Cleaning Inside of Fuel Tank

[Banjo]

We'll some days, you have a win; & yesterday was one of them.

My careful testing, was not too scientific, or in a laboratory, but it worked. It took over an hour to do this simple test, as the water had to be added a litre at a time, so that the water trickled into the tank, & didn't cause any "waves', or create disruption of the  water level surface.

The results of the test were very encouraging, although I won't be able to use all 14 off switched level point in the sensor, as the lowest one, is below the point, where fuel can be sucked out by the pump.  I could possibly lower the pick up point a bit, but think, I'll just stick with 13 switch points rather than the 14 available.

I don't specifically know how these sensors work, but it appears to be a number of magnetically switched reed switches, which are basically equally, distributed up the length of the vertical tube. 

I believe a little magnet, in the float, probably switches each one in turn.   The results are in the table above.

I will now draw a graph of with the capacity & LED switching points, overlaying them, & see how much non linearity there is.

Here is the LED Bar-O-Graph, indicating about half way through the test.

I was pleasantly surprised, that the two top LED & switching points arrived, before the tank filled. As soon the last top LED illuminated, at 57.5 litres, I stopped filling.  I then took the central pick up assembly out, & believe it or not, the water level was less than 1 cm below the top of the tank.  I carefully added more water, & after half a litre the tank was just over flowing.  The positioning of the top , or highest switching point, is nearly perfect.   It also indicates the tank is closer to 60 litres than I thought.

The useful capacity of the tank with the standard pick up point, is about 47.5 litre, or 45 litres, if you want to avoid running out of petrol.

I was so pleased with the way the electronics worked, I added a couple more inputs. One was an input from the parking or head light circuits, that dims the LEDs for night driving,, so they are not too bright, or distracting.  The other input was a need to disable the "damping function", of the sampling.  The last thing you want is the bar-o-graph LEDs, blinking on & off, as the fuel sloshes around in the tank. By taking several readings, say 1 minute apart, averaging those, say every 5 minutes, then updating the LEDs, results in a very stable display. However, if you are testing, as I have been, I need instant updates of the LED display, so can now switch between instant & delayed response.

I actually ran out of I/O on the 18  pin micro I was using,  but have just received a slightly bigger one with 20 pins & subsequently a bit more I/O capacity.

I will put this "rats nest" on a PCB now, transpose the code across to the bigger micro, & might actually put the tank in my KE30, & drive it around for  a couple of weeks, or tank fulls, to see how it performs in practice. I'll then pull, it out, & put the submersible hi-pressure EFI pump in the tank.  I'll come back & update this once I checked the linearisation, but based on what I've seen, it's not too far out.

Cheers Banjo 

 

 

 

 

 

 

 

 

Edited June 8, 2020 by Banjo

[Banjo]

In most automotive electrical installations, the frame or chassis of the car, is connected to the negative (-ve) terminal of the battery, and most 12 volt devices, complete their circuit, by one connection/wire of the device, connecting to the chassis. All lights, horn, radio, & instruments, work exactly the same way.  That's why the earth connection between the chassis & the battery (-ve) terminal is so important.  Most, if not all of us, have experienced dim lights, or poorly performing electrical items, as a result of poor earth connections.  This is also applicable to instruments.

The fuel tank sender, is such a device. It has a relatively low resistance; it's at the rear of the vehicle (long wire run); It uses only one wire to the sensor, with the chassis, being the return path.  In the electronic LED fuel guage design, I have built, I have eliminated the ground connection to the sensor altogether, so that, poor earth connections, will never effect it.

The design provides a constant current to the sensor, rather than a voltage.  The constant current, through the fuel level sensor resistance, will produce a stable voltage that is fed back to the meter behind the dash, via a two wire circuit.  The lowest voltage of the sensor output, is also elevated slightly above ground/earth/chassis potential.

It works extremely well. 

Unfortunately, this technique, cannot be applied to other standard automotive sensors like water temp sensors, as there is only one wire to them, & the return path, is via the engine block, as the sensor's screwed mounting, is the return connection.

However, I have in the past, taken a standard water temp sensor, drilled out the existing sensor element, & then glued into the hole, another NTC thermistor temp sensor with two (2) wires, neither of which, is connected to the chassis.

I've just come across a diagram, on the net, that depicts how the internals of these fuel level sensors work.

It appears it is a set of magnetic reed switches, that switch a string of resistances, in & out of the circuit.

I've also come across a Chinese supplier, that produces these same fuel level sensors, with a 4-20mA DC current, rather than a resistance output. I image, it is still a switched  output, so maybe it goes from 4 to 20mA, in 1mA steps.  I'll drop them a line, & see what they say.

There is a nice one pictured there, that combines sensor, breather & pick-up pipe, complete with filter, all in one assembly.  Would lve to get hold of one of them.

Cheers Banjo 

 

 

 

 

Edited June 11, 2020 by Banjo

[Banjo]

I plotted out the results on a bit of graph paper, to see what the linearity was like.

As the last 8 litres of the capacity of the tank is not usable, I have made that point 0% full.

If the linearity was perfect, then the spacings along the red line, of the LED switching points, between 0% & 100%, would be equally spaced.

It was never going to be perfect, as the positioning of the reed switches inside the vertical Stainless steel tube, are not equally spaced, as depicted by the bits of tape I stuck on the tube, at the point, where the switches actually switched.

The spacing between LEDs 4 & 5, & LEDs 7 & 8 switching on, are a bit wider than the others.  The switching points, right at the bottom & right at the top, are a little close together.  I could also not utilise switching points no: 2 & no: 13, which would linearise it a little better, & just use two less LEDs.

So I think, I might just pop it all in the car, & drive it around for a couple of weeks, or a few tank fulls, & see what it "feels like", in practice.

It's got to be better than the original needle gauge on my KE30, & much more reliable.

Cheers  Banjo