Making a non-linear resistance look linear

Pete Verdon · 2010-01-31T23:09:51+00:00

Hi,I'm planning to fit a fuel tank with a contents gauge. The sender that the tank manufacturer will put in is essentially a variable resistor, which reads 180 ohms when the tank is full, and 0 ohms when it is empty with, presumably, a linear resistance in between. There is a gauge that connects to this - presumably an ammeter and a means of passing current through the sender?My problem is that the tank is not square, and so the height of the sender float is not directly proportional to the amount of fuel actually remaining. More unfortunately, it's actually "the wrong way round", so that when the gauge reads half-full there will actually be significantly less than half a tank of fuel left - perhaps a quarter or less.I've wired up a couple of PICs and so on in my time, but have very little experience with basic analogue electronics. Is it possible to build a passive circuit (perhaps a network of resistors?) that will "convert" the resistance of the sender so that it's proportional to the amount of fuel remaining rather than the height of the sender float?Some sketches of the tank here: Note that the lame attempt at isometric projection actually makes the tank seem squarer than it is - the slope is in fact very pronounced indeed, it's basically wedge-shaped in all three dimensions, plus a sneaky little change of gradient in one side.This shape means the perfect correction is probably quite complex - but all I really want to do is pin 3/4, 1/2, 1/4 in their right places. I can pour water through the tank before installing it, to measure the "observed" resistance at those points.Note the "orientation" of the sender is indeed 0 ohms when empty - I believe the other way round is normal in the US but I haven't made a mistake! I could special-order the American sender if required, but it would then make my gauge read backwards unless flipped again as part of the circuit.Thanks for any help or suggestions,Pete

P

Pete Verdon 16 years ago

Hmm. Looking at some of the hair-raising suggestions being proposed for an analogue solution, you might have a point :-)

Pete

Vote

M

mike 16 years ago

Sometimes high technology ain't the best solution. Plot a non-linear meter scale and paste it on the meter face.

Vote

B

BobS 16 years ago

t.

ther.

The gauges you are talking about use quite a bit of current, having a sensor and gauge circuit resistance of 100-300 Ohms power to ground. A micro based circuit should use less current, especially if used with a voltmeter type gauge. For a micro look at PicAxe, $20 will get you running with an easy to use Basic interpreter.

Westach makes gauges that can be adjusted for zero and span. Also they offer fluid level sensors, from Centroid Products, that can be creatively bent to help linearize the output. Centroid offers both current and voltage output sensors with digital or analog displays. Both companies are worth calling about your application.

Calibrating a sensor with exposed resistive element in water vs fuel can affect you readings since the water is most likely conductive from contamination.

Vote

W

whit3rd 16 years ago

[ fill tank with water, observe sensor output versus fill level]

e

A couple of quad op amps and some diodes, a reference voltage and a network of (adjusted) resistors might be 'simple' in principle, but it scales in complexity with the number of 'breakpoints' required in the curve. Easier, to just re-print the scale behind the meter needle.

Vote

B

Bill Sloman 16 years ago

As everybody else has said, digitise, linearise and use some kind of digital to analog converter to drive the meter.

If you must go analog, there are any number of ways of getting a monotonic (always slopes the same way) non-linear relationship between input and output.

Here is an old patent that is probably a massive ovekill for your application

formatting link

National Semiconductor's Application Note 31

formatting link

is a compendium of op amp circuits - it includes a "nonlinear operational amplifier with temperature compensated breakpoints" at the top of page 26.

-- Bill Sloman, Nijmegen

Vote

Y

YD 16 years ago

Late at night, by candle light, Pete Verdon penned this immortal opus:

I missed the OP, so I may be off target here. Most fuel gauges I've seen don't worry too much about absolute linearity. The actual readings seem to be "full", "empty" and "time to think about filling up". In-between you just give it the occasional glance to see if it's still OK.

- YD.

Remove HAT if replying by mail.

Vote

P

Pete Verdon 16 years ago

That's all I need, but the shape of the tank (wedge shaped in two dimensions) is such that the non-linearity is rather extreme.

Pete

Vote

D

Don Lancaster 16 years ago

The more extreme, the better table lookup is.

Many thanks, Don Lancaster voice phone: (928)428-4073 Synergetics 3860 West First Street Box 809 Thatcher, AZ 85552 rss: http://www.tinaja.com/whtnu.xml email: don@tinaja.com Please visit my GURU's LAIR web site at http://www.tinaja.com

Vote

P

Pete Verdon 16 years ago

Pete Verdon wrote: > Is it possible to

Thanks for all the suggestions, but it looks like the actual solution will be simpler still. Apparently it only costs the tank manufacturers about £10 extra to get a custom sender made by their supplier, whose resistance at each point is calibrated to match the shape of the tank. Clearly that's the way to go - it hadn't occurred to me that it would even be possible.

Pete

Vote