Compression test vs Leak down test

[safe]

I'm chasing down a problem with my engine getting flooded from time to time when I run at very light load.
I'm pretty sure its fuel injection related, but I thought it might be prudent to check that it's not something mechanical.

So I started with a compression test (cold engine):
1: 150 psi
2: 155 psi
3: 160 psi
4: 160 psi
5: 155 psi
6: 170 psi

I thought that was a ok, within 13%, but not really that conclusive. So I got a cheap leak down tester, like the ones you can find on ebay around $50.

I tested at 100 psi and got very good results, 3-4% on number 1 and 4 an slightly less on the other cylinders.
Its suspiciously good... I could hear some leakage, more on 1 and 4, but on some cylinders almost nothing could be heard.

Could I have made some error during the test?
Never done one before and the instructions wasn't very clear....

I noticed that when I connected the tester and turned up the pressure the leakage was highly dependent on the pressure. The higher the pressure the lower the leakage, is that normal? The instruction said to use a pressure between 55 an 150 psi, so I used 100 and I'm not sure how I could have gone above that then that maxes the gauge on the tester...

This is number 1 for example:

[prschmn]

A leakdown tester should have a .040 orifice and be run at 80 PSI---an FAA standard.
So when you turn the pressure up the leakdown will seem to improve.

[Walt Fricke]

Mark - I'm not quite following this. I can see how, with the second gauge marked in percentages, it would have this effect. But my inexpensive gauge has both marked in PSI. I like to use 100 for the inlet because it makes figuring the percentages easier - 98 on the outlet = 2% leakage. So how would my end results vary if I used 80 psi instead of 100, and did the math.

Previous discussions of this suggested that while the FAA has a standard, there really isn't any standard in the automotive world (making our leakdowns less reliable)as the orifice diameter (and length) may vary.

In my case, I have noticed that the two gauges don't zero the same - 100 psi in looks more like 101 psi out with nothing connected. But this is easy to interpolate mentally.

I do think your two tests probably indicate that the issue you are dealing with is not related to worn rings or valves which don't seal. Maybe some adjustment in the EFI map might be in order?

Literally flooded, as in raw gasoline spitting out the exhaust? Engine dies and the plugs are wet with gasoline? Or just way rich (exhaust all black soot)?

[chris_seven]

Very interesting subject but where to start.

Cold compression tests will tend to find some type of serious problem more readily than a warm or hot compression test but may produce a little more general scatter in the results.

There may also be a difference between a 'wet' and dry compression test.

I would expect to see a variation of less than 10% so 13% does seem to indicate some wear.

Did you check the valve clearances before carrying out the compression test?

Leakdown can be tricky and it can be difficult to compare test results between different test systems.

Leakdown testers are basically miniature flowmeters. The measurement system compares the leakage in the engine to the flow in the orifice.

There is a pressure drop across the orifice and another pressure drop across the leak in the engine.

As the tester and the engine are connected in series the flow is the same across both the orifice and the leak.

If the size of the leak is the same as the size of the orifice the reading would be 50% as the pressure drops would be equal.

If there is no leak then there will be no pressure drop across the orifice and the reading would be 0%.

Most systems seem use the 1mm (0.40") orifice specified by the FAA but in itself this isn't a complete solution. A small leak in a large engine would be the same as a large leak in a small engine so some judgement may be needed depending on engine capacity.

Most systems seem to use 100psi gauges and so system pressures are often set to 100psi just for simplicity but this can create its own set of problems.

In reality any pressure above 15psi will work adequately for measurement purposes but it may be difficult to hear the leak.

Using 100psi can be an issue as this level of pressure could cause the engine to move way from TDC so this needs to be monitored during testing.

[911 tweaks]

my leakdown tester works such that I hook up the inlet pressure line form the shop which is around 90-115psi, and I zero out the inlet pressure gage by turning the inlet pressure knob to increase or decrease the inlet pressure so the inlet neelde is pointing to 0.
Then I hook up the output line if it is called that, to the line that is screwed into the cylinder in question (and the cyl is at top dead center) and then read the output gage for the % of loss...
Quite simple to do as long as the cylinder is at tdc which I verify by the crank pulley notch lining up with the case parting center line...
So, if you adj your inlet pressure to get input air pressure needle on 100%, than the other gage will indicate the % of leaking...
Looks like you did it correctly... good job...

[manbridge 74]

I've come to believe that leakdown tests on 911 engines can be somewhat inconclusive. Mainly due to tapping on a valve spring during test causing a change in results. I like to do them cold and then after it's close to normal operating temp. This combined with a compression test will give some good feedback. And watch the first "sweep" of the needle during warm CT. I like them all above 50 psi hopefully. Or at least all cylinders close to the same. Shoot for the same number of needle movements on each cylinder too.

[manbridge 74]

If cylinder moves at indicated TDC your crank pulley mark might be slightly off. Try it a wee bit to left or right of mark.

[chris_seven]

Originally Posted by manbridge 74

If cylinder moves at indicated TDC your crank pulley mark might be slightly off. Try it a wee bit to left or right of mark.

Has anyone measured the Z1 and compared to TDC checked with a dial indicator?


I don't think they are very accurately marked.

[manbridge 74]

I've never measured one myself but I remember someone posting here that claimed it can be off by as much as 3 degrees. Easy enough to find by using bolt stop method.

[otto_kretschmer]

Leak down test is the way to go. This is my BMW R50/5. I rebuilt the engine 4000 miles ago and the test results are good.

Upstream pressure is about 85 and downstream is 81

[boosted79]

Originally Posted by chris_seven

Has anyone measured the Z1 and compared to TDC checked with a dial indicator?


I don't think they are very accurately marked.

I checked mine and it was dead on. Degree wheel was set to TDC using a piston stop and dividing total degrees CW and CCW by two, the best way to do it.

[manbridge 74]

Nice!!

[Trackrash]

I recently checked my pulley Z1 for accuracy, since I am in the process of building a motor. It is very close. The only issue being the play in the pulley's locating pin would allow about 1-2 degrees movement.

OK, so I can make my own leakdown tester with two gauges with a 1mm orifice restrictor in between?

[safe]

Originally Posted by safe

I noticed that when I connected the tester and turned up the pressure the leakage was highly dependent on the pressure. The higher the pressure the lower the leakage, is that normal? The instruction said to use a pressure between 55 an 150 psi, so I used 100 and I'm not sure how I could have gone above that then that maxes the gauge on the tester...

I revisited this tonight, reread the instructions och watched some youtube....

The behavior with lower leakage, on the dial, the higher the pressure is, obviously, because the leakage gauge isn't a leakage gauge, its a pressure gauge... STUPID ME!!!

Now with the hose between the tester and engine disconnected I turned up the pressure until the leakage was 0 which corresponds to 7 bar on the pressure gauge. Then I connected the hose. This lowered the pressure to about 6.7 bar and upped the leakage to 6-7 % something on the worst cylinder that before showed 3-4%. I could hear slight hissing from the exhaust I believe.

But to me 6-7% is still pretty good, right?

Background:
This engine has an unknown amount of mileage on the bottom end and piston/cylinder/rings, best case 60-70.000 miles. I did a top end rebuild 6 years ago 30.000 miles ago.
This engine see a lot of hard use, 7-8 track days a year and as many auto-x's. Lots of heat, 250 F regular and 265 on some hot days. Even that its not recommended on a 3.2 stock bottom end, it has a soft 6800 rev limiter and a hard at 7000. I regularly let it go to the soft, that's my cue to shift....

[safe]

Originally Posted by Walt Fricke

I do think your two tests probably indicate that the issue you are dealing with is not related to worn rings or valves which don't seal. Maybe some adjustment in the EFI map might be in order?

Literally flooded, as in raw gasoline spitting out the exhaust? Engine dies and the plugs are wet with gasoline? Or just way rich (exhaust all black soot)?

Yes, probably EFI or synchronization issue. I have sent out the injectors for cleaning/testing and they were just fine.

Not spitting out gas. On the right bank the spark plugs are getting black and they stop firing and the engine dies...

I'm getting some new coils (I have VW/Audi COP) that's supposedly better, in any case they are red so they will at least make the engine bay a little more colorful.

[Reiver]

Yes 6-7 % is good