FJR Fuel gauge Linearisation

Linearising the FJR's Fuel Gauge

Start date 23/07/2012 15:20    Last updated 30/07/2012 16:39



We all know the fuel gauge bars are anything but linear in terms of bars showing v. quantity of fuel in the tank.

Typical gauge display
(Click on any image for larger view,
click on that for the original)

We also know that when the lowest bar starts flashing, we get a "miles in reserve" indication.

Whilst this is fine if you know there's always another fuel station just round the corner, my low fuel paranoia isn't comfortable, particularly when I'm in an area where fuel stations are scarce, or they're likely to be shut.

Also, my riding style varies considerably from three-mile city stop-start journeys, having fun in country lanes, or eating miles on the motorway. These all give very different ranges.

I've also noted that on the gauge, when it gets down to one bar, there's only about nine or ten miles before it starts flashing, so used for only about 1/25th of the tank's capacity.

As far as I'm concerned, this is a poor use of the gauge bars.

What I would like to do is alter how the bars are used in terms of fuel left, either more linear indication, or possibly compressing the top end to expand the lower.

Linearisation seems the more intuitive.

So my desire would be to split the gauge into its nine possible indications (8 bars down to 1 bar and 1 bar flashing) with equal quantities of fuel for each. I am assuming the sender can move sufficiently at the highest and lowest ends of the range.

To do this, I need to intercept the feed between sender and gauge, measure the sender resistance, and force an appropriate voltage onto the gauge.

Test of gauge bars

So I need to know a little more about the sender and the gauge.

The gauge is the easy part.

I lifted the tank, disconnected the sender connector, pushed wires into each of the connections going to the gauge. I could then connect variable resistor simulating the sender, and measure the variable resistor ohms, the voltage, and observe the bars on the gauge.

I did this and got the following results. The graph shows the results,the red line is calculated assuming a 220 ohm feed from an 11.2 voltage source.

Sender voltage & Bars v. Sender resistance

OhmsVoltageBars (0=flash)

Putting the data ready for driving the display, we plot bars against voltage:

Bars v. Sender voltage

Sender characteristic

The next test is to find the sender resistance for varying fuel level.

Detail of wires into connector
Close-up of wires into connector
Wires and connector in place
Close-up of wires into connector
Taped round connector
and to original sender wire
Wire tied to 'T' as strain relief
Detail of connections to wires
Meter display

To facilitate this, I've put a single strand from a piece of flex into each connector and carefully put the connector back in place. A quick resistance measurement shows they have connected. These I've soldered to a pair of wires to bring them out from under the tank so that I can hook my meter to them. I've taped this lead to the fairing so I have easy access to it on the forecourt.

The test would ideally be to start with a completely empty tank, then add fuel litre by litre (or quart by quart if you're in America) and measure the resistance at each value.

As it is unlikely I could manage to run out of fuel just as I reach a forecourt, I did a slightly different test.

Knowing I'd be doing a lot of local trips, what I did was to measure the fuel gauge sender resistance at each stop, this way I knew when it had reached its end stop. I ran it a bit further to reduce the amount of fuel in the tank, until the "low fuel" trip said about 30 miles.

Measuring jug in position
Measuring a "fill"
Resistance v. Litres

Now I raised the tank, disconnected the outlet pipe from the fuel rail, and lowered the tank back to its normal position. I was able to hold a container (my mum's kitchen measuring jug) beneath the free end of the outlet pipe.

Turning on the ignition starts the fuel pump. I'd expected the pump to continue pumping because there was obviously no pressure in the fuel rail, but the pump stopped after the usual couple of seconds. So I continued by cycling the ignition key.

When the container was reasonably full I stopped, measured the amount in it, poured that into a container, and repeated until no more fuel came out.

Having measured this original fuel, I put the fuel pipe back where it belongs, then put the drained fuel back into the tank. I then put in more fuel from a can, measuring quantity and sender resistance as I went, so that the sender resistance was well off its end stop

Then a short trip to the nearest petrol forecourt where I fill from the pump, stopping as each litre went in, writing down the quantity and the resistance.

Because of this three miles or so trip, my measurements would be a little out, so I "frigged" the results slightly. This won't make a lot of difference either way, but it does make my graph smoother!

Note this "frig" has nothing to do with the strange shape at the 5 litre mark, that part was exactly as measured. It could be that the sender stuck a little, but there is a data point part way up that very steep bit.

fuel Oz (Imperial)LitresTotal(litres)ResistanceVoltage if 13/220

The original gauge changes the single bar from solid to flashing at about 125 ohms. This corresponds to a fuel quantity of between 5.5 and 6 litres.

With my average fuel consumption of about 48 mpg or 10.5 miles per litre, this means an "blinking" range of 57 to 63 miles. With my worst fuel consumption of around 35 mpg, or 7.7 mpl, my "blinking" range is around 42 to 46 miles.

The first disappointment for me is that at the low level, there is no significant change in the sender's resistance below 5 litres.

This means my first wish for a better resolution at low fuel levels is not going to be as good as I'd hoped. The lowest sensible indication would be with a sender resistance of (say) 135 ohms, corresponding to about 4.7 litres, little better than the presesnt 5.5.