July 2013
I have a 40 year old Chevy. The fuel gauge bounces half the gauge reading when it gets under a 3/4 tank. I put in a new sending unit but that did not help. I tried a digital gauge and that helped some, but not enough. It is a 90 ohm sending unit, zero ohms (grounded) = empty and 90 ohms = full. I would like to find a circuit that will filter out the high and low spikes, and put out an average voltage reading based on the sending unit’s position.
Please log in to post an answer.
One of the suggestions was one that I was first tempted to suggest, but did not as it is potentially dangerous. The use of a filtering capacitor to dampen the voltage should not be tried. The sender circuit is low impedance and needs a very large value capacitor to be successful. Unfortunately that capacitor will store quite a bit of energy and the sender rheostat, in the presence of fuel and air, could create a spark that is sufficient to explosively ignite the fuel vapor. (Think TWA-800).
You have gotten a lot of good suggestions. The best ones deal with repairing the system as it was designed. You might want to make sure the sender is grounded properly and try an external wire back to the gauge to bypass the original one as a test.
The type of filter you need is called a "median" filter. I designed it in software when I worked for Marsh-McBirney, Inc developing flow meters. The data you have can be filtered with a PIC to keep anything from three to seven readings & then just select the value in the middle of the pack. Use even higher numbers if the readings aren't stable enough. With our units taking five readings, it did the trick and was immensely better than an "averaging" filter which took forever to ramp up or shut down, not to mention you never saw anything close to a real time value.
Most of the older vehicles used a +5.0 VDC voltage for all of the instrument cluster. And, the sending unit in the gas tank is one of the resistors in the Wheatstone Bridge circuit for the gauge. In my 1977 GMC 1/2 ton pickup I had the exact same problem. I fixed it by removing the dash group and then removing the gas gauge. The gauge had two long screws with loose nuts that depressed into some clips that connected to the flimsy circuit traces. I carefully tightened those nuts. I still had the typical GM problem of driving for a week to work, and at 1/2 tank I could only make it to work one more day, because the gas needle went to empty in 30 miles. I'd suggest checking your gas gauge connections in the cluster, loose nuts on the gauge's protruding screws, and the flimsy circuit trace connections.
On my old Dodge, (and this applies to Fords and Chevys too) the instrument panel has a thermal voltage regulator. The instruments operate at a voltage of about 5-7 volts. The guages have built in damping.
The thermal voltage regulator is nothing more than a bi-medal strip with a heating coil on it, that opens and closes a contact, thus maintaining a poor, but semi constant voltage. I've replaced these devices with simple voltage regulators like a LM317, set to a voltage that approximates 6 volts. More reading here. www.chevytrucks.org/tech/gasgauge.htm
Mentioned in the above link, is to check for a GOOD ground at the sender in the tank.
I saw the removal of a gas tank and the sending unit was a variable resistor. This resistor sits in the gas! I would not play with this system period!
As you know, a gas sending unit is just a variable resistor acting as a voltage divider. There are several possible reasons for a jumpy reading. One would be if the float is not well placed in the tank and the gas sloshing around just moves it up and down. If that is the case, a filter may help.
Since the gas level should change very slowly, a simple series connected capacitor/resistor combination connected from the output of the sending unit to ground would be a good filter. A large electrolytic with a high enough Voltage rating to withstand any spikes in the 12V system would be best.
You need a time constant that is longer than the period of the gauge's jumps; perhaps a couple of seconds. But you also need a resistance that is small enough to allow enough current to pass through it when needed to maintain the dial reading. TC(Time Constant) = R * C or C = TC/R. R is the value of the series resistor. I would take it to be a few Ohms, perhaps 10 as a first try.
It is a bit hard to calculate the power dissipated in this resistor as it depends on the frequency and severity of the fluctuations but I would start with at least a 1 Watt size and if that burns up, try 5 Watts. With a one second time constant we have: C = 1/10 = 0.1 Farads That is 100,000 uF which is probably impractical. I would try a nice large electrolytic and see what happens. 20,000 uF at 25 Volts would be a good first choice.
If that helps but is not enough, go larger. The negative (-) terminal on the capacitor would go to ground. It does not matter which side of the capacitor the resistor is on as long as it is in series with it. This will not effect the accuracy of the readings as long as the added resistor is not in series with the line to the gauge.
Another possible cause would be a bad sending unit, but you say you have already replaced it so that is not likely. A third cause would be bad wiring between the sending unit and the gauge. Before you try anything else, you might run an extra, temporary wire to see if it helps. If it does, repair the harness.
Finally, it could be a bad gauge. After trying the above filter and extra wire fixes, you could temporarily rig a 100 Ohm pot at the dashboard to activate the meter. Disconnect the wire from the sending unit and observe the meter as you drive down a bumpy street. If it still jumps, it is the gauge itself and that must be replaced.
