Fuel Level Sensor design 1

[CarlFST60L]

Hi guys,

I need to come up with an easy way to measure the fuel in a set of vehicle for telementry that is accurate to around 1L.

I dont know enough about how these cars actually mesure the fuel level, i believe most use a float and resistor wheel of some sort?

Most cars, i believe, do have access to the fuel level via the OBDII inerface.

I would much prefer to design an OP amp with a very high impedence input, measure the 'signal wire' and have a proportinal voltage output which i can use.

Any advice, tips, or even products that already do this would be appreciated.

Carl.

 

[itsmoked]

Fuel measurement in vehicles is a major pain in the rear. A seriously tough job to do, even to a relatively large liter accuracy. Of course the reason is the slosh and the shape. If tanks were 4 inches in diameter and 6 feet tall it would be easy, but sadly they are 6" deep and 3x3 feet wide/long. The fuel level sensors in cars are pretty crude. They are there to tell you to head to the station and nothing else.

If you look at how they do short run mileage stuff they often hang a small cylinder on the bumper for fuel.

So! If you try to tap into the fuel pickup you will have a non-linear rapidly changing reading.

What can you do or change? Why do you need this accuracy? Are these stock automobiles? Maybe we can come up with a solution based on your answers.



Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[CarlFST60L]

The measurement will only be taken at idle, and only on a flat surface.

 

[CarlFST60L]

The fuel level only needs to be checked from the factory sensor as the car comes to a stop at idle.

 

[itsmoked]

Yes but standing still?
Only on a test track?
A level flat surface?
Can you modify the tank?
Does the tank have a drain plug?
Will the tank always contain at least 8-10 liters?
Can the trunk be modified?
Can you pull the tanks?
Does this system have to work for standard road driving?

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[CarlFST60L]

Yes but standing still? Yes
Only on a test track? Why does it matter?
A level flat surface? Yes
Can you modify the tank? No
Does the tank have a drain plug? Stock car from the factory
Will the tank always contain at least 8-10 liters? no
Can the trunk be modified? no
Can you pull the tanks? its a fuel tank, standard, as they come from teh factory
Does this system have to work for standard road driving? No

All i want to do, is measure the fuel level in mum or dads car when its parked at home in there garage, they pull up, i log the fuel level. I was going to design an opamp to measure the fuel sender voltage, but i hope there is other easier ideas, or, someone knows something more about petrol sender voltages and measure fuel levels with only electrical connections

 

[Skogsgurra]

Is this device going to be mass produced?

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[itsmoked]

All i want to do, is measure the fuel level in mum or dads car when its parked at home in there garage

Why didn't you just say so.

Here is what I suggest.

Go to

http://www.alldatadiy.com/

You pay your $20 and for a year you can access the service documents for a particular car.

Hunt down the schematics and find the fuel sensor circuit. Figure out how it operates and what color the wire is and in which harness it resides, (this is easily worth the price of entry). Then once you find the actual wire in the car use a regular DMM to check the signal you see verse the fuel gauge reading. Once the tank level has dropped, say a quarter, check it again. You will then know what region the signal inhabits. Go ahead and design a high input resistance circuit to do what you need, as you've suggested.

I would consider using a micro to control any telemetry. You can purchase little RS232 to RF and back again modules. Then you just receive it from the car via a serial port on your computer..? If you really want to do it on the cheap you could use a PICSTAMP

http://www.parallax.com/html_pages/products/basicstamps/basic_stamps.asp

with its built in BASIC (easy software) to read the fuel level and then use cheap hacked TV type remotes to send stand number codes. To an IR sensing circuit.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[CarlFST60L]

It may be mass produced if it catches on, its not just a one off either, atleast 250units will be made for various vehicles...

This is the problem, now that i think you can see what im doing, i want to know if there is a standard the GM/Ford etc use, like, are they usually resistance, voltage?

I find it hard to believe there is no solution already there that would suite a majority of vehicles...

I need either digital or analogue output, best is analogue, as it will be connected to the analogue input of my GPRS telementry system...

I also have thought about using the OBDII interface, but im not sure what percentage of vehicles have the fuel level available via OBDII, or, where to get this info

 

[OperaHouse]

For race vehicles that have tanks that are foam filled they have a capacitive probe that has a metal sheath. The controled entrance to the probe eliminates a lot of sloshing problems. This is basically a dual 555 design with variable frequency, driving a fixed pulse, to an averaging circuit that produces a dc output. I remember a friend that built a digital readout that attached to the float sensor. He sat at the pump for about 45 minutes putting in .1 gal at a time. That was one pissed gas station owner!

 

[Comcokid]

There are, in a way, "standards". Note that I put "standards" in quotes.

Most fuel sensors are resistive. They consist of a wiper traveling on a thick-film ceramic resistor card. Older designs use a wiper on a wire-wound card.

In the 50's and 60's, different gauge/sensors manufacturers came up with designs that had certain end resistances and followed a certain curve in the middle. They are sometimes refered to by SW (Steward-Warner) or AC (AC Delco). There are others as well as some European auto makers had their own. Now, as each company copied each other, they may not have gotten the copy exactly the same as the information had to be reversed engineered. This has lead to variations in each of the "standards". This was acceptable as fuel gauges didn't need high accuracy (fuel tanks vary vehicle-to-vehicle) and manufacturers always had some reserve. These resistive senders are designed to work with heavily damped gauges. The gauges are designed to work with the fuel sender in such a way as to compensate for variations in the battery voltage.

Later in the 80's electronic insturment clusters began to appear, and in the 90's, different fuel sensor designs and methods. I'm not sure what all variations you might encounter today. In some of todays cars, it may have a mechanical pointer dashboard, but there is a micro inbetween the sender and gauge which may use look-up tables to linearize the non-linear profile of the fuel tank.

 

[LionelHutz]

I found this at

http://www.classicinstruments.com

when I looked at the Sender information page.

If Gauge is: Empty Full
GMC <=1964 0 30
GMC >=1966 0 90
Ford >=1987 16 158
Ford < 1987 AMC 73 10
Toyota-Nissan 1985 and later
90 0
Stewart-Warner 240 33
Auto-Meter 240 33

Ugly table but you get the idea.

So, there are standards used for the sending units. There will be standard ways these companies read the sending unit as well but I'm just not spending time to Google for this info as well. I'm sure you can use Google yourself. I seem to recall it's not a simple linear voltage but some type of switched signal.

The float and potentiometer always seem to be designed to give a fairly linear resistance change as it moves from the top to the bottom of it's travel. The problem is that most gas tanks are not square boxes so a depth measurement of the fuel does not provide an accurate measurement of the fuel in the tank. The 4th generation GM Camaro/Firebird platform was well known for having a fuel gauge that read 1/2 tank when the car had about 1/4 tank left and 1/4 tank when the car was almost empty as an example of this.

On top of that, the sender design really isn't accurate anyways so it certainly won't give you a fuel reading down to 1l accuracy over the usage of the whole tank of fuel. If you want 1l accuracy you will have to calibrate the module you build to the vehicle, likely with the use of a look-up table.

 

[OperaHouse]

I personally don't think you stand a chance of measuring 1L with the tank sender. Many of these senders have a step resistance function. Measure the rough length and width of the tank and then figure out the delta of what 1L would change the level. All the bits in the world won't make up for a sender. Averaging out a tank bouncing around a little will give you more bits out of a sender.

 

[CarlFST60L]

Thanks for your input.

The design is basically going to be an opamp to allow for all the diffrent electrical signals possible. That will feed an ADC on a processor. I will write software to do both timed and analogue signals. The cars will be calibrated by running empty, then adding 1L at a time while stationary. Obviously, i will run some filters in software and some hardware to get the signal smooth and 'accurate'.

 

[waross]

Let us know how it works.
Thanks

 

[Warpspeed]

I would say there is about a zero chance of this working.

The problems are not just at the tank sender unit. The tank sender unit has all the problems previously mentioned concerning non linearity of both the resistance element, and the shape of the fuel tank.

But there are even more horrendous problems at the dashboard end that supplies voltage to the tank sender.

These usually have a very crude bimetallic pulse width modulated voltage regulator that supplies an "average" regulated output voltage to all the various analog dashboard instruments.

This fiendish device typically makes and breaks the battery voltage at about 0.5 Hz using a heating element, a bimetallic strip, and an electrical contact. The clever part is that the on/off ratio varies not only with battery voltage but also AMBIENT TEMPERATURE.

This works, because the gauge elements also use a similar bimetallic strip to very slowly move the fuel (and engine temperature) gauges. The clever part is that this fiendish and diabolical voltage regulator system compensates for both temperature AND battery voltage over a fairly wide range of either.

The voltage going to the tank sensor will be pulsing on and off, going from zero to "whatever", depending on the battery and alternator voltage and the (?) amount of fuel, at an on/off duty cycle and frequency that is temperature dependent.

How you are going to measure and interpret a grossly non liner pulsing voltage that varies with battery voltage and temperature with an ADC I really do not know.

Probably the only realistic way to get anything like a reasonable figure for total fuel usage might be to measure the cumulative open time for one or more fuel injectors, and add a carefully chosen scale calibration factor.

That might be possible by "ANDing" the injector drive voltage with a high frequency clock (up somewhere in the MHZ range) and totalizing the bursts of clock pulses in a divide by "N" counter to set the scale factor. An electromechanical counter would then tick over in litres, or tenths of litres to indicate fuel used.

Scale factor can be adjusted by either changing the clock frequency, or changing the division ratio of the electronic totaliser. It is doable, and completely digital, so once set up it should be fairly consistent.

 

[CarlFST60L]

Thanks Warpspeed,

The reality of a practical solution is always far from the orginal concept.

After spending the last 8 years of doing R&D, I have seen projects fail after weeks, even months of development... I have put some special clauses in the contract which make special not of the final accuracy to ensure that this wont become a problem to complete the contract.

Luckly, the project is required to work with only two vehicles, so working some magic shouldn't be to much of a drama.

I'm not sure what information i can share, but i will share basically how the project goes.

Thanks again, if there is any more input, please keep it coming!!

Contracts arnt signed for a week, so the more input i get now, the more i can plan for, Warpspeed has just added a temp and main voltage test

Carl.

 

[itsmoked]

I believe Warp's comments may not apply to newer vehicles. It certainly does to older ones!

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[djroseman]

Commercial vehicle fuel tanks are a big problem since they are many different sizes and shapes.

Are you able to gain access to the inside of the tank? If so you could use any number of commercially made sensors. If you cannot its a bit more tricky.

On a plastic tank, you could use radar, which is useless on a metal tank. Also prices are not exactly cheap, but these units should provide the accuracy which you require.

If you only have to measure at idle, I would seriously recommend including an inclinometer to your solution and work out what calibration factor you would need to apply to your measurement according to the angle, and also waiting for a set time delay after the vehicle has stopped to give the fuel time to settle.

Dave

 

[Warpspeed]

Radar, optical or ultrasound all have some fairly significant problems in this application. Gasoline is not particularly measurement friendly stuff.

This is far from a simple problem to solve. Even the basic humble float requires a certain depth to raise it off the floor of the tank. So it can never be truly reliable at a very low fuel level with violently moving fuel.

Arguably the best method may be thermal sensing. Take a physically small thermistor and drive a small current through it. If it is located in air (or fuel vapor), it will self heat and the electrical resistance will significantly change. If it is submerged in fuel, the fuel will very effectively carry away the heat, and the electrical resistance will remain in the "normal" range.

We are talking about a very small pin head sized sensor here, with extremely low thermal mass, and low power with a thermal response of a small fractions of a second. The thermistor output would be an on/off, or wet/dry reading.

So you locate perhaps a hundred (or more) of these really small thermistors stacked one above the other in a vertical strip in the fuel tank. These sensors could be connected electrically in a 10 x 10 matrix, and scanned with a small microcontroller. The submerged sensors will read significantly different to the dry sensors, and the response will easily be fast enough to follow surging fuel.

The microcontroller can then do some clever things in software, such as compare the tank reading with a lookup table to correct for really strange tank shapes. It could also do the necessary smoothing and averaging to correct for surging fuel level due to vehicle motion. It could then even provide a suitable pulse width modulated output to drive the existing dash fuel gauge, or directly drive a digital display or a trip computer or even telemetry.

If I live to be over a hundred years old, I may one day attempt this myself for amusement.

It is so far, the only method I can see of getting an accurate enough fuel level measurement to be really useful.

The idea is simple enough, and if made in volume could I am sure, be produced at fairly low cost. Although the software would need to be fuel tank shape specific.