Fuel sender issue and resolution for erratic gauge behavior

[ULTM8Z]

I'm posting this here since this and and Pro-Touring.com are ground zero for swapping in EFI and aftermarket EFI fuel tanks. Figure the LSX swap forum here might be the best spot too in that regard. While my frame of reference is a 1971 Camaro, these EFI tanks often come with similar universal fuel sending units, so what I learned may help other people struggling against twitchy fuel gauge needles, or other erratic fuel gauge behavior.


Bottom line up front: I just resolved a ~10 year long issue with erratic fuel gauge behavior and the root cause was extremely subtle.

For reference, my fuel gauge is an Autometer Sport Comp II, full sweep electric (which I just found out from Autometer has a built-in buffer circuit to prevent rapid reaction to every little movement of the float arm).

The fuel sender that came with my Tanks EFI setup looks like a generic style universal sender that several companies sell.

https://www.summitracing.com/parts/cin-sn35

My post is deliberately lengthy since I want to fully describe the issue in case anyone suspects they have a similar problem.


Symptoms - Gauge needle position highly unreliable at fuel levels below about 3/4 tank. Needle would shift position for no apparent reason. There's the normal needle movement during fuel sloshing typical for these older cars, but this wasn't that case. The needle would move and then settle into a new position. It always seemed like the float arm was getting stuck, but out of the 4-5 times I dropped the tank over the last 10 years to try to replicate the problem I could never do it. I'd get the sender out on the bench, and the float arm would behave normally. Put it back in the tank and the problem would reappear. Looking inside the tank, there was clearly nothing obstructing the arm or anything.



Early troubleshooting – There are basically three things that can cause the problem. The gauge, the vehicle wiring, or the sending unit.

I did everything I could verifying the wiring the sending unit, the gauge etc. I’ve had the sending unit on the bench with an ohmmeter verifying smooth electrical operation. I even had the gauge out of the car and hooked directly to the sender in the tank to bypass the vehicle wiring.

With the sender in the tank, I could replicate the problem with the gauge hooked to the sender. Slosh the fuel around by rocking the car back and forth. Needle would move and settle into a new position.


Later I found if I even banged my fist on the tank floor, the needle would move (but always to a lower position). That was my biggest clue that the float arm was getting hung up, as banging the floor seemingly dislodged it. I could also replicate the behavior with the ohmmeter, which seemed to exonerate the gauge as well.


But when I pulled the sender unit out and tested in open air, I could never for the life of me replicate the problem, nor could I get the float arm to get stuck. I’d then give up and put the tank back in the car and live with it.


Fast forward to this weekend - This last time I dropped the tank, I swore I wasn't putting it back up until I was able to replicate the problem and fix it. I even bought a new sending unit.


With the tank out of the car and on the floor, I was again able to replicate the erratic behavior. But also once again, with the sender out of the tank it appeared to behave perfectly normally.


I drained the tank and was adding fuel back in gallon by gallon trying to figure out what was going on.



At one point, I got so frustrated, I took the sender out and just sat down staring at it and thinking. But while I was doing that, I was subconsciously moving the float arm around, only this time with very small movements. After about 30 seconds of this, I started feeling something like metal parts galling against each other.


That’s when it finally dawned on me that I found the problem. I set the sender in the bench vice and replicated that, and lo and behold I finally FINALLY got the arm to get stuck in open air. It’s almost like the skys opened up and Heavenly light shined down with a choir singing… lol.

I then grabbed the old sender and got the same behavior. That told me I found a design error on this thing.


As shown in the attached pdf…. Turns out the float arm is connected to a brass cylinder. That brass cylinder rotates against a steel cover plate on the rheostat. With large movements, it works fine. But if you subject it to small movements over a period of time, the friction between the brass and steel seems to cause galling and sticky behavior. The internal spring seems to apply just enough force to jam the brass cylinder into the steel.

So you figure while driving, you’re going to get a lot of small float arm movements as the fuel moves and sloshes, which then causes the galling issue. Banging on the tank with my fist is evidently enough to free up the cylinder allowing the float arm to return to normal.

The solution was to put an intermediate material between the cylinder and the plate. I had some .010” thick copper sheet, so I generated a shim. With the copper in there, the operation is silky smooth and you can’t get it to gall.


What compounds the problem is that for the dimensions of a 70-73 fuel tank, you have to trim the float arm length considerably. This reduces the amount of mass acting against the brass cylinder pivot point and it no longer has the leverage to free itself on its own (or keep from getting stuck in the first place).


After A LOT of “bench testing”, filling and emptying 10 gallons (one gallong a time), filling and emptying, forcibly sloshing the fuel around, all while watching the fuel gauge positioned next to me, etc… I was able to prove to myself that I had enough confidence to put the tank back up.

This also allowed me to precisely characterize what each gauge needle position meant in terms of gallons in the tank. The Harbor Freight electric fluid pump was probably one of the best ~$15 I ever spent… lol.


With five gallons added back to tank, I drove the car around…. And I’ll be damned if the thing isn’t rock solid now. Despite the dampening circuit, you still get the needle following the fuel level movement if you’re steady-state going up or down a hill, but that’s typical. But the dampening circuit does prevent the gauge from reacting to transient conditions like short term acceleration/decelerations, going around turns, etc.


The important thing is that once you’re back to level ground, the needle always returns to the original position. The problem always got worse with lower fuel level, so good behavior with only 5 gallons was proof in the pudding that this issue is finally resolved.


The only unknown right now is the durability of the copper washer. I used what I had on hand. But, if it wears out or something, it’s not a big deal, I can get some other more durable material later (dropping the tank on the Camaro isn’t a huge deal thankfully).


Any rate, I’m going around posting this in the various forums I belong too, since during my troubleshooting effort, my internet searches were finding other people having fuel gauge issues too. This sender appears to be universally available, and both of my senders (old and new) exhibit the same behavior, so this is clearly a design issue).


I sent the pdf to Tanks and Autometer (who were helping me in the diagnostics). Figure they can alert their fuel sender supplier to the problem.

Hopefully this helps some folks.

[dhutton]

I stopped using that style of sender years ago and only use the tube style senders which have worked well for me. It will be interesting to see if they fix the design of those senders. They are made in China and I have doubts.

[ULTM8Z]

I'm going to attempt to take this to the next level...

I think I have a poor man's solution to "averaging" that late models use and it's diabolically simple.

I found that if I disconnect the signal wire the gauge goes to "Empty".

However, if I disconnect the ground from the sending unit, the gauge holds its position, and it returns to normal function when re-connected.

To test that, I jacked the back end of the car up, initialized the gauge, then disconnected the ground. I then re-connected the ground and lowered the car, at which point the gauge returned to the straight-and-level reading.

What this means is I can have the gauge stop reading the fuel level when going up or downhill and then start reading it again when back on straight and level ground. Automter's built-in buffer circuit is already enough to prevent the gauge from reacting to short term transients.

So all I need to do is interrupt the sender ground with a tilt-switch. I can double stick tape the switch to the top of the fuel tank (using heavy duty tape) and the orientation of the switch follows the orientation of the tank.

Jacking up the back end of the car with a level, looks like a 5° angle is worth about a 1/8 tank of needle movement. A 10° angle is worth about 1/4 tank.

Although in real world driving, the response time of the switch is a lot faster than the change in terrain and the resulting gauge reaction (which by Autometer's buffer circuit design, is extremely slow).

So it may be the case that the switch detects the change in vehicle angle and deactivates the gauge before the gauge can fully react to a particular off-level angle.

They sell bi-directional programmable versions of tilt switches, so I may get one of those and experiment with different settings. The temp ratings are very high too, so I can probably safely mount on top of the fuel tank.