NefMoto

PUBLIC SERVICE ANNOUNCEMENT:

SO Sorry I forgot to create a thread when I stumbled onto this a few years ago, and perhaps it already been discussed: Mixing Premium Pump Gas and E85 at a 50/50 ratio will provide nearly full E85 power/knock resistance. 

One really nice thing about it is that many setups don't have the pump nor injector capacity to just make the switch from gas to E85, but many have enough to try out E45.   Stock injectors and pump for instance.   550cc EV14's on a K04 car as well.

For tuning, just:

• 1. Correct KRKTE for an approximate stoich AFR of 12:1 (varies slightly due to unknown % of ethanol in pump gasoline).
• 2. Drive around and verify fueling; correct if neccessary.
• 3. Take some runs and start bumping up your timing on boost until you start seeing more than 1 or 2 knock events per cylinder or you reach ZFZWOP timing.   Also check your logs for signs of AFR leaning out and injector duty cycle getting to 100%.
• 4. Laugh OUT LOUD and think, 'Good lord, I can't believe I thought all those E85 fanatics were full of sheit.'

For stock fuel pressure:
Stock injectors run KRKTE ~ .11
Bosch 550cc injectors run KRKTE ~ .063 (if you were running ~ .054 for gasoline)

You should calculate this for yourself of course.

Timing on my current ride:

Anyone have opinions on this?

The correct way to do E85 is ethanol sensor and blending.
You need to duplicate KFZW/KFZW2 blending the same way as it works on cars with adjustable exhaust cams.

You do linearization of the E85 sensor, and then you do a 1:1 blend on KRKTE, and a custom lookup table for blend factor for timing.
Basically about 50% E85 gets you already to MBT, so you are running the same timing past that.

Without E85 sensor, the timing is usually not a problem, but with a dumb map switch you are gonna have a lot of cold start and fueling issues...

Relying on knock triggered correction for tuning on high octane fuels is a dangerous game in my experience. Stock Kfzwop should keep things relatively safe I guess.

Interesting, would this be viable when used as an extra safety margin when on track, as apposed to adding power?

For example, im running Stage 2 on a BAM 1.8T, I could alter KRKTE as in the first post, and mix my 99 octane with ethanol 50/50 without altering anything else in the map?
This would give me huge knock prevention?

That's exactly what I plan to do on an ME7.9.10, the only thing I have not yet figured out is sensor linearization. I have both a linear ADC channel and a bias-resisted for temperature input one available on the ECU, but do not fully understand the principles of E85 sensor operation (yet). I could of course use something like E85 Zeitronix module, but really want to avoid extra equipment and wiring.

After some education in flex fuel sensors I see I should be rather looking for a free 0/1 input pin on the ECU with a free interrupt. I briefly did and it seems most of the obvious candidates (like the unused VSS input) are not soldered up on this ECU, so that looks like an external module to me, either fancy like Zeitronix, or home made based on Arduino mini or something similar. Not difficult, but needs to be done elegantly and cleanly. 

The sensor outputs a 5V PWM signal.
The ethanol content is represented by the frequency and the fuel temperature by the length of the on pulse.

I think an EGT input can be used for it, or rear o2 with a man-in-the-middle module.

Right, so the E% changes the frequency, the temperature the fill. For that reason I perceive feeding this signal directly to ADC as unreliable / imprecise, varying temperature at a constant E content will vary the output. Counting and measuring the edges should be better. That's the theory, the practice is something to be determined, and since I know you do not like academic discussions, I'll shut up and go cook up E85 sensor emulator on my Arduino, feed it to the ECU and see what ADC readouts I get. And promise to get back.


If ADC is to be used, this part I have sorted out already.

Now I have found the other flex fuel topic and I see what you are on about, will need to check the EGT connections on the ECU then.

The EGT connection is ready to be used on the ECU, just checked. But it is connected to ADC, the same way as an unused oil temperature sensor input, both with a 1K bias resistor, only the one for EGT looks more "serious". And there is a linear 0-5V input for a G-sensor (at least the Fiat docs say so). So plenty of stuff to use here.

But I think I found a better and more proper way to go. There is an unused and connected pin marked as "oil condition sensor" in the docs. On the ECU it is connected first to +5V through a pull-up of 3.1K, and then in series with 100K to the ECU pin. The ECU pin has a free CC interrupt available. Or so I think, the meter probe was slipping on me all the time, could not count the pins correctly, had a long day in general, will recheck tomorrow. Either this, or it has a neighbouring interrupt procedure that is already in place and seems like it is doing the right thing. Will re-check, seems like a good time to refresh interrupt programming on ST10, did that once a while back, forgot most of the stuff.

EGT on VAG has separate modules which output PWM.

For the ethanol sensor:

50hz = 0%
150hz = 100%

The pulse width can be 1 to 5ms.

1ms = -40C
5ms = 125C

So you can't just use an analog resistor/capacitor combo, because you're not measuring duty cycle (meaningless), but have to count edges, as the pulse length varies.

I'd set up a basic counter that increments every time it receives a rising edge on a pin.
Then if you want to update the ADC every 100ms with 1000ms latency, you can first observe for 1 second, then push the data to a stack (10 times), then every 100ms push a new value onto the stack, and pop off a value off the bottom of the stack. Subtract the values and write the result to the DAC.
Obviously check if the counter has rolled over. Or if you are lazy, just check every 1 second and reset the counter. Then the data is updated every 1 second only.

If you don't want the 1sec lag you need to write a smarter algorithm, but I think for this application it is unneccessarily complicated, and the 1 sec lag is not important.

I am not entirely sure what connection setup you have in mind describing this, especially the ADC/DAC part. I can measure the pulse width and period depending on the edge with a proper input capture interrupt handler. The only thing to do is to make sure I de-bounce the signal, just in case, but at these expected frequencies this is not a problem. And this gives me immediate reading with every egde (2 to be exact), to stabilise things I can indeed average a handful of readings. Can measure the temperature while at it - might come in handy, still have to research what consequences on cold starts E85 has. The only real issue is error handling, discovery and recovery. I do not want to run lean while on ethanol and the sensor is reading gasoline or is not reading at all.

Or least that's what my first attempt is going to be, will test it on a bench anyhow before I even cut the fuel lines to plug in the sensor, it will be a while...

So, first experiments are successful with Arduino emulated signal (apologies for poor focus):

https://www.youtube.com/watch?v=fC0STSnWaOQ

Syncing the signal after loosing it or late coming after ECU boot is still lacking a bit, do not yet see where the problem exactly is.

Debugging manually written assembly code that controls interrupts on a closed system is a bitch of a job, nevertheless, it works and now its downhill, need to now make a careful selection of maps to be controlled by this.

OK, so syncing after signal loss actually works flawlessly, it is just the damn Arduino MEGA that decides to sit on thinking how to boot for several seconds when USB-connected to PC. Still, more tests and proper signal error handling is still needed.

I've been running e45 mix in my mk4 Golf 30r setup for the last year gen2 1000ccinj, stock intank, 044 inside an IE surge inline. No ill results here other than the losing 10-15mpg. Running close to MBT timing with no timing pull here!

Hate to bump a old thread but I came across something very interesting recently. So my findings were the following the flex fuel sensor the one I tested at least from continental  technically only checks conductivity of the fuel adding water to the fuel will fool the sensor in to thinking it has more ethanol than is actually has. I have tested this on the bench with regular water and got a reading of E100 when it was just water. So unless you buying your e85 in  drums always make sure to take a sample and watch is separate.

Definitely going to test this. Thanks for the heads up!

Please do and report with findings. I would assume when buying e85 at the pump with tank below ground especially in the summer your gonna have more moisture in the fuel so technically I feel like it will be off more in the warm weather months rather than cold, but then again it depends on how much moisture is in the fuel tank below ground to begin with. I am curious if anyone else comes up with the same results. Also what flexfuel sensors are you guys using??

There was test posted years ago
https://youtu.be/XEPJo1FN32k

Precisely, old news. As for the sensor, effectively there is only one that is useful / reliable enough, the one from GM. (It is actually more than just a sensor, I think it has some meaningful circuitry in there).