Fuel gauge recalibrator project

Kaldek

2 AYC Bars
Premium Member
Location
Melbourne
First Name
Ed
Drive
2000 Legnum VR4 Manual, Ford Territory family runabout, BMW K1300R.
EDIT: This project is now complete and successul, however I recommend skipping to the posts from 2021 as most of the stuff in between is hot garbage because I had no idea what I was doing.

So folks, the Legnum fuel gauge non-linearity shits me and I want to fix it.

My plan is to put together an Arduino-based solution that does the following:

  • Disconnects the gauge from the senders
  • Sends a 5v signal to the fuel senders and measures their signal using a voltage divider with a 1kohm resistor
  • Connects the gauge to a digital potentiometer whose resistance is controlled by my calibrated resistances that will make the needle point where I want it to

The only problem with the job is that to calibrate the gauge I'll need to drain the tank to empty and then fill the tank an exact consistent amount (say, 5 or 10 litres) at a time, measuring the resistance of the fuel sender and needle position at each interval. I keep doing this until the tank has 60 litres in it. From here I build my table of original values and the needle position, and can use it to extrapolate a desired needle position for each amount of fuel consumed.

The goal will be for the needle to read as follows:
  • Full tank: F (duh)
  • 45 litres left: 3/4 mark
  • 30 litres left: 1/2 mark
  • 15 litres left: 1/4 mark
  • 10 litres left: Fuel warning light comes on (whatever needle position this needs to be)

It's funny but one of the guys I work wrote the code for the Magna and 380 dashboards when he worked for Siemens VDO. He'd probably have a good idea about the code within the dash itself. Might help!
 

Smurph

Leaving Skid Marks
Lifetime Member
Location
QLD
First Name
Symond
Drive
1997 VR4 Galant PFL
My understanding is because the fuel pump and hence the measuring instruments, like the float and the gauge etc are mounted lower in the tank. IE if you have a look the tank itself is larger and fills higher than the actual pump hence the full for ages and then once its low enough to give a correct reading you are already half a tank down so it in turn seems like its not linear. Just my thoughts
 

Kaldek

2 AYC Bars
Premium Member
Location
Melbourne
First Name
Ed
Drive
2000 Legnum VR4 Manual, Ford Territory family runabout, BMW K1300R.
My understanding is because the fuel pump and hence the measuring instruments, like the float and the gauge etc are mounted lower in the tank. IE if you have a look the tank itself is larger and fills higher than the actual pump hence the full for ages and then once its low enough to give a correct reading you are already half a tank down so it in turn seems like its not linear. Just my thoughts

Fair call. I may not be able to get the very top of the gauge to be accurate, but if I can at least get it close from 3/4 on down that would be enough for me.
 

Kaldek

2 AYC Bars
Premium Member
Location
Melbourne
First Name
Ed
Drive
2000 Legnum VR4 Manual, Ford Territory family runabout, BMW K1300R.
One thing I am curious about is if the fuel sender is getting battery voltage. If so I'll need to be careful because that would mean that the dashboard needle might be displaying based on knowledge of that (i.e. the car's voltage can be anywhere from 13 to 14.8 volts due to the alternator). If I tried to compensate for the fluctuating voltage in my own circuitry I might end up confusing the gauge in the dash.

I'm kinda hoping the sender is getting a regulated 5 volts. That would make life sooo much easier.
 

trotty

3 AYC Bars
Location
nsw
First Name
Daniel Trotman
Drive
2000 vr4 Legnum 5 sp man
2006 Suzuki Swift Sport 5 sp man
It's the shape of the tank that gives the inconsistent gauge. It bigger at the top so float moves less and u get more kms. The tail shaft hump is the reduction in the bottom of tank The low level light is a separate float/switch
 

Kaldek

2 AYC Bars
Premium Member
Location
Melbourne
First Name
Ed
Drive
2000 Legnum VR4 Manual, Ford Territory family runabout, BMW K1300R.
It's the shape of the tank that gives the inconsistent gauge. It bigger at the top so float moves less and u get more kms. The tail shaft hump is the reduction in the bottom of tank The low level light is a separate float/switch
The low level light being different is good then as it means I won't have to worry about calibrating to suit it.

However I'm also not sure if you're telling me to not do it, or just giving me some context.
 

BCX

Administrator
Moderator
Location
SA
First Name
Bill
Drive
2000 Galant Type-V
1997 MK Triton GLS [6G74 conversion]
2019 i30 N-Line
Hey mate,

Couple of things:

Fuel sender from memory is 5v from the cluster.
Fuel sender acts a reostat, not a potentionmeter... the gauge measures current flow, not voltage. Might want to check that out to be sure... again, this is from memory.

Here's a sample wiring diagram that is very similar to our car, just the colours are different (borrowed from magna/verada).

cluster.PNG





Low fuel light is seperate sensor (sensor that conducts when not submerged in fuel), this has a dedicated wire:
low%20fuel%20light.PNG



Something else to consider... maybe needs some signal averaging (in software) to smooth the signal... the float moves around quite a bit... but I guess the guage has that built-in so you might not need it.

Hope this helps?

Cheers,
Bill

PS. I still have to send you that stuff re: cruise control.
 

Kaldek

2 AYC Bars
Premium Member
Location
Melbourne
First Name
Ed
Drive
2000 Legnum VR4 Manual, Ford Territory family runabout, BMW K1300R.
Thanks mate. Sure helps! Also had a chat to my mate who wrote the dashboard code. He mentioned that for the magna and 380 they had to drive around for 10 hours all over the place, capturing fuel sender data every 80 milliseconds, all just to build the gauge smoothing algorithm. Mental.
 

Kaldek

2 AYC Bars
Premium Member
Location
Melbourne
First Name
Ed
Drive
2000 Legnum VR4 Manual, Ford Territory family runabout, BMW K1300R.
FYI I'm just delaying this project due to cost and time problems at the moment.
 

Kaldek

2 AYC Bars
Premium Member
Location
Melbourne
First Name
Ed
Drive
2000 Legnum VR4 Manual, Ford Territory family runabout, BMW K1300R.
Well this project has finally bubbled back to the surface as I've currently run out of other things to do on the car. The next step is to directly measure the voltages coming from the dashboard so I know voltages, then measure the resistance ranges of the fuel sender. After that I can store that information away and then work on a circuit that could intercept and fiddle with the signal.

I really don't what that circuit would look like right now. Do I disconnect the fuel sender from the dashboard, measure current with my own system and then simulate a fuel gauge using a computer controlled resistor? I'm no electrical wizzkid unlike @BCX but it's the learning I get a kick out of anyway.
 

Kaldek

2 AYC Bars
Premium Member
Location
Melbourne
First Name
Ed
Drive
2000 Legnum VR4 Manual, Ford Territory family runabout, BMW K1300R.
So I'm wondering why the fuel gauge in the car isn't just an ohmmeter and is instead a current driven device. Seems like this is historical, being:
  1. Gauges used to use the current passing through a bimetallic strip in the gauge which causes the strip to bend, pulling on the gauge needle; and
  2. It reduces wiring (no loop of wire required to go from the battery to the sender and back to the gauge.
What I don't know is if this design is somehow safer from a spark point of view? I mean, sparks in the fuel tank would be a death trap. Ka-boom central. I need to make sure I do NOT somehow increase the likelihood of this.

upload_2016-4-19_12-46-56.png


EDIT: After pulling apart the Legnum gauge it is shown to be an "air core" gauge where the current passing through the coils causes the needle to align with the magnetic field. Still current driven, but I assume a much smaller current than a bimetallic strip.
 

Kaldek

2 AYC Bars
Premium Member
Location
Melbourne
First Name
Ed
Drive
2000 Legnum VR4 Manual, Ford Territory family runabout, BMW K1300R.
And this is an interesting statement, because I'm essentially trying to create just such a thing.

upload_2016-4-19_12-51-49.png
 

Kaldek

2 AYC Bars
Premium Member
Location
Melbourne
First Name
Ed
Drive
2000 Legnum VR4 Manual, Ford Territory family runabout, BMW K1300R.
Getting interesting.

I could possibly measure the output voltage from the fuel sender and use that with an Arduino to control a digital potentiometer that bridges the voltage feed coming out of the gauge and ground, before the sender therefore affecting the current passing through the sender.

There is also a guide on using a manually controlled potentiometer to do a similar job, but it wouldn't help us here because what we're trying to solve is the non-linear shape of the fuel tank.

upload_2016-4-19_13-3-1.png



The basic functionality of this option is that after knowing exactly what the output voltages from the sender unit are at a known tank volume (by measuring voltage as I slowly measure doses of fuel into the tank), I can intercept and tweak the current passing through the gauge by directing that current through my digital pot rather than the fuel tank sender, thus fooling the gauge into displaying what I want it to.

Because it is a digitally controlled system I can program a non-linear reading for the gauge to match the non-linear shape of the tank. That's the theory anyway. :)
 

Kaldek

2 AYC Bars
Premium Member
Location
Melbourne
First Name
Ed
Drive
2000 Legnum VR4 Manual, Ford Territory family runabout, BMW K1300R.
OK I think step one is actually to bridge the output voltage from the gauge with a manual potentiometer and see if the basic theory of the plan works, just by watching the gauge. I figure I can do this just by back probing the wiring into the sender with T-pins and then hooking those up to a variable pot. I've got a couple lying around.
 

Kaldek

2 AYC Bars
Premium Member
Location
Melbourne
First Name
Ed
Drive
2000 Legnum VR4 Manual, Ford Territory family runabout, BMW K1300R.
Question for the masses - does the VR4 have two fuel senders or one?
 

trotty

3 AYC Bars
Location
nsw
First Name
Daniel Trotman
Drive
2000 vr4 Legnum 5 sp man
2006 Suzuki Swift Sport 5 sp man
Has fuel pump on drivers side and two low level sensors one on each side of the tank with a transfer pipe that that works on vacuum to draw fuel from passenger side.
 

Kaldek

2 AYC Bars
Premium Member
Location
Melbourne
First Name
Ed
Drive
2000 Legnum VR4 Manual, Ford Territory family runabout, BMW K1300R.
Well this just became a bit more interestig. There are two fuel senders, one on each side and what looks like as @trotty said a mechanism for transferring fuel from one side of the tank to the other. Sounds like this might be a jet pipe that pulls fuel across using a venturi effect. It's a neat solution because if the passenger side went dry for whatever reason, there is no pump over there to burn out.

upload_2016-4-20_9-2-4.png


I'm not sure yet if and how this complicates my project.
 

Kaldek

2 AYC Bars
Premium Member
Location
Melbourne
First Name
Ed
Drive
2000 Legnum VR4 Manual, Ford Territory family runabout, BMW K1300R.
OK so some project notes that are worthwhile. The service manual quotes that the main fuel sender is the one on the driver's side and the sub sender is on the passenger's side. The one with the six pin plug is the main one, the two pin plug is the sub.
The image below shows the Full and Empty resistances of the rheostats in the sending units.

upload_2016-4-20_10-9-17.png


At this point I am assuming that the two senders are connected in series as per the Evo 6 wiring diagram: Voltage is fed from the gauge through to the sub sender, then to the main sender. The resulting current drawn by both senders is what the gauge uses to display a reading. For a mid '90s design I'd still call this a total hack job as we have witnessed by the wildly inaccurate gauge. Put simply, oddly shaped fuel tanks need a computer controlled gauge, not a simple current meter programmed by fiddling with resistances and float throws.

Also I sort of confirmed what @trotty said about the low fuel warning light. On the driver's side there is a "Fuel Remaining Amount" sensor, and on the passenger side there is a "Fuel Level Warning Switch". I am not sure how these two function together as the passenger side is a simple float with a switch while the driver's side is a thermistor (that stays cool when submerged in gasoline). Maybe the driver's side one is able to detect a more subtle variance in level and is a better solution since that's where the pump is located.
 
Top Bottom