pasadena_commut

About a month ago we changed the oil on my wife's 4 cyl 1998 Accord EX automatic. While doing that it was observed that the coolant overflow tank was empty, and no fluid was visible at the top of the radiator. No blue splotches to indicate a coolant leak. Filled the radiator to the top, pulled and cleaned the overflow bottle, marked it half way (marked with a Sharpie) and filled it to there. Not sure how much fluid was added, perhaps a quart or quart and a half. The level kept going down when I checked it every couple of days (the car isn't used every day). Eventually it looked like maybe the radiator had cracked, as there was fluid all along the top. Since it was the original radiator, and now 26 years old, it was replaced with the appropriate model Denso from Rock Auto, also new top and bottom hoses and a new Gates radiator cap. The only tricky part was that the cap that came with that radiator was marked 0.9 (13 psi), and its flanges were too tight - it was very, very hard to get it off the radiator. The original radiator cap was marked 1.1 (16 psi). Anyway, refilled the radiator until coolant was pouring out the bleed valve. Closed that, started the car and kept it running until the fans came on, topping off the radiator fluid as needed. Heat was on, fan control to max, car was oriented facing up a driveway, albeit one without much of a slope. Turned it off, filled overflow tank to middle, sealed it up. Done?

Kept filling it every couple of days as it fell an inch or so. Maybe still air coming out of the system? Coolant was always still visible in top of radiator. This went on, fluid level kept falling.

Time for better diagnostic tools. This weekend we picked up a Powerbuilt 22 part cooling system loaner from Autozone. With car cold (last driven one the day before) put the adapter on the radiator, pumped it up to 14 psi and waited 10 minutes. Nothing, no drips underneath, pressure rock steady. Waited another 10 minutes and still nothing. The pressure may have fallen 0.1 psi or so, maybe. Not sure I trust a loaner device like this not to leak 0.1 psi in 20 minutes. Tested the brand new Gates radiator cap and it was bad - it would never open, even up to 40 psi. Good thing we caught that, it could have resulted in a big mess, or worse. Tested the radiator cap from the old radiator and it was also bad in that it didn't seal. At this point it had been sitting dry in a hot garage for two weeks though, it might have sealed OK when it was fresh off the car. Even though it leaked it still opened correctly at 16 psi.

Not a drop on the ground. Nothing visible on the front radiator hoses or nearby. We looked and felt around on the back of the motor as best we could and could not locate any signs of coolant. No signs or smells of a coolant leak in the cabin from the heater core.

So maybe a very slightly blown head gasket, one that opens up at temperature?. One that also didn't leave a hint of oil in the coolant or coolant in the oil. Pulled the plugs and inspected. All the plugs looked exactly the same, like, well, normal used plugs. Poked a Klein tools bore scope into each cylinder and saw piston tops that looked like one would expect, a bit dirty, no signs of fluid. (Not the right kind of bore scope, this only looks straight down.) If there is a head gasket leak, it isn't much of one. Safe to start though, definitely not hydro locked. Put the plugs and wires back in and started it up. Absolutely normal start, no hesitation at all, no white smoke, no odd coolant smell.

Since the Gates radiator cap was bad today I went to the local Honda dealer and purchased a 19045-PAA-A01. Checked it with the cap tester, it sealed normally and opened normally at 16 psi. Installed that on the car. Drove the car around until the fans started, then pulled over and looked under the car. No drips. Drove the car 20 minutes, came back, left it running, looked under the car for a couple of minutes. No drips. None. Upper and lower hoses were at appropriate temperatures, car temperature was where it has always been for this vehicle under normal operating conditions. Also, to the best of my knowledge this car has never overheated. (It was the in-laws car before it was ours.)

The only motor related thing this car does oddly concerns the idle. If the car is rev'd to 3000 RPM (for instance), and then pressure removed entirely from the gas pedal, the RPM will fall past idle and it will stall. However, in normal driving the car never stalls. There is a thread for that here:

https://honda-tech.com/forums/honda-...talls-3366772/

I'm at a bit of a loss now as to what else to check for. This is what I came up with, any other suggestions?

Jack up the car again and crawl under to inspect behind the motor better. I didn't notice any signs of coolant leaks when I was back there changing the oil, and don't expect to find anything.

Check for OBD2 codes. The CEL isn't lit though - does a car this old have any if the CEL isn't on?

Maybe there is still air in the cooling system from the low coolant/coolant change? Seems like kind of a lot,of air but I suppose that is possible.

Start keeping track of how much fluid is being added.

Add fluorescent dye? Pretty sure we would have spotted the blue form the coolant though.

Any other suggestions?


Thanks.

LukeM

So, are you still losing coolant? The correct way to use the bleeder screw is to start the engine cold, crack it open, and let the coolant flow out a little while until the bubbles stop. That way any air bubbles in the system get circulated out. The best way to test for a blown head gasket is to get a remote starter tool($10 at a store). Pull the ecu fuse, remove the radiator cap, and crank over the engine. If there is a leak bubbles will appear in the coolant. Did you check the color of the oil? If the water pump seals are failing it will leak out of a weep hole on the rear driver side of the engine. Often the slow trickle will evaporate before it hits the ground and you'll only know it by the smell.

pasadena_commut

Not sure if it has stopped losing coolant yet, but probably not, as the only change to the system, despite all the recent testing, was to put on an OEM radiator cap to replace the one that would never open.

We have run the car with the radiator cap off for extended periods and not observed any bubbles. Is there some reason that wouldn't make the same, or more, bubbles than the method you suggest? I also didn't observe any bubbles when the car was started by my son.

If "evaporate before drip" is common then I guess adding fluorescent dye is the logical next step. At least after we give it a couple more days to see if by some miracle the new cap fixed it.

Thanks.

LukeM

The bleeder screw is the high point of the coolant system so if air is going to be trapped it will be there.

pasadena_commut

Put UV dye in it, drove around 20 minutes, cooled off, then looked for leaks. Nada. Everything in front of the motor is pristine, not a speck glows. Going under the car, nothing at the weep hole or around/under the intake manifold. Hard to see into that mess under the throttle body but there were no glowing spots in the parts which were visible. No random glowing spots behind the motor either. Neither the oil nor the transmission fluid have picked up any fluorescence.

Is there some relatively easy way to look into the intake manifold? At this point there are two paths for coolant to get into the cylinders: past the head gasket to the cylinders and past the intake manifold gasket to the manifold to the cylinders. Given the choice I would rather deal with the second one. A bore scope into the cylinder will show drops in both cases with the coolant pressurized. Not sure I could distinguish between them based on those observations, even though they would be coming from different locations.

Other than the coolant loss one would never know there was anything wrong with the car. (Engine wise at least. The paint is going and the sunroof leaks from a not very good roof repair job back when my in-laws owned it.)

Edit: modified second paragraph.