Fuel gauge adapter

[jelliott]

That's excellent. (Especially considering that the battery configuration could change in the future, e.g. lithium-ion batteries that would run more like 13.0 V, or a regulated DC-DC converter supplying the 12 V bus.)

(Hopefully) one final question: As far as the filtering requirements are concerned, I saw a bunch of different capacitors mentioned in the datasheet, and you said 10 µF tantalum or 22 otherwise; the closest I see to that in the datasheet I have is '2.2 μF tantalum or 10 μF aluminum electrolytic capacitor is needed if leads to the LED supply are 6" or longer.' And because I'm fairly ignorant of capacitive filtering in general, are these minimum values, or specific requirements? (I don't have a 10 µF tantalum capacitor on hand, so I'm curious if I could substitute a larger one.)

Merry Christmas!

 

[alec_t]

Sorry, I was working from memory (not very reliable these days!). You can stick with the values mentioned in the datasheet. I would regard those as a minimum; higher values should be better. I know the circuit here isn't using LEDs on long leads, but any spurious voltages on the supply line or other common line are best avoided (and a car's electrics are likely to generate all sorts of spikes).

 

[jelliott]

Does this capacitor need to go between the chip power pins as you suggested earlier, or between diode power ("Bus1" in your schematic) and chip ground, as shown on the datasheet figure that included the "2.2 μF tantalum or 10 μF aluminum electrolytic capacitor is needed" guidance? Or both?

 

[alec_t]

It goes between the chip power pins, close to the pins, so that it can 'soak up' any nasty spikes before they reach the chip. Each chip needs its own cap.

Edit: It would be a good idea to add a similar cap from Bus1 to ground.

 

[jelliott]

I got this circuit built and installed in the car, but it appears as though only the second chip is working. The gauge reads a little above 50% (I run out of adjustment at that point, not physically on the R28 trimmer, the gauge reading just stops increasing beyond a certain point), and only the diodes going to the second chip are turned on (and their respective resistors show 12.5 V across them). The resistors going to the first chip show 6 to 8 V across them, and their respective diodes are reverse-biased accordingly. Any thoughts as to what kind of design or assembly errors might lead to this particular failure mode? I guess once the battery is 50% depleted I'll be able to ascertain whether the first chip is inoperative or just inadvertently in "dot mode" rather than "bar mode."

Happy new year (which I guess is just a few hours away for you) and thanks again!

 

[alec_t]

Sounds like the second IC is doing its job fine, but the problem lies with the first. One possibility is that R1 is the wrong value (too high), which would reduce the fixed current per stage in the first circuit. Check its value; also that there are no dry joints at its terminals. Double-check that pins 2,4,5,7,8 are wired as per the schematic (N.B. they are connected differently for the two ICs), and that the two lines labelled 'Bus1' are joined.

 

[jelliott]

I was able to reach pin 9 with the voltmeter probe with the PCB installed in the car, so I know it's not inadvertently in "dot mode." Hopefully I'll find time later today to take the board out of the car and check the continuity of all the solder joints that could affect the first IC. In the meantime, here are two photos I took of the PCB before I installed it in the car; maybe there's some error that will be immediately apparent to you that I haven't noticed in the course of checking, double-checking, and triple-checking my layout vs. your schematic. If it looks like the second IC is missing its +12 V power source, that's because I ran out of room on the 2" x 3" board, so power for the second chip is provided by a copper bus bar that runs across the two mounting standoffs upon installation. Also note that external connections #2, 3, and 8 (counting from the top) and associated components (2 resistors, 1 diode, 1 transistor) are not part of this circuit (they're for a battery-related warning light on the instrument panel) and that ground/earth is provided on both external connections #1 and 5.

 

[alec_t]

Good pics. It may be a trick of the light, or my poor colour vision, but the third band on R1 (bottom left, pic 1) is ambiguous to me. Is it orange (10k), or dirty yellow (100k)? Looking at pic 2 the solder joint at the right hand (grounded) end of R1 seems dry. Pins 2,4,5,7,8 and the Bus1 lines are confirmed wired as per the schematic.

 

[jelliott]

I still haven't pulled the board out of the car, but I did verify that R1 is a 10k ohm (orange) resistor and that it has good continuity to ground. I obviously can't measure the resistance when it's soldered to everything else, but there's 1.25 V across it when the circuit is powered up, if that tells you anything.

 

[alec_t]

1.25V across R1 is correct. But it would be that regardless of the R1 value. Can you measure the Bus1 voltage? It should be <1V when the gauge is at max.
Simulation shows that at gauge max only the 5 or 6 top steps (all in the second IC) have their diodes forward-biased, and the voltages across the first and second IC's resistors are broadly in agreement with what you measured, but ~ 10% higher. So both ICs seem to be working ok.
You could try increasing the value of R4.

 

[jelliott]

Bus1 voltage is 0.84 V. I think I bought extra resistors in case R4 needed to be adjusted up; maybe I'll try that tomorrow. Is the voltage divider arrangement of R3 and R4 such that I could easily achieve the same result by touching another resistor across R3 for testing purposes?

 

[alec_t]

Scrub the suggestion to increase R4 for now. If the circuit is still in the car, can you :-
1) confirm that the Bus1 voltage is the same (+- a few mV) as the gauge input terminal voltage (which should verify if the current amplifier and drive transistor parts of the circuit are ok)?
2) apply different known voltages (say 1,2,...5) to the circuit input and measure the resulting voltages at either Bus1 or the gauge input ?

If the above checks don't reveal any surprises then it's possible the gauge, in reality, doesn't match our simple 2-resistor model, so we need to verify the circuit performance out of the car and then go from there.
With the circuit on the bench :-
Power it from a stable 12V (or even 9V?) supply of >100mA capability.
Do test 2 above, using a dummy gauge consisting of two resistors (2W rating) as per the model (or at least one resistor of 120-220 Ohm in the R25 position).

 

[alec_t]

Our posts seem to have crossed in cyberspace.
Yes, you could just put another resistor across R3 for test purposes, but having seen your posted measurement I don't think that will do what's needed .
Your measured Bus1 0.84V is a pretty good match to the simulation's 0.76-0.84V (depending on temperature). It's beginning to look as though the circuit is doing what it was designed to do but the 2-resistor gauge model is suspect. Tests 1 and 2 above, with the circuit in-car, might give some clues.
A puzzle is that with Bus1 at 0.84V (where it should be) the gauge input (if the current amp is ok) is pulled almost as low as it can be without actually being grounded and yet the gauge only reads 50%.

 

[alec_t]

Do you happen to know (a) the resistance value of the original car's fuel sender unit, and (b) whether the sender unit connected between V+ and the gauge or (as we've assumed) between ground and the gauge?
This site gives an indication of some sender resistances, but doesn't cover European cars:-
**broken link removed**

 

[jelliott]

Gauge terminal voltage does not match Bus1. The former was 4.0 V.

With the battery monitor disconnected, and 3.0 V applied in its place, gauge terminal voltage was 4.6 V and the gauge showed slightly less than 50%. 1.5 V didn't deflect the gauge at all, and 6.0 V produced the exact same (slightly > 50%) deflection at the 5 V input from the battery monitor.

I don't have any firsthand knowledge of the Renault fuel sender unit, as the car was actually purchased from the Renault factory without engine or fuel system, and sold new as an EV by an American company in 1981 (see https://fr.wikipedia.org/wiki/Renault_5#Aux_.C3.89tats-Unis as there's no mention of it on English-language Wikipedia). However, Renault wiring diagrams verify that the original sender would have been connected between the gauge and ground. When I had the instrument panel out of the car over the summer I meant to test whether the gauge could be driven 'backwards' by applying a 7 to 12 V input instead of 5 to 0, but if I did, I never wrote down the results.

 

[alec_t]

Gauge terminal voltage does not match Bus1. The former was 4.0 V. With the battery monitor disconnected, and 3.0 V applied in its place, gauge terminal voltage was 4.6 V

That points to a fault with the current amplifier. Simulation can replicate your results if R29 goes high value (>~40k) or if Q2 is leaky (<~500 Ohm). So I suggest check R29 and its connections (resistance in circuit should measure <=2k7), and replace Q2 (any small pnp should do).

Renault wiring diagrams verify that the original sender would have been connected between the gauge and ground.

Useful info. Confirms our assumption. At least we won't have to do major circuit redesign ;-)

 

[jelliott]

So I suggest check R29 and its connections (resistance in circuit should measure <=2k7), and replace Q2 (any small pnp should do).

Resistance across R29 (specifically, from the +12 V bus to Q2's emitter) is 2.7k ohms. Unfortunately, I don't think I have any PNP transistors on hand, and stores are closed today for the New Year holiday. I will, however, still be here (visiting my family, where the car is) tomorrow and we should be able to get a new transistor from a local electronics store.

 

[jelliott]

Addendum: Touching another resistor (10 kΩ) across R29 does increase the gauge deflection, from roughly 60% to roughly 70%. Does that tell you anything new? Might it be possible to work around a leaky Q2 by fine tuning R29 to make the gauge read correctly?

 

[alec_t]

Hmmm. That suggests another possibility: Q3 could be low gain (I can't readily simulate that) and needs a greater base current than it's now getting. By all means try reducing R29 as you've done. Its value isn't critical (and doesn't require other resistor changes), but I wouldn't go below 1k (to avoid excessive Q2 base current).

 

[jelliott]

Success! Thanks again.