Fuel Gauge

[Jehu40]

Thanks to all, and for the diagram.

Two things:

It's definitely not the sender unit; I checked the resistance over its whole range and it was fine.

When I had the sender unit out I checked the incoming voltage at the plug, it was only about 5.5V, not 12V.

The fuel tank is plastic, and I didn't notice any earth (maybe because diesel isn't volatile like petrol), or maybe it relies on the earth in the plug.

[Sol]

Ok so if it's 2 wire then it must use current to drive the gauge level. I think the only options are:

Get the Haynes manual for the wiring and see what voltage is meant to go to the sender

Use DiagBox to read the level sent to the BSI and work back from there at the sender and wiring between them.

[Jehu40]

Ok so if it's 2 wire then it must use current to drive the gauge level. I think the only options are:

I'm assured (by Citroen too) that the gauge is controlled indirectly by the BSI. For some reason the two wires become four just above the sender unit. Yes, I need to find out hat the current should be.

[Gryffindor]

Okay, I did wonder before but thought it unlikely so said nowt, but four wires means canbus (+,- and 2 x data), therefore there must be a small plastic thing at the point where the 2 wires become 4, containing an analog to digital converter. The 2 wires feed an analog measurement in, and the 4 wires send a digitally converted value on to the BSI.

At least, that might be the case. If it is, your problem is probably in the 2-wire section somewhere.

But having 4 wires from the tank to the BSI does seem counter to the whole idea of multiplexing a car to reduce wiring.

[Jehu40]

I'm only up on the most basic electrics, like resistance and voltage on a multimeter. I understand what a DAC is too. I don't have Diagbox or even heard of it. I think I might eventually have to go to an autoelectrician, althought I doubt the average one willl know much about a Berlingo BSI to fuel gauge.

[Gryffindor]

Th BSI side is in wiring terms simple - it's a bit like USB. The four wires are ground and +12v, plus two data wires on which binary data much like Morse Code is sent. It needs two wires to prevent electrical interference - these two wires are twisted together along their length.

The fuel sender side of things you already understand fine.

But I still think it odd to have 4 wires the length of the car when 2 would do, so my interpretation could be wrong. I am also an amateur - I did electronics at hobby level (building magazine kits) in the early 80s and have only just got interested again.

[oilyrag]

The Haynes manual shows the 2 wire arrangement - with a transistor in the dashboard fuel gauge to operate the low fuel level warning light.  Looking at the tank sensor, the resistance increases as the float drops down and at some point towards the bottom end of its travel the warning light is also triggered.  
So, a simple test might be to ground the 'live' input wire that feeds the tank sensor. If the gauge is working as it should, then the level will show as full on the gauge, and drop to zero when the ignition is switched off. 
Then put a low wattage bulb in series between the 'live' input wire and earth ( mimicking the tank resistor) and the gauge should then show empty and the gauge low fuel warning light come on?
This might at least eliminate the gauge as the issue.
The comb on the fuel level sensor or moving contact might work OK on test, but maybe still have an unseen fault? 
I'm not sure about the low voltage that is being measured at the live feed but most cars have a stabilised voltage for the fuel gauge - maybe that is the correct stabilised voltage!
Good luck.

[Jehu40]

Thanks all. I was also told by someone else that the low voltage feed could be correct. Someone else suggested running an earth from the sender unit plug, so I'll concentrate on that end and do all the suggested tests.
I'll get back on that.

[Jehu40]

OK, tried the suggestions out, or most of them:
Drained most of the: tank

Unplugged the two wire connection from the sender unit from the four wire plug:
There was a resistance from the sender unit, so the wiring is intact:

On the four wire plug:
The voltage is 5.25V);
The fuel gauge moves to max when the live is earthed against the car (I tried all four points on the plug, only one responded):
After a short while the gauge went down to a bar under half, lower than previously;
Connected again, gauge went back up to max but didnt come back down, after severeal attempts -I tried shorting out all the points in turn, but nothing:
Then noticed that after two or three minutes the gauge went to zero, but no empty light;
Plugged the sender unit back in, the gauge went to where it normally sits, just over half, with next to no fuel; returned fuel to the tank, about 15 litres, gauge still at same place.

PS - some obserrvations: I went for a 40 mile drive yeaterday: the gauge started where it is now (just over half full) and went down nearly a whole line on the gauge. When the gauge went to full on the first earthing it returned to where it had been after the 40 mile trip. After plugging into the sender unit it ent to just over half again!
There is obviously some memory at play.
It seems therefore that the gauge responds to complete earth but is not recognising the correct resistance. It also responds to the sender unit being plugged in but does not adjust much.

[Sol]

What you don't know now, is what the gauge does if plugged into the sender, with the sender in your hand, moving the arm from empty, to 1/4, to 1/2 to 3/4 and then to full. The sender unit is maybe the whole problem. And you proved there is an earth involved.