Hondata

Hello,

I have a car with an ethanol content sensor installed inline on the car's return line.

Attached are a few logs of the car at WOT, and as you can see, under higher load, the sensor briefly reads lower ethanol content values, then comes back up.

The engine of topic is fed by twin 340lph fuel pumps, an 8AN feed line, and 6AN return line and ID2000's. The ethanol content sensor is 6 inches away from the outlet of the regulator. I cannot imagine it is having fuel supply issues.

The sensor is grounded to the K-Pro board's "GND" wire on the same breakout connector the ethanol input is connected to. Should this be grounded directly to chassis instead? Or are we seeing the regulator close off flow to the sensor?

Is the car better off installing a "tee" or "y" fitting on the feed for ethanol content sensor readings instead of the return?

Thank you!

Hey Derek,

This kind of behavior isn't unusual, I see the same thing on both of our flex fuel cars (have a couple on different engine management systems), a couple percent is pretty normal. Even on my Infinity stand alone car I see a 2% or so variance and the sensor is being run directly by the ECU.

For example here's an Infinity pull. I don't believe you have anything to worry about and your install is good.

Fair enough.

More or less, I am interested in knowing if the sensor should be grounded separately or directly to the board.

The sensor's reading in K-Pro differs from the Kent Moore J44175 Fuel Composition Tester I have in-house. The fuel sample from the car tests at a rock-solid 135Hz output (85%), whereas the car's sensor is logging at 77-79%. I realize the difference is slight, and can be accounted for in the calibration when making compensations, but should the frequency output not match within maybe 1-3Hz?

Side-note: Vit, is that a DynaPack test log? It sure looks like one!

It is a pulse width / frequency input, so it either works or doesn't work. So your ground is ok.
I have noticed that the sensor needs a reasonable flow through it to give an accurate signal. On my car it is only 1% while running, but if the fuel pump stops with the engine off the reading drops to about 25%-40% after a second or two.
If you're getting reading changes whilst running then I can increase the filtering / smoothing.

D-Rob wrote:Fair enough.

More or less, I am interested in knowing if the sensor should be grounded separately or directly to the board.

The sensor's reading in K-Pro differs from the Kent Moore J44175 Fuel Composition Tester I have in-house. The fuel sample from the car tests at a rock-solid 135Hz output (85%), whereas the car's sensor is logging at 77-79%. I realize the difference is slight, and can be accounted for in the calibration when making compensations, but should the frequency output not match within maybe 1-3Hz?

I've seen the same thing, manual tests get me 85% on my E85, the sensor is ~82%. It's not a big deal and doesn't really have any effect on the tune. The fact I had a couple cars exhibit this behavior basically put my mind at ease.

D-Rob wrote:Side-note: Vit, is that a DynaPack test log? It sure looks like one!

It is, we use a Dynapack 3000 at the shop. Really need to upgrade it to the 4k pods, my cars max out the 3k pods and I'm doing 160mph+ pulls in high gear now, lol.

Hondata wrote:It is a pulse width / frequency input, so it either works or doesn't work. So your ground is ok.
I have noticed that the sensor needs a reasonable flow through it to give an accurate signal. On my car it is only 1% while running, but if the fuel pump stops with the engine off the reading drops to about 25%-40% after a second or two.
If you're getting reading changes whilst running then I can increase the filtering / smoothing.

That's odd -- at idle it's always been spot on for me o_O

Manual test? Mixing in water and determining the separation point?

Yes, Summit even carries a marked test tube to make it easy. Always get 85-88% or so. GM flex sensor is around 82% and drops a couple percent under throttle in my FRS and S2000

The flex fuel sensors (all made by continental) quote an accuracy of +-5%.

I do feel the filtering can smooth it over a bit, and make the signal more consistent.

Can we perhaps have access to the filtering of the signal?

Hondata wrote:The flex fuel sensors (all made by continental) quote an accuracy of +-5%.

It makes sense. That is nominal. I've run E98 in cars with the flex sensor and it would only spit out 94-95%, heh.

D-Rob wrote:I do feel the filtering can smooth it over a bit, and make the signal more consistent.

Can we perhaps have access to the filtering of the signal?

What would be cool is just to use the last flex content value after a certain MAP threshold is hit. One can assume the E content isn't changing in boost/WOT, and it can be a failsafe in case the signal ever cuts out (or loss of fuel in the return on high powered cars).

If the signal from the sensor is out of range then the last good value is used. The issue is that the sensor is not 100% accurate.

VitViper wrote:

D-Rob wrote:I do feel the filtering can smooth it over a bit, and make the signal more consistent.

Can we perhaps have access to the filtering of the signal?

What would be cool is just to use the last flex content value after a certain MAP threshold is hit. One can assume the E content isn't changing in boost/WOT, and it can be a failsafe in case the signal ever cuts out (or loss of fuel in the return on high powered cars).

Vit, I do like the idea of defining a load and/or engine speed threshold to hold the last reading from.

Hondata, is this something, along with some filtering change, that we can perhaps implement?

Hate to be a squeaky wheel, just looking for any input regarding changing the filtering speed.

Thank you!