REVELTRONICS

I have been calibrating the Fuel Level on the UTCOMP. I have got it pretty accurate now. The Fuel Tank screen shows quite accurate level information. The Fuel remaining on the Tacho screen however does show the same figure but only when the vehicle is stationary. If the vehicle is moving and cornering, the Fuel remaining figure is very unreliable and can vary instantly by more than a gallon !. The Fuel Tank display seems to handle the variations in tank level when cornering, and keeps an accurate display. I presume the Fuel Tank display must be sampling at a much slower rate to keep the display accurate. But it looks like the Fuel Remaining display on the Tacho screen must be using a much faster sampling, and consequently the figure varies a lot. Can this sampling rate be slowed down for this screen ?
Thanks.

I think that in tacho or ecu is more sophisticated algorithm than just sample averaging - you can check this if you fuel up on running engine - it will probably do not detect additional fuel or it detect only a little until you switch ignition off. There is no workaround for it in UTCOMP (I'm using fuel level from fuel consumption), you can try to add small capacitor on input so it will smooth the characteristics over time but it needs some experiments.

Ok thanks for that. I thought you would have used the same fuel tank level signal as used by the Fuel Tank screen, rather than try and calculate it from the fuel consumption. Hopefully the "Miles Remaining" on the same Tacho screen will not fluctuate as much as the Fuel Remaining.

Miles remaining depends on fuel level. Some fuel tanks, especially in older cars, does not have any partitions inside and fuel goes from from band to band on every corner so fuel level also fluctuate. As I remember in latest firmware there is averaging from 6 seconds, please make sure that you have latest fw viewtopic.php?f=21&t=569

I have the latest V3.8 software. There is no averaging available on the ADC tab for Fuel Level, only for other sensors.

Ady, I checked in source code in latest firmware and fuel level is averaged from maximum number of samples, which gives averaging from 5 seconds. It's not configurable, because it's already maximum number of samples for averaging (255).

I have built a small circuit to dampen the fuel sender voltage fluctuations before they go to the UTCOMP. It uses an op-amp configured as a unity gain high impedance buffer, and then a simple capacitor and resistor as a low pass filter to dampen the variations. There is a micro relay to discharge the capacitor rapidly when powered off. When switched on, the voltage output to the UTCOMP rises up to it's final reading in about a minute (in effect starting at tank full and slowly dropping to it's final figure). When the car is driving on the road, the Fuel Remaining and Miles remaining are considerably more stable and now are able to be used as reliable figures.

Ady, are more stable after averaging from 1 minute or are stable as rock? I'm happy that you found workaround, but I'm just curious, what averaging in this case is required.

The circuit is obviously not any form of complex digital sampling and averaging, I am not clever enough to design anything like that. It is simply a method of slowing down any rapid voltage readings from the tank sender which is the only thing I could think of to help the situation. I think the low pass filter has a cut off of about 0.1Hz, but is only a single order filter, so the roll off is pretty gradual. On the test bench, any rapid voltage variations at the input are almost completely damped, with no noticeable voltage variation at the output. Slower variations are let through but at a lower voltage variation. The output of this circuit is then fed into the UTCOMP, where you say the signal is sampled and averaged over about 5 seconds which makes it even more stable on the display. The combination of these two seem to have made the system work better. It isn't perfect and is not steady as a rock and in reality it could do with even more damping. The Fuel remaining and Miles remaining are MUCH steadier when driving. They do vary, but whereas before the Fuel remaining display could easily vary by over 1 gallon whilst driving along, it now only varies by maybe 0.1 to 0.2 gallons, and only then when a radical turn or acceleration occurs. In reality if the UTCOMP sampled and averaged over a much longer period, maybe 60 seconds or more, this extra circuit would not be required and a steadier accurate reading of the tank level would be displayed. 5 seconds of sampling and averaging does not seem to be enough for an open none baffled tank like mine.

After further testing I found that even though the display was now stable, it was in fact displaying the wrong Fuel level !. On the bench everything tested ok, but once connected to the UTCOMP, the output of my circuit was being dragged down by the input of the UTCOMP, reducing the output voltage of the circuit causing an incorrect (high) fuel level displayed. I have now used the other op-amp in the LM358 package to buffer the output of the circuit into the UTCOMP input. It is now allowing the correct damped voltage to be fed into the UTCOMP and I am now getting correct readings combined with a damped display.