I picked up a Jaeger fuel tank level sensor in a junk yard with the intention to connect it to an arduino mega board. Originally the sender didn't work, but after doing some researching, I found the connection that needed to be cleaned. Now when I attach a multimeter on the ohms setting set at 10 k I get a reading from 0 ohms to 40 ohms depending on where the float level is set. Based on my research I assume this is the way the sensor should work.
My question is how do I determine what voltage should be run through this sensor to make it work correctly. At this time I am not trying to hook it up to an arduino. I am just trying to measure a varying voltage with my multimeter so I can breadboard it and make sure I don't toast the arduino when I get that far. I have hooked it up to 5 volts, but I don't get any variance on the voltage when I move the float. It just reads a solid 5 volts at all positions of the float. I also tried 12 volts, but like the 5 volts I don't get any variance. It reads a solid 12 volts.
I am guessing that I am missing something pretty basic here, but I have run out of ideas. Any help would be greatly appreciated.
You need to go to the car model that the sender was removed from.
The complete meter circuit needs to be obtained in the automotive manual for that model.
There are different methods of powering meter/sender circuits.
Older models us a switching supply, some use a bridge, some a straight divider.
I would be concerned on frying the sender. Most are extremely fine nichrome wire. I had to repair a rare VW Razor sender by replacing a hair strand of spring loaded 26ga wire. All be cause I put 12 direct to it.
post the car and year if you would.
Standard practice used to be a regulated 10 volt supply within the instrument panel that drove the temperature sender, oil pressure sender, and fuel guage. The meters that showed these functions on the dashboard were then just crude volt meters, calibrated to show full scale deflection at 10 volts.
I don't know if modern vehicles follow the same practice.
My VW fuel gauge didn't even have electricity: it was a Bowden cable attached to a lever on the float mechanism, and the inner cable movement moved the needle on the dash. Mind you, that was a '62 T1.
If memory serves, that was a Jaeger also. Or VDO maybe.
(It made a very crude accelerometer too btw 8) )
@JimboZA
My father once borrowed one one of those vehicles to take us camping in Cornwall. The brakes were very iffy when he collected it but finally stopped working completely. Upon inspection it turned out that the front brakes were completely inopperative and the brake cable to the back had finally snapped and got mangled. Unperturbed he Jury rigged one of the tent guy ropes and routed it through the cab to the front. When he wanted to stop he just shouted "brake" and we'd all pull on it together. I doubt such a practice would be technically legal nowadays. 
I don't have the model of the car this sender came from. I obtained it from a local junk yard. i.e. the tank was in the light steel pile with all the rest of the trash. What can I say, the price was right. The junk yard owner sold it to me for a dollar. I didn't loose much if I couldn't make it work and I thought it was worth an education in tank senders. My ultimate goal is to have an arduino monitor the level of the fuel in the tank and then light a series of LEDs to display the level.
I will look into trying to send a 10 volt regulated supply. I haven't fried the sender yet because I still get a reliable return on the ohms meter.
Thank you for the thoughts and ideas.
10 is good for a 12volt car, but if it was a 6V then???
Also: 
make sure you have a current limiting method, as stated wire are very small and work on microamps.
Some senders us wire as small as 36ga.
I do like your project effort as I have been diagramming similar wiring and ideas for my Dune Buggy project and plan to apply the same effort will apply to an old restomod 53 Hudson Hornet.
Will run some Arduino Minis over I2C to LCD display.
I hope you understand that the kind of sender this appears to be from your description requires the resistance of the meter itself as part of the circuit , and is not intended to have 12V connected directly to it or it will burn up when it is in the Full position.
I think the way I'd use it would be to connect a relatively large resistor in series with it and measure the voltage drop across the sender. If the resistor is 10 or 20 times the resistance of the sender's highest resistance, the reading should be relatively linear, as long as the resistance change is linear.
Keeping in mind that the resistance may -not- be a linear function of position, it may have been wound so that, in conjunction with the fuel gauge resistance, the fuel gauge position would be fairly linear.
Well, if you used a 5 volt supply and put a 40 ohm resistor in series with the fuel level sensor you would be reading a feedback voltge of zero to 2.5 volts.. That would be running the circuit at 125ma.. Might be a little high but it shouldn't burn it out..
If you were worried you could go up to an 80 ohm resistor from a 5 volt reference and that would cut the max current in half, but it would limit your voltage switch from zero to only 1.25 volts.. Your losing resolution/accuracy but you are lowering the current so that would be up to you to decide how accurate you need to be..
just google voltage divider calculator.. plug in 5 volts and the second resistor would be the sender(between 0 and 40 ohms).. try plugging in varying resistors for the first one and you'll see that I mean..
What liquid will you be using?? Is this for gasoline? Diesel? just curious...
You can put a constant-current through the sender and measure the voltage difference
across it. A 25mA constant current source would give you a full scale of 1V, which
if you selected analogReference(INTERNAL) would give good accuracy I think.
There are various constant current circuits, some with transistors, some with
opamps (note 25mA may be too much for many opamps). Or just use 150 ohm
resistor from 5V to the sender - slightly non linear but hey, the level/resistance
response is unlikely to be linear in the first place.
So:
5V ---- 150ohm ---- A0 ----- sender ---- GND
void setup ()
{
analogReference (INTERNAL) ;
}
#define CURRENT 0.025
#define VREF 1.1
void loop ()
{
float ohms = analogRead (A0) * VREF / CURRENT / 1024 ;
// float here means floating point, not fuel float!!
...
}