Fuel gauge recalibrator project

[TME_Steve]

Just a quick status update. I haven't ordered an external fuel pump yet, but once I do that I will work on getting a calibration going.

Yeah just get a cheap whatever pump. We had ours lying around from something. Pretty sure it doesn't work like it used to anymore

 

[Kaldek]

View attachment 43880

Lucky for me I have spares floating around. If u need more picks just holla

@trotty Do you have any ability to take pictures of the colours of the wires that go into the plugs on all those units? I'm trying to work out if the fuel senders are wired in series here or further up the wiring harness.

This assumes you have the fuel tank wiring harness somewhere...

 

[Kaldek]

Just another quick update. I've selected the Analog Devices AD8400 digital potentiometer as the simulator for the fuel tank sender. This is a 1kohm potentiometer with 256 steps, meaning that each step is roughly 3.9 ohms. The fuel sender in the VR4 reads 4 ohms when the tank is full and about 107 ohms when the tank is empty (based on the workshop manual, not based on a calibration yet).

With a maximum resistance of 107 ohms that means I can utilise 28 of the 256 potentiometer steps. Dividing this by four to slice the gauge up into quarters, this gives me seven steps (or gauge needle positions) I can use within each quarter of the gauge.

The chip requires 5 volt power, which the Arduino can provide. I don't know though if having the poteniometer pins hook into a 12 volt circuit is a problem. I assume not, but then I'm still new to this stuff.

 

[Kaldek]

Just for a little extra evidence of Mitsubishi throwing the fuel gauge circuit together for the VR4, note below this grab from the Japanese workshop manual.

With the 2WD Galant/Legnum and its simple fuel tank shape, the rheostat in the single fuel tank sender is 4 ohms when full, 112 ohms when empty. The combined resistance of the two fuel tank senders in the VR4 attempts to mimic this.

What does this suggest? There is no direct evidence but it suggests that Mitsubishi took an off the shelf part - the gauge circuitry for the dashboard - and knocked together two rheostats in series for the fuel tank sender in an attempt to make the off the shelf part work. Again, hence the wonky way the fuel gauge works with its long pauses and then sudden drops.

 

[trotty]

Unfortunately not. Rear half of car went to scrappers and loom with it.

 

[Kaldek]

Unfortunately not. Rear half of car went to scrappers and loom with it.

That's OK. Mitsfix has a couple he's gonna let me take a look at.

 

[Kaldek]

Here's a question related to getting the calibration data. When filling the tank, does the fuel neck split the incoming fuel into both sides of the tank? Or does it fill the driver's side first and then spill over the W hump?

I'm wondering if it means I can't get good calibration data unless I empty the tank from full version and measure resistance every litre, rather than filling the tank from empty and doing it. The former version would be much harder as I can't open the tank when it's full!

I really need to go inspect a tank at Mitsfix. God damn job and no more working from home!

 

[trotty]

Single fill point. Overflows to passenger side.

 

[Kaldek]

Single fill point. Overflows to passenger side.

OK then I think I've worked this out. I need to stop focusing on measuring the system as "litres remaining". The difference between my calibration test and a normally filled tank is just how long it would take the gauge to drop when you were driving around.

For example, if I put 20 litres in the tank and it only partially filled the driver's side of the tank, the only difference between that and a tank which was filled completely is that when the filled tank got down to 20 litres after driving, it would just take longer for the needle to move because the fuel delivery system is pulling fuel from both sides of the tank and therefore the sender float would be dropping slower.

Not sure if that description makes sense aside from inside my own head.

 

[Kaldek]

Quick update: finally got my external fuel pump delivered (chinese copy of a Bosch 044). Will go and seek connectors and hoses for this before I progress the work. I also got a bit sidetracked with my UTCOMP installation too.

 

[Kaldek]

Another interesting update. To complete this project, I can get the calibration data without draining the tank.. My UTCOMP trip computer is now so accurately calibrated that I can measure the fuel sender resistance values as I consume fuel. I only need to measure the fuel sender resistance every three litres to gather enough data points.

Only problem with this approach is it would take a while and require that I keep pulling over.

 

[Kaldek]

@TME_Steve I bought a Chinese Bosch 044 clone. Need to buy something that will screw into the inlet side of the pump though. Any guidance on something cheap I can do? I need to get a fitting in there that I can get some fuel hose onto.

 

[wintertidenz]

Is there a way you can tap into the UTCOMP fuel sender resistance values and log them with something like an Arduino? That would save a lot of time and messing about.

 

[Kaldek]

Is there a way you can tap into the UTCOMP fuel sender resistance values and log them with something like an Arduino? That would save a lot of time and messing about.

I don't have the UTCOMP connected that way - it's just told when I fill the tank and it assumes I did so, then measures fuel consumed. Assuming I always did that, I could ignore the whole project and just use this very accurate fuel gauge in the UTCOMP, as @trotty suggested many posts ago.

But where's the challenge in that?

 

[wintertidenz]

Very true. I'd like a more accurate fuel gauge, but I just don't have the time to start messing around with it - so I'm following this project with interest.

 

[trotty]

In all honesty I trust it that much now I just watch the distance to empty and it's always spot on with the fuel light. But it's all relevant to the average usage too as that plays a vital role in calculating dte

 

[Kaldek]

Now that I have a very accurately calibrated UTCOMP in the car, I have some interesting pictures to show you.

I noticed that the top half of the gauge for accuracy wasn't too far off. At 23 litres consumed, the gauge was slightly above the halfway mark - close to being accurate. But at ~29 litres consumed, it has dropped by a quarter of a tank reading. Six litres for a quarter of a tank!! My target calibration is one needle "notch" per every three litres, whereas here between halfway and one quarter, it has moved almost five notches. That should normally require a consumption of 15 litres - more than double what was actually consumed to make the needle drop this far.

 

[Kaldek]

I have also realised that with the UTCOMP I don't even need to measure the resistance (which would require pulling out the rear seat and stopping all the time to plug in a multimeter). I have an accurately calibrated fuel consumption meter now, which means all I have to do is:

1. Build and test the digital potentiometer circuitry and install it in the car (bypassing the fuel senders)
2. Set the digital potentiometer to 4 ohms to simulate a full tank. Monitor gauge needle position and adjust resistance to force needle to point at the "F" mark
3. Fill the tank, reset the UTCOMP fuel gauge
4. Consume three litres, monitoring this by checking the UTCOMP digital fuel gauge
5. Tweak the Arduino calibration until the needle matches where it needs to be for the amount of fuel consumed and note this resistance down
6. Repeat items 4-5 until the tank is empty
7. Save the calibration data
8. Profit!

 

[Kaldek]

Righto, AD8400 ordered. The build begins!

Also ordered a voltage regulator for the Arduino. Total cost - $1 including shipping.

 

[Kaldek]

Found a thread at the Arduino forums which describes what I'm trying to do.

The consensus is either a digital pot to drive the gauge like my solution or to drive the gauge directly with PWM using an Op-Amp. I honestly had not considered that!

I also clarifed from this forum post and my own disassembly of the gauge cluster that the fuel gauge is an "air core" gauge. The theory of operation for this type of gauge is that current flowing through the coils in the gauge cause the magnet to align with the coils. It is not a bimetallic strip (and I suppose that technology has not been used for a long time anyway).