What does "% leak down" really mean?

[ewave]

When you do a leak down test, I know you apply pressure to a cylinder through the spark pug hole, and you compare the ratio of two pressures. There's something that doesn't make sense to me about this.

Do you apply say 100 psi, and then wait a certain amount of time, and then read the pressure remaining in the cylinder? Or, is there a standard size orifice that has a given pressure drop for a given amount of airflow? Is this really standardized? And what actual size is the orifice? Also, while it's very easy to convert to % if you use 100 psi, how are the ratios affected if you use higher psi, like 150 lbs? Could a car have only 3% leak down at 100 psi, but say 6% at 150 psi?

I have my own leak down gage, and I've done my own measurements, but I have always really wondered about how standard these tests really are...

[Rot 911]

You are comparing the air pressure being applied to the cylinder and the pressure the cylinder will hold that is being applied. For example your input pressure is 100 lbs to the cylinder, but the cylinder pressure is reading 95 lbs, 5% of the air is escaping, hence a 5% leakdown. You keep the input pressure constant. Most people use a lower pressure because 100 psi tends to move the piston.

[ewave]

You need gas movement to get the pressure drop. 5 psi of pressure drop would be moving an enormous amount of air through a spark plug hole. It would move a tiny amount of air through a pin hole. I get that air is leaking out of the cylinder, but the pressure drop will be affected by how fast the air is being replaced.

[speeder]

Originally posted by Kurt V
Most people use a lower pressure because 100 psi tends to move the piston.

I always thought that you needed to lock the crankshaft, (ie. put the car in gear and set e-brake), in order to do this test(?)

[ewave]

Originally posted by speeder
I always thought that you needed to lock the crankshaft, (ie. put the car in gear and set e-brake), in order to do this test(?)

No: As long as you have the piston at Top Dead Center, it doesn't go anywhere.

[Rot 911]

Originally posted by speeder
I always thought that you needed to lock the crankshaft, (ie. put the car in gear and set e-brake), in order to do this test(?)

Or just use less pressure. Keep in mind the cylinder needs to be at TDC to keep it from moving.

[Rot 911]

Originally posted by ewave
No: As long as you have the piston at Top Dead Center, it doesn't go anywhere.

On the contrary, even at TDC 100 psi is enough to get a 3.2 engine piston moving. Ask me how I know!

[ewave]

If it moved, I'd guess you didn't have it at TDC. I had no problem when I did my test on a 3.2L engine at 100 psi. Both with engine out of car, in car, and in car when hot.

[Rot 911]

Originally posted by ewave
If it moved, I'd guess you didn't have it at TDC. I had no problem when I did my test on a 3.2L engine at 100 psi. Both with engine out of car, in car, and in car when hot.

Gee you are probably right.

[Grady Clay]

Paul,

Cylinder leak test is a relative measurement. In other words
it is valid when comparing one cylinder to another on the
same engine, with the same gauge, with the same supply
pressure, etc. With a consistent gauge, that relative
measurement can be extended to other engines. When
using various different gauges, the absolute measurement
becomes more subjective.

I am a proponent of regular measurements, sort of like keeping
track of your blood pressure, cholesterol, and PSA. It is a
trend you are looking for and a sudden change is a red flag.
We did cranking compression and cylinder leak measurements
as part of most 911 6k maintenances and tracked the numbers.

Zero percent is the same for all gauges. Six percent might
vary +/- 20% of reading from one gauge brand to another.
Twelve percent - possibly more variation.

I have two identical Sun Electric gauges. They were very
expensive, about $600 in the late ’60 and early ‘70s. With
care they vary less than 0.2% measurement when measuring
leakage in the 0.0-4.0% range, even today. I built calibration
jets for them at 2.0%, 4.0% and 10% using modified Weber idle
jets. We used filtered and well regulated supply air for the
tests. I felt it necessary to go to these lengths because we and
our customers used to rely on these numbers for very important
decisions about the condition of engines. This is why I am also a
proponent of not relying on a single measurement without some
spirited driving in between. I didn’t want any questionable
numbers.

Do a Pelican search on this subject; there is a LOT of
valuable information.

Here is the tester I use.



Here is the sparlplug adapter I made.




Best,
Grady

[air-cool-me]

at what point is the % leakdown a problem, if all the #'s are about equal?

8%? 10% 12?

[ewave]

Grady,

Thanks!!! That's the explanation I was looking for. I bought a cheap tester, about $120, and I got 0% leak down with the engine tested at 100 psi hot and cold. While it made me feel really good, it really didn't seem like reality. This is what the unit looks like... I can't remember the exact manufacturer.



I'm going to guess that the unit is missing the high precision orifice, and therefor its readings are of no value.

Thanks again for the very useful reply!
Paul

[Grady Clay]

With my gauges:
We never saw less than 1.0% and less than 2.0% on very rare occasions.
A normally good engine was 2.0% to 4.0% and would usually come down to 2.0% with spirited driving and an immediate test.
Many normally driven street engines would test in the 2% to 6% range but would return to 2-4% after driving. Occasionally there was a 12% cylinder that returned to 2% after just starting and raping it a few times – a piece of carbon under the exhaust valve.

Engines that start to show wear may have 4-12%. Some 150k engines that have significant wear can still measure 6%.

Cylinders with 12% to 20% you need to find out why.

If you find significant variation among cylinders after several tests and interspersed with spirited driving; that is an issue for concern.

The beauty of the cylinder leak test is that if there is a problem, it tells you where it is. You can tell if the air is leaking past the rings, valves, head gasket, and more. Some of the causes can be worn or broken rings, broken or pulled head studs, burnt or worn valves, too tight valve adjustment, cracked head or cylinder, and other.


Again:
Never rely on a single test unless there are other major indications.
Cylinder leak tests are part of an overall analysis of an engine.
Multiple tests over time set a baseline.
This is a BIG DEAL measurement and diagnostic tool.

Best,
Grady

[ChrisBennet]

Originally posted by ewave
Grady,

Thanks!!! That's the explanation I was looking for. I bought a cheap tester, about $120, and I got 0% leak down with the engine tested at 100 psi hot and cold. While it made me feel really good, it really didn't seem like reality. This is what the unit looks like... I can't remember the exact manufacturer.



I'm going to guess that the unit is missing the high precision orifice, and therefor its readings are of no value.

Thanks again for the very useful reply!
Paul

This is going to sound nuts but is there any chance you accidently used the hose from your compression tester by accident?
The hoses for my compression tester and my leakdown tester are physically interchangable but they readings won't be right (I tried it once.) My compression tester hose has some sort of check valve in it.

I've never had a problem with 100psi turning the motor. I usually use 80psi I can test longer before my air compressor kicks in again.
-Chris

[RoninLB]

Originally posted by ewave

I bought a cheap tester, about $120, and I got 0% leak down with the engine tested at 100 psi hot and cold.

I'm going to guess that the unit is missing the high precision orifice, and therefor its readings are of no value.

mine is also inexpensive and I go thru the 100psi routine like you. It was explained a few months ago that 100psi was too high to use. I think the gentleman said 30psi, but I registered 50psi for my next test.. as per my un-precision gauges and ease of quickly figuring #'s.

my low precision readings are plenty fine for my routine. My engine see's one gauge. If my needle returns to the same numerical over time I assume things haven't changed. All analog gauges that I know of have different variable precision % as it moves from the center reading.. ie: if I were looking to dial in 10psi, I would have more confidence in a 0-20psi ga than a 0-200psi ga.

The beauty of Grady's post is the way creditibility becomes a concern when dealing with very serious projects.. or understanding the background of the numbers being fed to you by others.

So anyway.. I think your ga set is fine.

[konish]

ewave,
I may be wrong, but there is not controlled leak through any kind of orfice designed into the tester. The "leak" comes from the rings, intake valve seat or exhaust valve seat. In a perfect world, a cylinder may theorectially experience "0%" leak-down. I would think the test would be invalidated if there was indeed a controlled leak through an orfice as it may never yield where the problem exists...it would ALWAYS leak down faster at the orfice than anywhere else unless something was worng with your motor.

Wrong?

R/
Dustin

[Early_S_Man]

This is a great thread, but there are a couple of misconceptions still floating around ...

There is no controlled leak!

The 'main' gauge reading the percent value [whether giving direct numbers counting 'up' in leakdown or down in percent held from input] ... is reading a differential pressure across a specific orfice size, and the dual-gauge set illustrated above DOES have the orfice and should not be assumed to be a cheap or untrustworthy gauge, unless damaged or giviing non-repeatable results. If there is no airflow into the cyliunder being tested ... there is no differential pressure, and the reading will be 0% Leakdown or 100% retained pressure!

A cheap gauge would be a $30 item from China of obvious shoddy construction ...

Many suppliers have advertised $80 - $120 2-gauge leakdown testers like the one above ... in Autoweek over the past 25 years, and I have no reason to doubt their quality, nor have I ever heard any complaints about their performance or quality!

Grady's points about keeping records and watching for trends or changes on a specific engine ... measured with one specific gauge is a very good one. Just such a policy is the only one that makes much sense!

Leakdown numbers tossed around here from one single PPI test are essentially meaningless ... unless the shop is known and trusted. The idea of declaring an engine healthy and 'good' just because someone throws out a number from an unobserved test at an unkown shop is pure lunacy!!! Yet they are tossed around and touted here as the 'Gold Standard' ... by people that don't know the compression numbers, or whether the test was done hot or cold! If a technician reports 7% or worse leakdown and doesn't report where the leakage was traced down and heard on each cylinder ... you aren't getting your money's worth -- or a quality test!

[DavidI]

My wife constantly asks why I go on this site so much if I am not working on a project on the P-car. I told her that there is so much information that it is like going to Porsche School of mechanics with a bunch of friends.

This is another fine example of an educational thread. Thanks Grady and Warren!

David

[konish]

Grady & Warren,
Thanks so much for clearing that up!

David,
I agree completely...this place is simply awesome!

R/
Dustin

[Zeke]

Pardon my ignorance for asking, but why must the piston be at TDC?

If I understand the purpose of a leakdown test on a 4 cycle engine (we used to do leakdowns on 2 cycle kart engines regularly because a leak in the case caused a leaning out of the motor and either a burned piston or a seized one) is to show if one or more of three things is not performing. The 2 valves and the ring set. If you know wyou are on the compression stroke, why can't the test be performed at various points of the piston travel until the exhaust valve begins to open?

Or course, this would necessitate holding the crankshaft to prevent rotation. Is this the only reason that the piston is preferred to be at TDC? What if the rings don't want to hold half way down? Isn't that cause for concern performance wise?