A driver of a waterboxer Vanagon had this complaint:
Driving along coolant light on dash starts flashing, pull over and check the pressure tank in the engine bay. It is visibly down by a lot, and the reserve tank behind the license plate door is overfull. Left it overnight to cool down. Next day the pressure tank level hasn’t risen. Siphoned some coolant from the reserve tank because it was too full and put in back in the pressure tank.
This same sequence happens over and over whenever I drive it for awhile. Had coolant system pressure-tested, test pressure never dropped. No external leaks can be found.
Could this be from a bad cap on the pressure tank?
When coolant keeps transferring to the reserve tank, two easy checks will distinguish a pressure cap failing to hold pressure from an early developing fire ring* leak:
1. If, after running at full operating temp and before cooling down, the reserve tank level has been seen to rise substantially, but the pressure tank is still full with liquid, it’s just the pressure cap failing to maintain the in-system pressure. It needs to hold pressure up to 1 bar or14.5psi. At normal temp the in-system pressure will be around 10psi, but there can be transients when high engine load causes momentary increases in coolant temp, and thereby pressure. You would not have a low coolant level warning in this instance while the engine is still hot because the liquid level in the tank is high. In this case coolant has escaped due solely to expansion, the system is still completely filled with liquid but the portion that couldn’t fit was released to the reserve tank.
If the pressure cap’s return valve works properly and the overflow line to the reserve tank is airtight, the coolant contracting as it cools off should draw almost the full volume back into the pressure tank. In that case, if you didn’t see the reserve tank full while the engine was still hot, you might not notice the signs of the pressure cap’s failure to hold the spec pressure.
You also might not notice the indicated temp increase because a lower system pressure means a lower boiling temp for the coolant, and it won’t get hotter than its boil point temp; adding more heat just causes it to boil more furiously, but the temp doesn’t rise. Too low or total lack of coolant system pressure can result in the situation where the indicated temp looks fine, even a bit lower than normal, while the cylinder heads can be overheating because coolant in contact is boiling. To absorb heat effectively liquid, not vapor, needs to be in direct contact with the hot metal surfaces. This puts the lie to the folks who claim they intentionally run their cooling system at zero pressure to lower stress on the system and it’s all OK because the temp gauge still reads normal. They forget that the temp gauge senses coolant temp, not cylinder head temp.
With a bad pressure cap, it is not uncommon, while idling hot, to see bubbles rising in the pressure tank; don’t be fooled, those aren’t necessarily combustion gas bubbles. If the cap is bad they are more likely steam bubbles due to the coolant spot-boiling in the cylinder heads because boiling is not being suppressed by maintaining a high in-system pressure, which is the cap’s responsibility. Most of the bubbles recondense to liquid once they detach from hot surfaces and move into the coolant flow, but some may survive long enough to be entrained to the pressure tank where they are visible. The vapor bubbles entrained toward the radiator will soon recondense to liquid coolant, which is why a bad pressure cap alone doesn’t cause gases to accumulate there.
The pressure cap doesn’t have to be failed wide open for this phenomenon to occur, I have seen bubbles rising in the pressure tank at hot idle with a cap that was still holding 6-7psi. Installing a cap that bench tested as holding the specified 1 bar made the bubbles stop. The same engine was leakdown tested as having excellent combustion containment. If it were a fire ring leak and it was able to produce gas bubbles at the very low in-cylinder pressures when idling, the leak would have to be a big one. Such a leak would be obvious on a leakdown test, and would be big enough to allow coolant to be forced into the cylinder after shutdown, which would cause the spark plug to look very clean, if one were to look, or could even push enough liquid in to cause hydro-lock on the next starting attempt. But such a large leak would almost never be seen in real life because long before it reached such a state the engine would have become practically unusable due to the dynamics explained in the next section.
2. If, after driving awhile at full operating temp and before cooling down, the reserve tank level has been seen to rise substantially and the pressure tank liquid level is visibly low (this is what triggers the Vanagon low coolant warning flashing LED, and on earlier vans the temp needle will also peg high), then remove the front grille and crack open the radiator bleeder bolt while the engine is still warm. If a lot of gas escapes before it comes out as steady liquid, that’s a strong sign that there is a high pressure fire ring leak. In this case the excess coolant was initially released from the pressure tank not only due to expansion but due to displacement as well. Combustion gases were forced past the fire ring at high cylinder pressures (large throttle openings).The overall system pressure increases due to the gases being injected at pressure, so while the pressure tank is full of liquid it is released by the pressure cap’s regulating valve to the reserve tank. Some of the gas bubbles were entrained with coolant flow to the radiator, where they accumulate at the top and displace liquid to the pressure tank, causing more to be released. Some of the gases also get entrained to the pressure tank. Once a bunch of liquid is released and more gases are accumulating in the pressure tank and radiator top, the pressure tank’s liquid level will lower because it is being depressed by the accumulating gases on top while the trapped gas volume elsewhere is compressed to accommodate it.
It’s at that point where the coolant level sensor detects the liquid level dropping and activates the flashing LED. The engine temp gauge would already be rising because by the time things get this bad the trapped gases in the radiator have interrupted the radiator pumping loop which must be filled with liquid for the water pump to induce flow, but it may not be noticed until the low coolant level warning LED gets the driver’s attention. At that point he’ll see that the temp is notably higher and rising fast on an ’86 and later van; on an earlier van the temp gauge would already have shot to the top of the scale along with the LED warning.
When you stop and note the low liquid level in the pressure tank and overflowing reserve tank, if you were to open the pressure cap while the engine is still warm the liquid in the pressure tank will rise immediately and may even surge out the top. This is because of the two trapped gas volumes at opposite ends of the system being pressurised and in equilibrium, and when you release the pressure at one end the gases in the radiator are able to expand and displace the liquid back the the pressure tank. This is a sure sign you are going to find a substantial volume of gases trapped in the radiator. But it’s better not to open the pressure cap, instead use the existing system pressure to force the gas out the radiator bleeder, and if you do see plenty of gas escape before liquid comes out, you know you have a fire ring leak.
An early-onset fire ring leak will not show up on any compression, cylinder leakdown, or cooling system pressure test for simple reasons. The fire ring only leaks, initially, when in-cylinder combustion pressures are very high. Those high pressures only occur at large throttle openings (when you vary the throttle opening with the accelerator pedal, what you are doing is regulating the mass of fuel/air charge that can enter the cylinders, which varies the end combustion pressure and thereby the amount of torque the engine produces). Full-load in-cylinder pressures are 800psi and higher, while low-load (part throttle) cylinder pressures may be only a few hundred psi. So, when the fire ring leak is just beginning to develop, gases only escape at high engine loads, when there is over 6 tons of force trying to lift the cylinder head, and very hot gas at over 800psi trying to find its way out.
A compression test will develop 190 psi at most. A leakdown test is using usually only 100psi test pressure. A coolant system pressure test is trying to force a rather viscous liquid (coolant) at only 15psi thru a gap that probably doesn’t even exist at that moment. But hot burning gases at 800psi can force their way out.
The hot gases burn away material as they escape, so once a fire ring leak begins to develop it will steadily worsen to where a leakdown test will eventually show it up, but by then the vehicle has become unusable because the time it can be driven before coolant flow is interrupted by the gas accumulation at the top of the radiator becomes shorter and shorter. Before you get around to testing, you’ll already know very well what the problem is. But you can diagnose it very early on by observing the simple checks I began with here.
So, now that you understand the physics involved, in summary:
With a bad pressure cap, after driving awhile and while the engine’s still hot you’ll see the reserve tank level rise while the pressure tank appears to still be full. You may see bubbles rising in the pressure tank. There won’t be gases trapped in the radiator.
With a developing fire ring leak,after driving awhile and while the engine’s still hot you’ll see the liquid level in the pressure tank drop and a commensurate rise in the reserve tank. You may see bubbles rising in the pressure tank. You will find gases trapped in the radiator. It’s also common to see gooey black residue on the inside of the upper part of the pressure tank.
*”Fire ring” is a common name for combustion seals, usually of multi-layer steel (MLS) construction, although some are copper or aluminum. These are integrated with laminated head gaskets on most engines, each cylinder will have its own fire ring integrated within the overall gasket. The rest of the gasket material seals all the other openings that go between the engine deck surface and the cylinder head, for coolant to circulate and for oil to be delivered to the cam and/or rockers and drain back to the sump. A waterboxer has individual, separate MLS rings that seal the top of each cylinder to the cylinder head. The water jacket seal, often called the “head gasket” is a separate seal that merely seals the gap between the top of the crankcase water jackets and the bottom surface of the cylinder heads. On a wbx, either can leak without affecting the other, although if a fire ring leak becomes extreme it can increase pressure in the water jacket so fast that the jacket seal ruptures.
Thank you for this, it is a perfect description of what I have experienced. I had a fire ring failure this year and got it all fixed about 600 miles ago.
Any thoughts on what one might see on a water pressure gauge that’s dash mounted ? I’ve recently seen mine creep to 20 psi while on the highway. It usually goes back down eventually whereas when the fire ring failure occured it would be always be 20psi while driving.
Hey Seth, thanks for stopping by.
Yeah I would think you would see coolant pressure elevated above normal, but it would depend on the severity of the leak, and the momentary engine load. If the leak was in the early, intermittent stage it’s possible you might be able to spot a pressure surge when you open the throttle and see it go back down at part throttle. That would be a useful diagnostic, but of course few people have a gauge for coolant pressure. I think the first sign most people get is the sudden cloud of steam behind when the reserve tank overflows onto the exhaust! unless they saw the blinking low-coolant level warning first.
Twenty psi all the time sounds like perhaps your p-cap isn’t opening, though, maybe you should test it, it should limit pressure to 1 bar. Could be gauge error as well. System pressure at light load with normal temp ought to be around 10psi.
I ran a shop from 1977 to 1998. At first we had few liquid cooled cars, we specialized in German stuff. As soon as the liquid cooled Rabbits, Jettas, etc started coming in we had problems with the cooling system caps. You probably know that the front engine car’s caps are the same as the Vanagon except for the hose fitting for the overflow. We got a Stant pressure tester and the VW adapter. We would order up 10 caps at a time and test them all as there were so many problems. Half or more would fail. Would not hold pressure or relief was too high(over 20 psi) or the return/low pressure part was bad. These were mostly from the VW dealer. After we got the tester and rejected the out of spec caps there were almost 0 problems.
BTW we were fixing the Vanagon head gaskets in early 1984. A quick analysis showed that the main problem was that the black rubber gasket was being eroded by corroded aluminum. Obviously the fix was to seal the gasket to the head so the coolant could not get in between and there would be near zero corrosion. We used “Ultra” silicone probably a year before VW started speccing that as a fix. After that the only time the gaskets failed was when there was an overheat usually due to a failed hose or part and a near empty system.
Soon after we encountered the “fire ring” failure. Our diagnosis was to put shop air pressure, 160 psi into the cylinders. Usually that would have bubbles in the expansion tank.
Many Vanagons that we fixed had gotten new radiators, water pumps, thermostats, etc before we fixed them.
Lucky for everyone the original Rimco was not far away and they could fix most of the corroded heads as at that time the only source was the dealer. Big $$$.
Thanks for the excellent summary of what’s going on with a fire ring leak.
In my case I can drive the van all day around town no problem, but on a trip climbing hills with heavy throttle the overflow tank fills and the coolant doesn’t return.
I’m not sure if the fire rings are passing pressure or possibly a crack in the head opening at heavy throttle.
I guess I’ll find out when I pull the heads.
If there is no sign of fire ring leakage I guess I’ll order a pair of heads from you.
Are all the fire ring gasket sets pretty equal- or does one stand out? (I’ve never pulled a VW engine before and would like to only do it once)
Thanks again for making things clear.
Hey Dan, thanks for stopping in. Yes, the signs you describe are what one would expect from a nascent fire ring leak. The clincher on diagnosing this is, after an episode like that, to check the radiator bleed to see if gases have accumulated there, it’s easiest to just crack open the bleeder bolt while the system’s still warm and pressurised and see if a bunch of bubbles come out before liquid flows.
A cracked head isn’t impossible and could act this way, but it’s far less likely than the fire ring.
There are basically two brands of wbx engine gasket sets as well as some of the individual seals and gaskets: Reinz (now part of Dana Corp.) and Elring. I’ve used plenty of both and there’s no significant differences, although I prefer the Reinz set for the Reinzosil sealant that’s included and a couple other minor things.
Before you get started, you should definitely read my article on how to help ensure this doesn’t happen again:
This was a great writeup thank you! I recently purchased an 85 with stock engine in Colorado and tried driving it back to the east coast. I made it 6 hours into Nebraska with no problems. The next morning I got the flashing led and the temp needle started to rise but never topped out. The overflow tank was overflowing, but I couldn’t tell anything with the pressure tank as it’s black. Needless to say I ended up towing it back to PA. When I drove it back to my house I got the needle starting to rise a lot but no led light. Pulled over about 5 times to cool it off and limp it home. When I got home the overflow was back to normal. I found a pressure cap test on YouTube which showed a guy blowing into the tube on the cap and it made a duck call noise. He showed a bad cap that wouldn’t make the noise and one that would. My cap doesn’t make the noise. Could this be my problem or is it the fire ring?? Here’s the link to the video: https://youtu.be/RT8Ew43mNg4
Thanks for your comment, I’m glad this article helped you understand better what might be happening with your engine. Note that the key test to ascertain a fire ring leak as opposed to a bad coolant pressure cap is finding gases accumulating in the radiator after one of the overheating episodes.
I’m not here to do individual advice or analysis, but I have another article that addresses the video your refer to, this will help you understand the limited value of what was shown there:
Thank you very much for your prompt reply. I’m glad you’ve done an article on that video. I was hoping to start with the cheapest options first, so if the new cap doesn’t fix it I know now where to go. Thanks for writing these.
Very interesting article.
I had new pistons and barrels and head put on my 91 Vanagon recently. Since the work my O2 sensor is failing.
I put a new sensor on and within 2 weeks the car is bogging down with an over rich mixture and belching black smoke. Unplug the O2 sensor and its OK. Put another O2 sensor on and the cycle repeats.
Could a fire ring failure be contaminating the O2 with coolant? Or doesn’t the coolant get into the combustion chamber with a fire ring failure?
No, the 4th and 5th paragraphs of section 2 explain why that won’t happen while a fire ring leak is still small and intermittent.
Excellent write up! Your description of the issue at the start of the article is precisely what is happening with my 1991 Syncro.
I have four tank caps that all (save one) fail at restoring cold coolant to the main pressure tank. I am hopeful that the issue is limited to the cap only, and the fact that they show this fault is an indication that they suck and I just have to find a good one. I do have one that allows the return of cold coolant, my hope is that it suffers from an inability to hold hot coolant pressures. It would appear from your wisdom that the best way to diagnose is by cracking the bleeder screw.
I happen to also have your heavy duty heater valve installed with the bleeder tube that connects via a banjo fitting to the bleeder bolt. Would the clear tube be a diagnostic tool? For example, if it is empty, does that mean it is full of gas and cracking the bolt is not necessary? If no, do I crack the bleed bolt despite the banjo fitting, and behave as if it is not there?