The new ECU is here – a whole new world! Temporary wiring was added so that it can be powered up on the bench in order to learn it before it goes into the car, since it has to be powered up (ignition on) to do much of anything. There’s also time needed to upgrade the MAP and O2 sensors along with their wiring, along with relocating the manifold vacuum block off the KPro ECU. The idea is to get the ECU configured before it ever sees the car, then swap out the appropriate sensors right before integrating everything. That way the car can be driven right up until then (though even now it always comes down to: do I drive the car today or work on the next upgrade?). Once the switch happens, the plan (hah) is to get it idling decent so that everything can be checked out and the car driven onto the trailer for tuning. I don’t want to arrive there, find something wrong, then get turned away. When this happens will happen depends somewhat on the next track event, which depends on when we’re ready, equipment-wise, money-wise, and mental-wise.
A heavy-duty relay was added near the fuel pump, along with a direct connection to the alternator through #10 wire – there won’t be much voltage drop on this setup.
A short tube was welded across the GM fuel composition sensor – pictures when I get around to it. Curious if I’d hurt the sensor by welding to it, it was temporarily wired up to see if it was working. Somewhat surprisingly, all the online wiring diagrams for the sensor were wrong, but since the +12V supply feeds the center pin on the 3-pin connector, it didn’t blow it up. Connected correctly, it output a 55-Hz square wave, indicating that the fuel contained 5% ethanol – cool.
Because the fancy capacitive fuel level sensor won’t work with ethanol, it was replaced by an old-school float type sensor. Not wanting to mess with converting the sensor current to a voltage for the dash, a corresponding old-school gas gauge will take the place of the water/methanol flow meter, which is coming out in anticipation of the ethanol setup.
A month ago Midlana ran Pomona speedway without a hitch. At the end of the day it was driven onto the trailer, then backed off the trailer and into the garage, where it sat for a month while the new fuel system was fabricated. With the fuel system finally done, 6 gallons of gas was added (no leaks!) and the car taken out for a test drive…. and it ran like crap!
As the car limped home, it would run fine at idle and up to about 2000 rpm, but then die back to idle before speeding up again. Back home, fuel pump voltage was confirmed good, but the worry was that there might be something wrong with the new pump, a hose, or the filter, meaning that the tank had to come out. The “good news” was that the problem was easily producible; if the gas pedal was floored in neutral, engine rpm would shoot up to about 2000 rpm, then immediately die, coasting down to maybe 500 rpm, then repeat. If it was a fuel flow problem I’d expect the issue to be load-sensitive – that is, in neutral when it consumes very little fuel, it should rev to as high as desired, yet the cut-off speed didn’t seem to care whether the engine was loaded or not.
Fuel pressure was monitored during the oscillations and was fine, which made things even more mysterious. Connected the laptop to check for ECU errors and there were none – but suddenly, there was the issue. The throttle position sensor (TPS) output should increment smoothly between 0 – 100% as the pedal moves to the floor. It worked fine – up to about 25%, then jumped around erratically. This failure wasn’t unexpected, but it happening now was – what did a gas tank replacement have to do with the TPS failing? Apparently nothing.
The Honda K-series TPS isn’t very reliable; what makes it worse is that it’s part of the throttle body assembly – you can’t buy just the sensor from Honda. Out of this arose a business opportunity, and K-Tuned sells a Hall Effect TPS sensor. The cool thing is that there are no sliding contacts like inside the OEM sensor, so theoretically it should last forever. Because of numerous Honda forum posts about bad throttle sensors, it seemed like it was only a matter of time before mine failed as well, and so it had. Fortunately I’d ordered the K-Tuned part last Fall – and today it started its new job.
The next step is a biggie, switching to the new ECU.
Over the last week or so, “tank V2.0” was fabricated.
Reasons to redo the tank include: wanting to run ethanol, an incompatible fuel level sensor, an incompatible external fuel pump, the new ethanol pump is in-tank only, and the fuel tank sealant swells when exposed to ethanol. More subjective was that tank V1.0 had several weeping leaks, never enough to drip, but an annoyance. Also, the tank had warped badly during welding, something that only bothered me, but just knowing that it’s there, saying “you can do better” helped me decide to just that.
The pictures pretty much spell out how it went together. I splurged and went for “real” fuel tank doors just to see what you get. These are the low-end units (“only” $30 each!). They reminded me a lot of the covers on outdoor electrical outlets – in fact I wonder if those could have been used instead, except anything aluminum is a no-no for ethanol.
The layout’s pretty obvious, the dividers doing double duty of keeping the fuel load from sloshing side-to-side, and guiding it into the accumulator. The forward section of the tank makes due with one fuel door, such that if there’s cornering with any acceleration, the fuel will always enter the accumulator.
As can be seen, the water jet parts came out really nice – they provide a nice finishing touch to the tank plus providing a truly flat surface for the gasket to seat against.
The Walbro “E85 400 LPH” pump is only available as an in-tank unit, necessitating a mount, hose, and power, with everything sealed to keep fuel from leaking up past the threaded fittings. Stat-o-seals were used on the larger fittings and Honda-Bond gasket sealant was dabbed onto the screw threads. The fuel fittings shown are -8 for fuel supply, -6 fuel return, -3 vent, and a -8 fuel vapor return which vents vapor to the top of the filler hose. This was added because tank V1.0 constantly caused gas station fuel nozzles to trip off – very annoying. The idea is that fuel vapor flows up the vent hose to above the nozzle instead of causing vapor to rush back up against the incoming fuel, causing it to splash on the trip mechanism on the filler nozzle.
The 45-degree red and blue fitting that doesn’t go anywhere near the pump is actually the fuel return. Without it, returning fuel would drop all the way to the bottom of the accumulator, introducing air bubbles into the fuel right near the fuel pump inlet. The 45 degree coupler points the fuel at the wall of the tank so that it smoothly runs down the side – or that’s the idea.
The sheet metal shop was unable to close the sheet, meaning three long weld seams. Fearing the extreme warping that happened with tank V1.0, it was welded in 1″ increments, moving back and forth, side to side, and was rewarded with virtually no warping. Seen in the pictures below is the comparison of the tanks V1.0 and V2.0. Tank V1.0 warped, bending the top 4″ section of 0.065″ stainless on edge nearly 1.5″ over 44″. I can hardly imagine the enormous force needed to bend a 4″ thick section of stainless! Tank V2.0 shows practically no warping at all – very happy about that.
The last picture is a GM flex-fuel sensor. It’s a pretty cool device that outputs a duty cycle measuring the exact amount of ethanol in the fuel, while the frequency measures fuel temperature. Knowing the ethanol content of the fuel allows the ECU to adjust timing, fuel, and boost instantaneously. The sensor gets plumbed into the fuel return line and while the ports are fairly small, reader “JR” recommended running fuel through the sensor, but also around it, so that 80% of the fuel takes the easy way around while 20% actually flows through the sensor.
If there’s ever a tank V3.0, the dividers will be welded to both the top and bottom skins, like tank V1.0 was. The need for this was made clear during leak testing where the tank was pressurized, with a distinct “Bong” and seeing the sides of the tank physically balloon slightly. Speaking of leak testing, yes there were leaks, but not as many as before. Also, this time I put on my old-man glasses, really strong reading glasses which allow me to get about 2″ away from the welds. Between that and a flashlight, questionable spots were touched up, while soapy water handled finding all the obvious leaks. While the tank was leak-checked three times, I admit that I wonder if it’s still going to leak. I suppose I could fill it with water and let it sit, but I think air is more able to get through really small holes than water is. Someone suggested pressurizing the tank and seeing if it holds pressure. Yes, but if the pressure drops, then what? So there’s a leak, but how can it be found? Soapy water and air pressure, so that’s what I went with right from the start.