Recalibrating the fuel gauge using a Spiyda Gauge Wizard

[John Vine]

Thanks, JK.

Your question is a lot more tricky than perhaps you'd imagined!  I think that, without the wizard, my bending would have produced these readings:

• Full tank (24 ohms) between 1/2 and 3/4 on the gauge
• Empty tank (163) about 1/4 on the gauge

JV

[Ivaan]

I've tie wrapped my float to the arm. If they come off, fishing them out from the tank is a PITA.

[Jonathan Kay]

Your question is a lot more tricky than perhaps you'd imagined!

 

:-)

 

I'd imagined that you'd plotted the response characteristics of the instrument...

 

Jonathan

[John Vine]

I'd imagined that you'd plotted the response characteristics of the instrument...

I'm not sure I understand what you mean, Jonathan.  I thought maybe you were asking "what would the gauge now display in normal driving without the Wizard connected?"

Anyway, on the basis that a picture is worth a thousand words, here are a couple of plots:

1. Before adjusting the float arm:

2. After adjusting the float arm:

You'll see that the precision in the 6-14 litre range has improved dramatically.

I now need to reset the Wizard calibration points, and then do a (prolonged) road test.

JV

[Mark w]

Really useful info John .

Thanks for taking the trouble to post all of this .

I am at Graph 2 stage ! 

From my crude observations in use the second plot seems to mirror my gauge response following float arm bend . 

 

[Garth]

Very informative thanks John

[John Vine]

Well, I've just been testing the Wizard set-up in anger, and encountered a truly bizarre situation.

The tank contained about 32 litres.  The sender resistance was 9 ohms (see Chart 2  in post #29), and the gauge pointed firmly at "F" (I had already adjusted the needle to that position using the Wizard).  I turned on the ignition, and the resistance and gauge remained unchanged.  But when I started the engine, the gauge dropped immediately to just over 3/4 and the resistance increased to 73 ohms.

Thinking that maybe I'd affected the Wizard in some way by taking its feed off the gauge 12v switched feed, I made a new switched connection off the washer/wiper feed.  But the same thing happened.  I then went for a 30-mile run (using, say, 4.5 litres).  With the engine running, the sender resistance had risen slightly to 75 ohms and the gauge had dropped to just below 3/4.  With the engine off, the resistance was 15 (corresponding to about 27.5 litres).

The relative changes in gauge and resistance with the engine off were as expected, but can anyone suggest what was going on with the engine running?

JV

[bjw]

Hi John,

In post #19 you wrote:  To measure sender resistances, touch the multimeter probes to the green/black and black gauge wires (or, if you've already hooked up the Wizard, to the green and black terminals in the Wizard connector block).

It's unlikely that you will obtain a correct measurement of the sender resistance with it connected to the Wizard even with power off,  but with power on you certainly won't.

How were your setup values measured?

Cheers,

Barry

[John Vine]

It's unlikely that you will obtain a correct measurement of the sender resistance with it connected to the Wizard even with power off,  but with power on you certainly won't.

Hi Barry,

Yes, I agree.

The Wizard comes in two parts: (a) the unit itself with circuit board and edge connector, and (b) the connector block, with fly-leads.  The connector block slides onto the edge connector.  

Apologies, but I'm afraid my comments were a tad ambiguous.  I should have made it clear that, when I said "if you've already hooked up the Wizard", I meant "if you have already connected the fly-leads to the car's wiring".  I didn't mean that the Wizard unit itself should be plugged in!  (Indeed, if it were, it would be impossible to make contact with the fly-leads with the multimeter probes.)

I measured my resistance readings (without anything else connected, and with the ignition off) in two ways: initially, across the car's sender wire (green/black) and chassis earth, and later (after I'd wired in the connector block) across the green and black fly-leads of the connector block (which were in turn connected directly to the car's sender wire and chassis earth, respectively). 

These amount to the same thing, I think.  At least, I hope so.  (Once I had wired in the fly-leads, the second method was much easier as the terminals in the block were very easy to probe.)

JV

[bjw]

Ah, I see.

So when you say: >>But when I started the engine, the gauge dropped immediately to just over 3/4 and the resistance increased to 73 ohms.<<

You do actually mean the resistance of the disconnected sensor changes ????  That surely can't be right ???

Cheers,

Barry

[John Vine]

Yes, that's exactly what I mean, which is why I thought it bizarre. I plan to try a couple of things:

1. Observe how the gauge and resistances behave when I disconnect the Wizard completely and revert to the standard wiring connections - the only change then being the bent float arm.
2. Beg or borrow another gauge and see how that compares.

One possible factor in all this is that Caerbont had previously recalibrated the gauge (see the link in post #1). This was unsuccessful but maybe because of the weird behaviour I noted yesterday.

JV

[bjw]

1: The two major electrical differences between engine off and engine running are the increased system voltage, and additional noise on the supply. The wizard seems to be a well established product so noise really shouldn't be an issue.

2: A small spurious +ve voltage developed at the 0v terminal of the sender would produce an apparent increase in sender resistance.

It's not terribly obvious however how #1 might result in #2 !!

Just to eliminate the possibility I would check to see if there's any voltage on the disconnected sensor wire with the engine running.

Cheers

[Geoff Brown]

I considered 'buggering' around with the float arm but in the end I went the 'suicide, fuel anxiety route' & literally let the car run out of fuel & calibrated by watching the gauge move as I slowly added fuel.

This was done very locally so I had no reason to worry. So now I know that when my fuel gauge reaches zero, & you have to monitor for this, that I have approximately 22 miles or approximately 4 litres of fuel remaining if I drive 'carefully'. A tad difficult in a R400D ! The 4 litre 'addition' works all the way from zero to half. Which is OK for mentally working out the expected range.

At the other end of the scale when the tank is topped to the very brim, fuel lapping at the top of the neck, the needle takes ages to drop. In fact at least forty miles worth of fuel. So the half to full calibration is not so important.

[revilla]

Is it possible that the fuel return flow to the tank is directly hitting the float and displacing it? Or even the new float position allowing it to be "sucked down" by the pump?

[John Vine]

Barry: Thanks for the suggestion.  I'll investigate.

Geoff:  I guess your R400D is similar to mine, but does it have the 1/4-sweep gauges or the newer 3/4-sweep ones?  My fuel gauge (2008, 1/4-sweep) has never given accurate readings. With the tank full it reads 3/4, and when down to around 8 litres it reads minus 1/4.  If you have the 1/4-sweep gauge and fancied playing with your multimeter, I'd be very interested to know the sender resistance at Full, with and without the engine running.

Andrew:  The Duratec doesn't have a fuel return line (unlike your K).  When I took my "bizarre" readings, the pump wasn't running anyway (having already primed the system).

One thing I didn't add before:  the 73 ohm value wasn't a rock-steady reading (unlike the 9) but seemed to oscillate by +/- 1 ohm or so around the 73 mark.

More news later.

JV

[John Vine]

An update...

Just to eliminate the possibility I would check to see if there's any voltage on the disconnected sensor wire with the engine running.

You were right, Barry. There was a small voltage on the disconnected sender wire when the engine was running:

 

Observe how the gauge and resistances behave when I disconnect the Wizard completely and revert to the standard wiring connections - the only change then being the bent float arm.

This was interesting.  With around 28 litres in the tank (sufficiently full to get close to a full gauge reading -- see Chart 2 above), I got the following gauge responses:

where 1 = "full".  So, it looks as though I'm getting at least a realistic reading at the upper end of the scale (compared to the 3/4 I was getting before). 

My next step is to record afresh the sender resistances per litre added, but this time including gauge readings with the standard wiring set-up.

JV

 

[bjw]

I think the first priority would be to eliminate the spurious voltage. This is of course speculation, but it could well be a grounding problem, where there's a common between the sender ground and something else drawing power which together are not making good contact with the chassis.

The obvious nearby candidate is the fuel pump, which I had dismissed thinking it ran all the time the ignition was on. (I have Webers and forgot that on injection cars things are different...) I believe your fuel pump runs for a short time to prime the system then turns off until cranking is detected ?

Your measurement with 'ignition on' would then be with the fuel pump off, and with 'engine running' with the fuel pump on ?

The way the wizard measures the resistance is likely to use less voltage across the sender than the conventional system, so may well be much more sensitive to any spurious voltage, hence the consistency in readings when you reverted.

Cheers

Barry

[John Vine]

I believe your fuel pump runs for a short time to prime the system then turns off until cranking is detected ?

Correct.

Your measurement with 'ignition on' would then be with the fuel pump off,

Yes, once the system was primed (just a couple of seconds).

and with 'engine running' with the fuel pump on ?

Yes, I guess so, but I don't know whether the pump runs continuously or just when the pressure drops below a threshold.

I'm taking a series of more detailed measurements.  I'll post them here when they're complete.  Meanwhile, I'll give some thought to possible earthing faults.

JV

[Jonathan Kay]

I thought maybe you were asking "what would the gauge now display in normal driving without the Wizard connected?"

I was, and whether the improved accuracy at low levels would do the job? But now it's got much more interesting... looking forward to what you discover.

Jonathan

[John Vine]

Well, I’ve now taken a more detailed set of readings, and very illuminating they’ve proved to be.

In short, it appears that simply bending the float arm has corrected my gauge readings, to such a point that I’m now very confident that I’ll be able to monitor levels much more accurately between 1/4 and Empty.  This appears to have rendered the Spiyda Wizard superfluous as far as my particular car/gauge is concerned, although I’m still looking at ways to implement a low-fuel warning lamp. 

The gauge readings are nice and linear too:

Notes on readings:

1. All readings taken with the Wizard disconnected
2. Direct:  the sender wire is not connected to the gauge
3. G to S:  the sender wire is connected to the gauge
4. RP: Rest Position of needle (far left, beyond “R”)
5. OC?: possibly an Open Circuit?
6. G reading (G to S): measured in fractions from the LH edge of “R” (= empty) to “1” (see photo of gauge below)
7. G reading (G to S%):  fractions converted to %
8. Engine running: I didn’t attempt to take these readings until there was sufficient fuel in the tank (7 litres)

Two things still puzzle me:

• The apparent open circuit (OC?) with IGN on when I tried to measure the resistance of the sender (across the gauge earth and sender terminals) with the sender wire connected to the gauge.
• The readings for the same but without IGN on.  The values drop steadily but seem to have no relationship with the empirical values in Col 2.  I’ve no idea what these values represent.

JV

[revilla]

John,

I have to admit I was getting rather confused trying to follow the odd readings you were getting, but assuming I'm understanding the concise summary at the end of post #45 correctly:

You can't really measure the resistance of the sender with the gauge connected. The gauge itself has a resistance which is in parallel with the sender, so you are measuring the resistance of the parallel combination of sender and gauge. Worse still, the gauge has a power supply. Multimeters don't really measure resistance, they apply a small known voltage across the terminals being tested and measure the current which flows (or alternatively, they feed a small known current into the circuit from one terminal to the other and measure the voltage it takes to do so). To avoid damaging components, these voltages and currents are normally very small. If you connect the meter to a circuit which has external power supplies applied, the voltages and currents applied by the meter will in all likeliness be completely swamped by the voltages applied, leading to meaningless readings. The voltage across the terminals could easily even be of the opposite polarity to the current being supplied by the meter, which in theory would lead to a negative resistance display; I doubt the meter is going to do this, so you would probably see open circuit indications or something when the meter is just not able to make sense of what it is seeing.

Andrew

[John Vine]

Hi Andrew,

That makes a lot of sense -- thanks.  I have to admit that my approach was to measure anything and everything just in case, and even then only to avoid having to repeat tank draining and refilling!  For me, the key measurements are Col 1, Col 7 and last.

But I'm left wondering what actual resistance values the gauge is receiving when the ignition is on?  Clearly, the gauge is reacting to something, but what?

I'm also wondering if I've tried to calibrate the Wizard the wrong way.  The instructions give three methods (paraphrased here) applicable to oddly-shaped tanks like ours:

1. Set each of the five individual calibration points (E, 1/4, 1/2, 3/4, F) one by one, by filling the tank to the required level. This is the most accurate method. Combine it with method 2 and measure the resistances with a multimeter for each calibration point in case for some reason you need to reprogram the inputs or make an error.
2. Fill the tank as before, but this time measure the resistance with a multimeter.
3. Enter each of the five individual calibration values by entering their resistance in Ohms.

Method 1 didn't allow for the volume of fuel sitting below the pump/sender (about six litres in my case).  So, I followed #2 and then #3 instead, but measured the resistance of the unconnected sender.  Perhaps I should have measured the resistance with the sender connected to the gauge (Col 3 in my chart)?

JV

[revilla]

You need to measure the resistance values of the sender alone, disconnected from everything else (not just with the ignition off). This will give you the correct resistances. The actual resistances will then be exactly the same with the gauge connected and with the ignition on, none of that should have any effect on the actual resistances; it just isn't valid to try to measure with a multimeter under those circumstances. So the resistances the gauge responds to when connected are the ones you measure with it disconnected.

[bjw]

Andrew,

Note that in post #41 that John is seeing a voltage on the sender when it's disconnected and the engine is running. Even if the wizard is abandoned this is a significantly peculiar observation that I think should be investigated in case it's an indication of another problem elsewhere. Maybe I'm too fussy, but...

(To summarise - that voltage appears to be what caused the gauge to read low only with the engine running following the initial calibration. It's backing off the wizard measuring voltage and making the resistance appear higher than it is, hence reducing the gauge reading.)

In the last set of data direct measurements of the disconnected sender are still showing higher with the engine running.

Cheers

Barry

[revilla]

Barry,

Yes I did say I was getting a bit lost!

I admit that looks a bit odd and needs explaining. Sorry if this has been covered already, I flew back in from two weeks in Indonesia on Saturday night and in totally jetlagged, my body is telling me it's 3am...

Do the fuel pump and sender have completely separate wiring or do they share an earth? I guess even if they have separate wires from the unit they will share an earth at some point.

I suspect your post #42 and John's reply in post #43 cover it. The fuel pump will prime for a short time to pressurise the rail then stop until the ECU detects that the engine is turning, then it will run continuously (if it works the same as a K). The pump could be drawing anything up to say 10A. If that is giving rise to a voltage of 0.14V (as I understand from post #41), that would imply a resistance in the earth line of only 14 milliohms, which is tiny. 2mm2 / 25A thin wall cable has a resistance of 9.42 milliohms per metre so it's only equivalent to the resistance of about 1.5m of cable . As you increase the engine speed, you will increase the MAP and this is coupled to the fuel pressure regulator to increase the fuel rail pressure to keep the same pressure differential across the injectors (again assuming it works the same as a K), so the pump will work a little harder, draw a little more current and the voltage seen will increase a bit. You are getting 0.15V at 2000rpm.

You could confirm this my measuring the voltage on the earth connection at the pump relative to the battery -ve. I suspect you will see the small additional voltage on the earth.

Good practice would be to separate high current feeds and low level signals so that they don't flow in the same wires, so the sender and pump would be wires separately right back to the battery. In practice I doubt it is wires like that, and probably uses a single connection to the chassis for both?

Cheers,

Andrew