O-Ring Failure Information

Last updated: April 5, 2002

Date: Wed, 16 Feb 2000 22:02:30 -0500
From: "David Lane" (dlane@peabody.jhu.edu)

A Primer on O-rings

In response to a recent request for information about O-ring failures, and given the number of new members on the list, I offer the following general article on the subject. I do this with a nod towards Felix Miata, and to others who have covered it to one extent or another on their web sites. This version is a little more in depth, including diagnostics, preventive maintenance, and a section on limping the car to a shop.

The O-rings in question are the large ones that are responsible for providing the seal between the alternating steel and aluminum sections of the engine. There are two sets of O-rings at each section joint--an inner one and an outer one. These O-rings function to keep coolant flowing through the passages from section to section. The outer O-ring keeps coolant from leaking to the outside of the engine. The inner O-ring has a rougher life since it has coolant flowing on one side, and combustion gasses burning on the other. This is the O-ring that usually fails, and the one we will be talking about.

In theory, the sections of the engine are flat enough so that the O- rings compress in their grooves--mostly out of harms way. A typical O-ring failure is caused when the engine overheats., warping the alternating layers of aluminum and steel in the engine "sandwich" relative to each other. This small separation exposes the fragile O- ring to direct pressure and heat from combustion (on one side), and to coolant pressure on the other side. Thus exposed, the O-ring can't hold out for very long, and eventually it breaks (usually near the spark plugs). Once broken, coolant and combustion gasses can trade places. Symptoms vary according to the size of the break.


When the engine is running, combustion pressures far exceed pressure in the cooling system, so combustion gasses work their way into the coolant.

If damage to the O-ring is slight, you will get what looks like a slow, frustrating coolant leak somewhere. Every so often, your Low Coolant light will come on, but you won't be able to see evidence of an external leak. You may notice that for some reason you have more coolant in the overflow bottle than normal. If this occurs, first check your radiator cap for a tight seal. Then check your overflow tube for a split. Either situation can result in coolant coming to the overflow tank in the usual way, but air being sucked back into the radiator due to the air gap (either at the cap or in the hose). If the problem persists suspect an O-ring problem.

If the O-ring has a larger break, you will have a significant amount of combustion gasses trying to displace the coolant in the system. This will tend to push coolant out of the radiator, into the overflow tank, and possibly right out of the top. With the engine running (thermostat open) you will see a steady stream of bubbles coming from the submerged end of the overflow line. The resulting hydrocarbons can be detected by a smog sniffer.

With a really large break, combustion gasses will not be able to work their way out of the cooling system fast enough, and the system will tend to blow coolant out of the path of least resistance--usually the overflow bottle.


When the engine is shut down, combustion pressures go to zero, but the coolant remains pressurized for a time, causing it to push into the combustion chamber.

If damage to the O-ring is slight, you will probably not notice anything. Possibly, the break in the O-ring won't be great enough to pass coolant, or maybe it will tend to close itself up. Another possibility is that a small amount of coolant will get past the O-ring, but it will evaporate. Small O-ring leaks can be difficult to diagnose. One method is to get a radiator pressure tester. You may have to leave it on over night to see the drop.

A moderately damaged O-ring will allow more coolant to pass. If you shut down the engine for a short time (at a gas station, for instance) there may be enough coolant in the combustion chamber to create a rather large cloud of white smoke from the tail pipe when you restart the engine. You will notice a pronounced aroma of "Eau De Antifreeze" in the cloud.

A severe break in the O-ring will allow enough coolant into the combustion chamber to prevent the plug from firing until it burns off. You will probably see it and smell it if you remove the plug. Any problem which allows liquid into the rotor housing can cause the engine to seize, so a broken O-ring needs to be dealt with as soon as possible.


If you are fortunate enough to have a good rotary shop within a reasonable driving distance, and are phobic about having the car towed (and the related cost), you may want to try and drive the car there. Much depends on how badly the O-ring is compromised, but you can try the following technique for a short drive, and if it works, you should be able to get where you are going.

  1. Remove about a quarter inch of coolant from your radiator. Close the radiator cap tightly. This will leave room for a flow of combustion gasses to enter the cooling system, circulate, and exit through the overflow tank. Without this room, coolant will be forced out.

  2. Drain your overflow tank and leave the cap loose. The empty overflow tank will be able to accommodate a small build-up of coolant if it should get pushed out of the radiator. The loose cap will allow any excess to spill without building up pressure.

  3. Expect your Low Coolant warning to come on when you first start the car. As soon as the thermostat opens up, the entering bubbles will displace coolant and will raise the level in the radiator. The warning will go off.

  4. You may be tempted to leave the radiator cap loose. Don't. It works fine if you have a leaking hose on the outside of the engine and don't want the usual pressure in the cooling system to build up. However, with a bad O-ring, the pressure in the cooling system is the only force resisting the flow of combustion gasses through the break. If you leave the radiator cap loose, you will see something like "Old Faithful" occur as soon as the thermostat opens--causing rapid loss of coolant. Using these techniques, I was able to drive my car for a hundred miles to the shop. This was on Interstates, as well as on typical "urban sprawl" four lane roads. Even though I was carrying five gallons of spare water, I did not need it.


Replacing an O-ring requires disassembling the engine. And since at least some of the housings are likely to be warped, a rebuild is the only solution. The best strategy for mental health is to start thinking of the situation as an excuse to have your dream engine built for the car.

Early rotaries were more prone to O-ring failures than more current cars. As an early fix, Mazda offered O-ring "shields." I believe they function as a thin metallic strip that goes into the O-ring groove, and shields the rubbery O-rings from combustion temperatures, but I have not actually seen one. These shields were still available a few years ago, and you might consider them if your engine sees stresses beyond stock. Another option is to go with aftermarket O- rings. Hayes, for instance, makes some teflon/silicone versions.

Much like fancy apex seals, fancy O-rings offer additional protection. But no apex seal can survive repeated detonation, and no O-ring will last long in an engine with warped housings. In the end, the stock components are just fine if nothing goes wrong with the cooling system. This brings us to...


The best way to prevent O-ring failure is to prevent overheating. Keep your cooling system in top shape, and get an aftermarket water temperature gauge. Hoses are especially prone to failure. You may think you can keep an eye on the condition of the hoses, but it is harder to do this than you think. Water hoses tend to rot from the inside out, and it is hard to see them getting weak. By the time you get what looks like a pin-hole leak, there could be a three- inch crack on the inside of the hose--only held together by the thin outer skin.

In addition to the radiator and heater hoses, most rotaries run smaller coolant hoses to and from the intake section for various reasons. These must also be changed. How often should you replace coolant hoses? I have no data on that, but I would guess that 60,000 miles would assure safety, and 80,000 is probably on the high side. Certainly if you have a 1st gen or a 2nd gen, with no record of the hoses ever being changed, do it now. Change your belts at the same time. You don't want to find yourself without a working water pump.

Much has been written about stock Mazda water temperature gauges. In short, they suck. The gauges in first generation cars are slow to respond. The gauges in third generation cars come up to the mid point quickly and stick there through a range of about 35 degrees (F.), only budging if the temperature rises a bit above the figure at which the electric fans are supposed to cut in. This is only fifteen or twenty degrees below the point at which the engine can be damaged, so you get very little warning. I have less experience with second generation water temp gauges, but logic would suggest they are somewhere between the first and third gen gauges in usefulness. Such is "progress." An aftermarket gauge will show you a linear progression as the engine heats up, and you will know when to start watching things more closely. An aftermarket water temperature gauge should be considered an insurance policy and an early warning system for heat related problems..

The warping that exposes O-rings to trouble comes from the fact that the steel from which the intermediate and end housings are made expands and contracts differently from the alloy from which the rotor housings are made. As you can imagine, any rapid build-up of heat will cause the engine to pass through a period of dimensional change. For this reason, it is essential that you warm up your engine under light throttle, and at moderate revs. In addition to warming up the car's lubricants, it will allow the engine to stabilize before you ask it to put out full power. A short period of driving under similar conditions is an adequate cool-down technique for anything except track work of the kind that doesn't allow a cool-down lap. Only under those conditions do you need worry about letting the engine idle for a minute of two.


The information in this article is based on my experience with an O- ring failure, as well as many discussions on the net over the years. Some of the techniques mentioned are difficult with 3rd gens due to the overflow tank being hidden. Obviously, I can't take responsibility for what might happen to you or to your car if you set out on a long drive with bad O-rings using the suggestions above. They worked for me, but caution is indicated--and you might as well bring the phone number of a towing service if you set out that way.


Felix Miata's explanation of what O-rings are.


Date: Thu, 17 Jun 1999 09:44:42 +0000
From: "David Lane" (dlane@gigue.peabody.jhu.edu)

There are several symptoms of internal o-ring seals leaking. When one of these seals goes, the combustion chamber ahd coolant passages connect, so you get water in the combustion chamber, and combustion pressure in the cooling system.

When the car is running, there is high pressure in the combustion chambers, so gases migrate into the coolant and displace liquid. This can force coolant out of the overflow bottle. Symptoms are a full overflow bottle, and a mysterious loss of coolant. If you can see it on a late 2-gen (I don't have one), you will see a stream of bubbles exiting from the bottom of the tube that goes into the overflow bottle when the engine is idling warm with the thermostat open.

When the car is shut down, the combustion pressure drops to zero, but the coolant pressure stays up for awhile, pushing coolant into the combustion chambers. If you start the car before the coolant has a chance to evaporate, you will see white smoke which smells like antifreeze coming from the exhaust. In bad cases, a stop for fuel will result in a veritable fog of white smoke, and missing from one of the rotors until the coolant gets out of there.

Other tests for o-ring failure is to have someone put a smog sniffer over your open overflow bottle cap while the car is running to see if combustion gasses are present. You can also have a pressure test done on the radiator (hold pressure for a very long time to detect a tiny o-ring leak) and see if coolant finds its way into the combustion chambers. These tests will confirm a small break in the seal. The symptoms I mentioned earlier will alert you if/when the break in the o-ring is larger.

An engine rebuild is the only solution to an o-ring failure. The cause is most often overheating, which warps the alternate steel and aluminum sections of a rotary engine, exposing the o-rings to the heat of combustion.


Mike took a poll of people that have had their O-rings fail. Here are the results. --Steve

Date: Fri, 20 Feb 1998 02:49:11 -0800 From: Smeagol

I received 7 responses (including myself) to the oringinal inquiry concerning the water o-ring failure on the 3rd generation cars. Thanks for taking the time to reply with your individual information. Following are the results as best as I can summarize:

  1. year/model of FD?
	All were 93s: 
	93 R1		- 3
	93 base	- 1
	93 Touring	- 2
	93 ? 		- 1

  2. miles on engine at failure?

  3. original engine?
	- yes on all counts

  4. symptoms observed?
	- low coolant light
	- having to add coolant after driving (low coolant level)
	- difficulty starting
	- gurgling
	- overflowing tank
	- empty air sep. tank
	- rough running at first
	- wet spark plugs

  5. determined cause of failure?
	- increased pressure in cooling system
	- excess heat

  6. covered under warranty (extended)?
	not covered	- 4
	covered	- 3

  7. if so, what was covered?
	- new caps
	- engine & necessary parts for replacement, but not
	  hoses/belts or other "wear" items

  8. cost to you (if any)?
	$2500, used engine for $2500, parts - $600, $25 deductible,
	$4800 (included other work)

  9. how long did the work take?
	- 5 weeks
	- a few days (parts only)
	- 3 weeks (estimate)
	- without car for 3 months
	- 1 month

 10. if not covered, list options
	- Pettit rebuild
	- none unless paying out of own pocket

 11. did you rebuild?
	- Mazda remanufactured engine - 3
	- waiting to replace engine - 1
	- yes - 1
	- no - 2
 12. who did the rebuild work?
	- Rotary Performance, Pleasanton, CA
	- Mazda remanufactured - 3
13. cost?
	- approximately $3500 total for the extended warranty company
	- $2100 for engine

Date: Tue, 24 Feb 1998 08:54:39 -0500
From: Stephen Ziegler

(snip) If you want to learn more about what I am talking about, please visit Mazdatrix.

See http://www.mazdatrix.com/faq/oring.htm for the scoop on the O-ring.

Date: Tue, 24 Feb 1998 12:26:40 +0000
From: "David Lane"

Joe Ramos said:

I am unaware of any preventive measures. The failure occurs at the hottest point in the engine about 1 to 1.5 inches below the L2 plug. Not being an engine builder (I just destroy them), here are some GUESSES: Perhaps the water channels could be bigger? Maybe different plugs? Maybe lower boost?

In earlier engines (pre '83 12A's, I think), Mazda installed a protective shield on the o-ring which gave it added protection. Rick Soderberg who rebuilt my engine had a set of these in his parts bin which he installed. Mazda doesn't use them any more as they changed the composition of the o-ring to better with- stand the heat. This may have worked for the 13B/13BT, but certainly not for the 13B-REW.


At Joe's suggestion, I also had these old-style "o-ring protectors" installed when the engine was rebuilt on my car. I would guess Bret did not have them on hand, so they may still be available from Mazda.

Mazdatrix lists a procedure it can do when building an engine destined to host an aftermarket turbo. I am working from memory here, but I think they groove the water passages adjacent to the spark plugs to allow additional heat transfer. Again, if memory serves, Mazdatrix only offers this service on new rotor housings purchased from them. It is in their catalog. I have not checked the web site for it. Dave Lemon will have more information on the best applications for this mod.

Also, there is the previous discussion about the possibility that the main bolts holding the engine sandwich together may (for one reason or another) need to be re-torqued after X years or miles. The upshot of the previous list dialog on the subject was that Cam at Pettit seemed to think this was important, but other list members doubted it. Also, it is a nasty job which requires removal of the clutch housing. My own take on it is that any structure that is made of layers of aluminum and steel; which involves sealants, gaskets, and very long bolts holding it together, and which experiences numerous heat cycles over a long period of time simply has to change in some dimentional way. How much metal fatigue or minute looseness would be necessary to open enough of a gap for the o-rings to get exposed to direct heat from the combustion chambers?

Finally, when the Mazda rebuild that was on my car for a short time (and croaked do to o-ring failure) was torn down, the water passages were badly corroded, and the coolant was a rusty brown gook from the long period of time the car was inactive. It couldn't hurt to be sure to change coolant at the recommended intervals.

The problem here is the difficulty in establishing a cause and effect scenario. Our cars run the gammut from being stock to being highly modified--from being daily drivers only to being streetable track machines. To add to the confusion, any rotary can suffer a near instant O-ring failure as a result of overheating. Really getting to a cause/effect level with the O-ring situation would take a large sample and some serious delving into the circumstances for each individual car.

It seems that a prudent course of preventative actions would include:

  1. Periodic coolant changes.
  2. Periodic attention to all coolant hoses.
  3. Periodic replacement of radiator cap and thermostat
  4. Frequent oil changes (oil also cools the engine) goes without saying.
  5. Torquing down the main engine bolts if it ever becomes relatively convenient (maybe when the clutch goes).
  6. Installation of Trev's fancy O-rings, or at least old-style O-ring protectors if worse comes to worse.
  7. Opting for the Mazdatrix modified rotor housings if your engine sees exceptionally hard duty.

Date: Fri, 26 Jun 1998 11:10:58 +0000
From: "David Lane"

Symptoms for a bad rotor--apex seal:

1. Lumpy Idle

2. Buy a cheap compression tester. Pull out the little valve that holds the reading. Disable the fuel pump (pull the "engine" fuse) or do something else to keep the car from starting. Crank the car and watch the needle on the compression meter. High/Low/Low shows a bad apex seal (with the two adjacent rotor faces having low compression). High/High/Low shows a bad side seal (which only effects one rotor face).

Symptoms for bad water seal (o-rings):

1. Coolant gets into rotor chambers and you see white smoke on start-up--possibly with missing until the coolant burns off. You may also see coolant on the plugs.

2. Combustion gases get into cooling system, forcing coolant out of the radiator and into the overflow bottle--sometimes even past the overflow bottle onto the road. You get a low-coolant light, but see that the overflow bottle is full. At idle, with the engine warm (thermostat open) you can see a steady stream of bubbles coming from the overflow bottle.

3. Hold a cloth over the exhaust pipe and sniff it for the smell of coolant.

4. Have someone with a smog sniffer use it in the top of the radiator to see if hydrocarbons are present.

Note: On first or second gens, a swollen gasket on the radiator cap or a hole in the overflow hose can cause coolant to move from the radiator to the overflow bottle and not be able to get back. Double check that before sending your engine in for a rebuild if all you are experiencing is an unexplained movement of coolant in that direction.

O-ring Fix <-- page on how to fix this problem (besides the normal rebuild. This may be temporary, but people seem to have had good luck with it for a significant amount of time.

[Mail me] [To Lightning home] [To my home page] [Copyright Notice]