow meter and RPM, with Acceleration enrichment [basically a digital pumper carb])

[sh00ter]

Well, I pulled the trigger. Ordered a Megasquirt 1 kit. I will use the Airflow sensor for accurate fueling. Tunerstudio MS has a very good autotune feature. So I will be riding around with a laptop strapped to the back of the bike.

Basically, the initial setup will be Megasquirt 1, with a wideband O2 sensor controller. Stock AFM, actual throttle position sensor (rather than a switch).

I will either tune using Alpha N (throttle position and RPM) or MAF (air flow meter and RPM, with Acceleration enrichment [basically a digital pumper carb])

Max, super excited to hear how that project goes.  Besides just seeing if it works, what's the goal?

[wmax351]

Well, I pulled the trigger. Ordered a Megasquirt 1 kit. I will use the Airflow sensor for accurate fueling. Tunerstudio MS has a very good autotune feature. So I will be riding around with a laptop strapped to the back of the bike.

Basically, the initial setup will be Megasquirt 1, with a wideband O2 sensor controller. Stock AFM, actual throttle position sensor (rather than a switch).

I will either tune using Alpha N (throttle position and RPM) or MAF (air flow meter and RPM, with Acceleration enrichment [basically a digital pumper carb])

Max, super excited to hear how that project goes.  Besides just seeing if it works, what's the goal?

Some of the goal is that its interesting. But I will be able to improve power somewhat, improve throttle response greatly, and significantly improve fuel economy. The old ljet doesnt respond to changes, and runs super rich at high rpm. I will be able to control fuel mixxture accurately, with closed loop feedback. keep it at 15-16:1 while cruising, 14.7 in the middle, and 13.5 at wot.

[wmax351]

Got most of the computer and related items set up. O2 sensor bung welded into the exhaust, and all the things needed to wire it up.

I am going to start out using the control method that the K1100's use, Alpha-N. This compares throttle opening, RPM, air temperature/pressure, and coolant temperature. Then it trims the fuel delivery with feedback from the oxygen sensor.

[rbm]

Max,

What will you be using for your TPS?  The K75 uses just a switch whereas the K1100 unit outputs a signal proportional to the throttle opening.  Are you adapting the K1100 TPS to your K75 throttle bodies ?  I was toying once with the idea of fitting K1100 TBs to the K75 engine; never got past the contemplation stage though I think it could work.  In your case, if you fitted the K1100 TBs, you'd get both larger throat opening and proportional TPS for your Megasquirt.

[wmax351]

I just used a cheap proportional tps. Pretty much universal. Paired with a quick and easy adapter (prototype as of now).

[wmax351]

Almost ready to install the Megasquirt unit.

I have a proportional TPS, and I have modified the circuit to provide inverse log response (for more accuracy) at the low end of the throttle.

All that is needed is to cut one wire on the air flow meter, and install.

I used a cheap volvo 740 L-Jetronic computer for its connector. I made a little circuit board with connectors for other circuits and sensors, as well as the main L-jet connector. This will all go into the old enclosure for the volvo ecu.

I have a wideband O2 sensor controller in a separate box, which feeds a 0-5 volt signal for the Air Fuel ratio (from 9:1 to 19:1 with gasoline).

[Rick G]

Max,

What will you be using for your TPS?  The K75 uses just a switch whereas the K1100 unit outputs a signal proportional to the throttle opening.  Are you adapting the K1100 TPS to your K75 throttle bodies ?  I was toying once with the idea of fitting K1100 TBs to the K75 engine; never got past the contemplation stage though I think it could work.  In your case, if you fitted the K1100 TBs, you'd get both larger throat opening and proportional TPS for your Megasquirt.

Hi Rob
TWB on the K100 forum put the K1100 TBs on his K100RS and it worked fine but he does not have a TPS working because the mounting is different .  It goes well without the TPS.

Max How are you getting round the intergration of the megasquirt ECU and the K75 ignition and revlimiter?

[wmax351]

Max,

What will you be using for your TPS?  The K75 uses just a switch whereas the K1100 unit outputs a signal proportional to the throttle opening.  Are you adapting the K1100 TPS to your K75 throttle bodies ?  I was toying once with the idea of fitting K1100 TBs to the K75 engine; never got past the contemplation stage though I think it could work.  In your case, if you fitted the K1100 TBs, you'd get both larger throat opening and proportional TPS for your Megasquirt.

Hi Rob
TWB on the K100 forum put the K1100 TBs on his K100RS and it worked fine but he does not have a TPS working because the mounting is different .  It goes well without the TPS.

Max How are you getting round the intergration of the megasquirt ECU and the K75 ignition and revlimiter?

The K100s also use a Air Flow Meter like the k75, which doesn't need the tps.

The K75 has the bosch ignition controller, which handles revlimits and ignition. It outputs a tach signal to the L-jet controller, which I am using with the Megasquirt.

I will enable a revlimiter on the megasquirt as a secondary safety.

[wmax351]

The megasquirt and j-jet adapter board. Mounted on the stock ecu tray. The wideband o2 controller is in the box on the right.

[wmax351]

Project is held up for a few days waiting for a $0.28 crystal. Those little buggers are delicate when installing. This time, I will glue it in place before soldering it.

[wmax351]

AND IT RUNS!!!!

Started right up. Needed to tweak a couple of settings, then it started on the first crank. Now to tune it.


I've been able to get a good idle vacuum signal (impressive with a Individual Throttle Body engine) at around 50kpa. This should make tuning much easier.

[wmax351]

Hooked it up to a laptop, and drove around today. Got a functional tune working so far. Great (far better than stock) response and power at lower rpms, and equal or better at higher. Its easier to do the low rpm tuning around town and on the twisties above berkeley, so understandably this part is tuned better. I need some more open straight road to dial in the higher rpms and higher loads. With the lower gears, it shoots through the rpms too quick to get a good reading. So for now, I have the high rpm set rich to keep the engine safe, and then I will let the tuning software (Tunerstudio MS) dial it back.

Interestingly, it starts far easier now. Even without priming the fuel pump (which the megasquirt usually does). Touch the starter switch, and it fires up even more quickly than stock.

I have it set up taking the ignition signal from the middle coil, which is the same cylinder as I am taking the vacuum (load) signal from. I have it configured in the software as a 2 cylinder, with a single coil. Basically, that is to give the correct reading for the crank position, and to allow me to split the fuel injection into two events per cycle (720 degrees) for smoother idle. I would do three times per cycle, but the resolution of the fuel injector decreases as the pulsewidth decreases. At idle, I am at about 3.5 milliseconds, divided into two 1.75 ms squirts. As long as the pulswidth is around 2 ms, it is usually fine.

Will keep working on it and update here. Looks like this will be fairly useful. Thus far I have left the stock fuel injection wiring harness on the bike, and tapped into it for some connections. But based on the results of todays testing, I am thinking I will want to make this a permanent install, with a custom, high quality wiring harness, perhaps this winter.

[wmax351]

Tuning setup.

[rbm]

Fantastic news WMax. :2thumbup: :2thumbup:

Will you consider writing up your project for others to learn from when the system matures and becomes more "beta"? :)

[frankenduck]

Interesting project to follow you have going on here.  Can't wait until it is finalized.

[wmax351]

Fantastic news WMax. :2thumbup: :2thumbup:

Will you consider writing up your project for others to learn from when the system matures and becomes more "beta"? :)

I will indeed. Most of it is very well documented in general (see megamanual.com and MSextra.com), but there are a few tricks I had to use.

Basically what is needed is to tap the Water temperature sensor wire that goes to pin 10 of the L-Jet connector, make an adapter plate for a volvo throttle position sensor, add a temperature sensor to the intake (I screwed a GM coolant sensor into the intake snorkel. If it was turbo or supercharged, this would be more critical, need to be in the plenum, and would need a faster response), weld a oxygen sensor bung to the header collector box, make a harness for the injectors (which taps from the fuel pump supply wire, and supplies the wideband oxygen sensor controller), and then finally tap the #2 coil (-) and the #2 cylinder's vacuum port (with a vacuum line to the built in air pressure sensor). A pre-assembled wideband oxygen sensor controller is attached as well. All of this is hooked into the controller box.

A laptop is used for tuning, most of which occurs with an automatic algorithm with Tunerstudio software(http://www.tunerstudio.com/). The base map had it running well enough to get going after a bit of fidgeting with the idle and off idle mixture.


From here, I will continue tuning the fuel maps, and then add acceleration enrichment (like a pumper carb/accelerator pump). I will also add exhaust gas oxygen control, which updates the mixture on the fly.

So far it has better response in the lower rpms than it ever has had. Huge increase in torque and smoothness. I haven't gotten the higher rpm as well tuned yet. I have it set super rich for the moment as a safety precaution, and it has pretty good power at full throttle. 

In general, I have been starting with the autotune function set for around 13.5:1 air fuel ratio, richer at higher loads, leaner down with lower loads. From there, I will lean it out for better mpg and emissions at cruise and other speeds. Will aim for a stochiometric (14.7:1) ratio at cruise, 13:1 at wide open throttle for peak power and torque.

I am considering at some point implementing a full fuel and spark control system, and potentially sequential injection. I would need to make a shutter wheel to replace the timing mechanism in the bike, as the 2-hall system will not work for the megasquirt, as far as I can tell. I may be able to make a microprocessor controller which will take the output from the hall sensors and convert it to something for the megasquirt. The stock bmw system is pretty ingenious. For the K75, it can tell which cylinder of 3 to fire (with wasted spark) by which of 2 sensors are active. The wheel has 2 slots 120 degrees from each other. With arbitrary cylinder numbering:

Cylinder	hall 1	hall2
1	On	off
2	On	on
3	off	on

Will keep posting as I update it. I'll happily share the maps with anyone who is interested in doing this.

[wmax351]

Well, did about 50 miles on it on friday. I changed the injection cycle to once per engine revolution. Much smoother response and idle, and better off idle performance, even without acceleration enrichment.

Was having some trouble tuning at higher speeds/rpm due to vibration/bumps causing read errors on my beater laptop. This will be resolved with a flash drive with linux running off of it, not using the hard drive.

Also had some issues with jiggling cables breaking the DB9 serial port on the megasquirt. I have made a new cable, which is a much lower profile, which prevents there from being leverage to damage the port. I have also added a 5 dollar bluetooth module. This allows me to use my phone (droid) as a datalogger and dashboard, and to change settings on the road, if I desire.

It is running great. Much more fun to ride: better performance down low in the rpms, and stays good throughout. I may be taking a trip next weekend, which will be a great trial run and tuning opportunity.


I attached the tune file, if anyone is interested in looking at it in tunerstudio. Its a free download at www.tunerstudio.com and you don't need to register.  Its using MS2/extra version 3.24

[wmax351]

Well, more work has been done. The system is working well thus far, so I removed the old EFI harness and Air flow meter. Removing the AFM should give a bit better flow at peak torque (6000 rpm) where a lower manifold pressure at WOT suggests some restriction.

 

[wmax351]

Well, more tuning completed. Its running well. However, I am hitting the maximum Duty Cycle for the Injectors at WOT in several spots. There are a few things I can do here. This basically means I am trying to inject more fuel than the injectors allow. I may switch the injectors to fire once every 720 degrees rather than every 360. I can also increase the fuel pressure, which I will probably do.