PDA

View Full Version : Troubleshooting your fuel level sender


TitaniumTT
01-10-2014, 05:41 PM
I could find ZERO definitive info on this on the "other forum so I took a crack at it myself and decided to post up what I found in here.

First, a general understanding of the fuel system in the RX8.

As many of you know it is a saddle type fuel tank. This poses many problems for both fuel delivery and proper reporting of fuel level to the driver. There are two separate senders in each half of the tank. The fuel pump module is on the drivers side.

Lets start with an empty tank shall we? You're bone dry pulling up to the pump. Shut the car off and the first thing you do is put about a pint of premix in the tank :squint: Then you start pumping.... generally about 14 gallons from what I've experienced...

The drivers side section starts to fill first. As the drivers side section fills up, it starts to spill over into the passenger side section. You top off the tank and start driving. As your Renni consumes fuel its being pumped out of the tank regulated @ roughly 58psi through the drivers side. Now you may think you're draining the drivers side first, in reality though, the module is using vacuum created from the pump to pull fuel from the passenger side to keep the drivers side full, in theory. In reality though, and my friend Mike can attest to this, anything under a 1/4 of a tank is susceptible to fuel starvation. We both experienced this at Limerock Park last June.

So... the next logical question is, if the pump is emptying out the passenger side first, and you've got a 1/4 of a tank left on the gauge, that means the drivers side is half full and shouldn't go dry... right? Not to the best that I can tell. It appears that the gauge is smarter than we all think it is. From what I can tell from f*cking around with it for the last few hours, it's actually really clever. It seems that it will take an average of the two sending units and report this level to the driver, yet it seems to revert to the drivers side for the last half, and the passenger for the 1st half.

So, as your driving along the highway in a straight and level manner, the fuel will be pulled away from the passenger side and fill the drivers side. As the passenger side empties, the level in that side drops to zero. The drivers side is still full however. As the level in the drivers side begins to fall, the gauge reports this drop down to the point where the fuel light comes on and at this point you've got 30 miles of highway driving to find a gas station.

That's how it's supposed to work. What happens if something fails?

Lets say that the fuel pump is failing and isn't creating enough vacuum to pull fuel over from the passenger side to the drivers side. Well, this is where Mazda did a really good job at reporting a level to keep you from running out of gas. If the level in the passenger side remains full, yet the drivers side (where the pump is) starts to fall, the gauge responds accordingly. Of you're OCD like I am, you're constantly keeping track of fuel mileage. You may think you're getting the worst mileage you've ever seen, but when you get to the pump you get 6ish gallons in before it clicks off. Now you start thinking WTF? That's not right... psshhhh, derrrr I can get more in ... and away you pump until it overflows and now you're fuming mad. More than likely what happened here is the pump failed and the passenger side of the saddle is still full because the pump doesn't have the balls to pull the fuel over. Or the siphon tube is clogged. The test here is to make sure the siphon tube is still connected surely, there are no obstructions, no kinks etc etc. If that's the case, you need a new fuel pump.

Lets say now that you're cruising down the highway and you think you're getting the best mileage you've ever gotten on a road trip... the needle doesn't want to move from the 1/2 way mark and you're cruising along oblivious to the world and then all of a sudden the car stutters and dies... you've got ~270-300 miles on this tank of gas and you're wondering what's going on... you've got a 1/2 tank of gas and you're just getting 40mpg on this physics defying roadtrip through time and space.... no... you're out of gas and the drivers side fuel level sender puked. No a big deal at all actually, once you get the fuel pump pod out of the way all you have to do is stick your hand in there and rotate the sender 90* CCW and it'll pop right off. A beautiful design. Now, getting the pod out of the way is a ROYAL MOTHERF*CKING C*CKSUCKING C*NT of a job getting that ring to spin. Best approach I've found WITHOUT the special factory tool is a VERY blunt common screwdriver and a mallet. Whack it 2 or 3 times on one side, then rotate 180* and hit it 2 or three times on the other side. If it's the first time it's coming off, it'll take a good 10 minutes of slowly working it 90-180* before it will start to move by hand.

So, we know how it works, and how the system reports in normal and in a failing mode, you're just screwed like I was if the drivers side is the one that takes a dump. However, what about on the track like what Mike and I experienced. Well, anyone who knows LimeRock Park knows that the majority of the turns are right handers with one good lefthander. What I discovered playing around with both senders while my dead one decided to work for a few minutes, is that if the senders are moving opposite or if the passenger side is lifted slightly when the drivers is at rest, it moves by what is almost like a calculated value. So at the end of the day, the level in the tank is correct, but you don't know how much is in which side. Both Mike and I experience fuel starvation on the No-Name straight which is right after the left hander and we both had a 1/4 of a tank when it happened. My recommendation is to not run the stock RX8 below a 1/3 of a tank on the track unless you've got some sort of aftermkt surge tank inline... he and I are actually developing one ourselves.

Some troubleshooting tips -
If you think you're getting really shitty gas mileage and you can only get 6 or 7 gallons in the tank, more than likely your pump is failing.
If you can never get the needle to go above 1/2, it's probably the passenger side sender that puked.
If you think your mileage is defying the laws of rotary physics and not moving past a little under 1/2, it's more than likely the drivers sender that puked

It was infact my drivers side sender that puked. Me being me, even after draining the tank completely and playing with the senders, I decided to test the sender on the bench. Surprisingly it tested just fine. I know right? When I plugged it in and started playing with it though and watching the gauge, it would only go down as far as an 1/8 of a tank and not trigger the light. For a dealercost of $90 (no I don't go to the dealer :D ) I wanted to be sure. Just test twice and order once approach.

Anywho..... I hope that helps someone diagnose a fuel level issue as I could not find ANYTHING on the other site and since I'm anti-IB, here it will stay....

Cheers

-Brian

JL1RX7
01-10-2014, 07:46 PM
Thanks for the info! Finally getting around to this from the summer I see! I wondered how they read the saddle tank. Averaging sensor makes sense and relatively cheap.

TitaniumTT
01-10-2014, 08:02 PM
Yup.... the thing died Labor Day weekend and I've driven her maybe a handful of times when other cars have been a PITA or Maggie needed my Jeep for a week... aside from that she's been in "storage."

Thinking about how moving one lever independently of the other and going through the math in my head, yes, it's simple math and averaging. The thing to worry about though is when on the track, and when the drivers side pukes.

JL1RX7
01-10-2014, 08:13 PM
Hard to say if they are using the computer to do the averaging or if it is true resistance that doing the work. It's been awhile since I read a cut sheet on level sensor. Most of the time the ones I work with are high or low level limit switches. Next week when I am bored I will read up on them.

TitaniumTT
01-10-2014, 11:07 PM
I tired to find a wiring diagram of them online to get a better understanding of how they work but couldn't find one.... so I gave up and dove into it blind. From what I understand from reading one or two posts on the other site, there are two different inputs into the dash from the senders and the faggotry and witchcraft takes place in there....

RETed
01-10-2014, 11:14 PM
WTF!

I guess the "engineers" don't understand KISS???

Big band-aid to a stupid problem...
Can you put some kinda equalizer pipe between the two halves (run underneath) to minimize this headache?


-Ted

TitaniumTT
01-10-2014, 11:23 PM
Not really. I think it would be VERY susceptible to getting ripped off and would be very close to the exhaust., unless it's at the very bottom of the tank, it's never going to get every last drop of fuel out of the passenger side. I was actually impressed taking the passenger side apart. It was BONE dry... I mean BONE dry... kudos to them even though it sucks in its own right....

This is why my plate is NFURY8.... everything in that car relating to performance was pretty much a no compromise approach to make the car handle better, stop better, go as fast as it could. The suspension and drivetrain were designed with zero fucks given to future service or repairs. It's a nightmare to do ANTYTHING on an RX8, but fuck do they stick like mutant spiders dipped in glue....

On the otherhand, I could strip the interior with a plastic trim pry tool, #2 phillips and a 10, 12, 14mm socket in about 17 minutes and have everything in a pile next to the car... because interior trim doesn't effect performance.

GySgtFrank
01-11-2014, 07:23 AM
How hard would it be to put a decent sized fuel cell in there and do away with the split tank?

TitaniumTT
01-13-2014, 01:13 PM
You really wouldn't get more than 6 or 7 gallons on one side without severe chassis modifications. You've got the truck but that's certainly not ideal for weight placement.... which brings a decent point for me to remember... always fill the tank when corner balancing.... 1/2 tank will through everything off as all the fuel is on the drivers side...

The answer in my opinion would be a surge tank with an in-tank pump like I built for the fuel cell in my FC. This will allow you to keep the full quantity of what you've got in the tank, and give you the ability to run the tank virtually dry without risking fuel starvation. Like I said, Mike and I are working on something that will be featured in Bawbs FB as well as my 'vert.

The only mods you would need to make to the tank would be to add a return line.

Dannobre
01-13-2014, 03:30 PM
I've been everywhere down this road......and have tried so many different things over the years...

Few things...

If you go to a return system you had better make sure you are getting the fuel return to the fuel pump cup/siphon...or it will make things worse

The siphon will only work if there is a significant amount of fuel going through it...so if you increase your fuel requirements you will not get enough fuel transfer to get the fuel out of the passenger side.

If you put another pump in the passenger side...it will run dry a lot of the time on the street...not a good thing :) So you need a switch...and the siphon intact if you plan on street driving the car

Basically after a lot of experimenting....I now have a transfer pump in both sides of the tank...and a large surge tank in the trunk with a Bosch 044 pump. With 500+ Hp in a REW swap I have never had fuel starvation this way....down till the transfer pumps are both sucking air :)

Overly complicated...yes...but the only thing I have found that works without a hiccup

TitaniumTT
01-13-2014, 04:00 PM
Interesting... write up with pics somewhere? Also curious about the REW swap as well, how you got around various emmisions requirements and keeping the DSC and electric PS working....

Dannobre
01-14-2014, 01:16 AM
PS is not an issue...DSC on the other hand I disabled a long time ago :)

Guess i should do a build thread at some point....in process of doing a dry sump currently...so maybe when I do the install

Sinfidelity
07-12-2015, 12:41 PM
Hard to say if they are using the computer to do the averaging or if it is true resistance that doing the work. It's been awhile since I read a cut sheet on level sensor. Most of the time the ones I work with are high or low level limit switches. Next week when I am bored I will read up on them.

I've been taking readings all week from the right side sender in order to get the right sender in my new lpg tank.

The readings below.

Fuel gauge reading
Full 47ohm
3/4 210ohm
1/2 243ohm
1/4 326ohm
Empty 326ohm

So I'm guessing this is not gonna be accurate if rx8 somehow averages out from two senders?
I noticed the reading of 326 never changed between below half and empty !!? Drove me crazy
Maybe the two senders merge their wires parallel and cld be measured at that point?

Sinfidelity
07-13-2015, 09:49 AM
So today I followed the wiring back to cluster.
Turns out the two senders share the black/ brown trace and send the other two wires along with a single merged black/yellow-ish trace back to cluster.
So a total of 3wires go directly to cluste from the two senders.
I pulled the two plugs from a spare loom and twisted the unshared two wires (one from each sender) ( blue/red trace and white /green trace) together.
Filled the tank and measured impedience basically as parallel from both senders and got 9ohms which I believe what is stated in the SM.
ILL test at all the gauge increments to see if it seems consistent and post results.
I don't think the ecu is making any calculations what so ever as the car I traced these wires to cluster had no ecu attached.

RICE RACING
07-13-2015, 06:01 PM
good thread, should be more like this!

Sinfidelity
07-17-2015, 02:40 AM
Damn it's weird!

So I cut out the fuel sender plugs n cable from an old loom to use as a test cable enabling me to read my running car and test in parallel, solo and in series to try figure this out.

All seemed to be good till I hit half tank while testing both senders parallel
Full 7-9ohm
3/4 55ohm
1/2 94 ohm
1/2 142 ohm left side only

Not sure what to follow here seeing as the right side isn't gonna change now till empty.

JL1RX7
07-17-2015, 04:06 AM
Now since it is a averaging sensor I wonder if it only takes input from certain OHM readings. So say anything above 90 OHM it only reads from one side vs the other.

Most of the sensor's I use are for temperature across a large area. Not a split sensor system like this. But trying to see the logic of what they are trying to read, then using a sensor to read it. I could see it doing something like that.

Sinfidelity
07-20-2015, 09:29 AM
Now since it is a averaging sensor I wonder if it only takes input from certain OHM readings. So say anything above 90 OHM it only reads from one side vs the other.

Most of the sensor's I use are for temperature across a large area. Not a split sensor system like this. But trying to see the logic of what they are trying to read, then using a sensor to read it. I could see it doing something like that.

I think I figured it.
The gauge cuts the circuit to the right sender once it gets below half. Much like how it activates the fuel light at near empty.
Then only reads level from the left.

My final readings were this

Full 7-9 together
3/4 55 together
1/2 142 left side only
1/4 200 left only
E 326 left only

I'm still not convinced it's 100% accurate but I think it's close enough to program a new single sender

bleedingEdge
07-29-2017, 10:59 AM
Just want to relate my experience with the fuel level senders, for anyone else having this problem, and because this thread was the closest thing I could find from days of googling this.

Problem: fuel gauge doesn't fall below 1/2 - 1/3. Started abruptly, not going below 1/4 for a couple months, then got worse. 2004 RX8 MT, 260000 miles.

Cause: both fuel level senders' resistor cards became less accurate toward the bottom (tank empty) ends.

This really confounded me for a while because my brain really hates it when it seems like two distinctly different parts fail in exactly the same way at exactly the same time -- especially after so many years/miles. But it's true. I removed both senders and tested on the bench. They both returned 10Ω on the high (tank full) side, and 263Ω on the low (tank empty) side. Identical. It wasn't until I actually consulted the shop manual that I realized the low side is supposed to return ~325Ω.

I used a little pick tool and tiny wire brush to clean off the varnish from the contact strips, and sure enough, both senders low end results came back into spec. I have two new senders on order, but for now, these old ones are working fine again.

Moral of the story: there is a remote chance that, if your gauge isn't falling below 1/2, it could be both fuel senders are a little faulty.

On a side note, there is disappointingly little info about this coming back from my internet searching. I never found anyone discussing why this varnish thing matters or how to deal with it aside from replacing the part. I don't know much about electronics, but my guess is that the varnish is a pretty good conductor, therefore the more of it there is, the less the electricity is forced through the resisting bits, therefore the less resistance Ω gets added to the circuit?