Testing valve seal with acetone?

[Grant]

tl;dr: I don't have a vacuum tester, so I've always filled the ports up with acetone to look for leaks. Is this a valid way to test valve seal?

I got a junkyard EZ36D (Subaru flat 6) from a junkyard to go in my rallycross car. When I get the motor I do a leakdown and find the exhaust valves leak. One cylinder has 70% leakdown. I pull the heads and lap them excessively, but they continue to leak acetone on the bench.

The valve faces appear to have little pits on the mating surface. The rest of the engine looks great (as it should being out of a 2011), so I send the heads off for a port and valve job.

When I get the heads back I notice they still leak acetone. Not only did acetone get by, but light did too. I call the shop up, and the employee I talked to (who did the port work; the machine work was outsourced) seemed genuinely baffled. They give me some shipping labels to send the heads back.

A week later I get an email:

We have a few issues, we have never "tested" a valve job by a flashlight method, We do 5-10 sets of heads a week for 10 years and have never used this technique. This is not the industry standard. They are done via a vacuum test on the ports. We asked multiple head shops in the area including our own and confirm the same. the valves will seat in quickly regardless of light . We then tested several other sets here and we get the same results. If you hand lapped the valves this would get rid of the light you can see. We and most shops don't offer that service.

The standard is 20-25 inHg, yours tested at 21-23" both with valves that do and do not have light showing.

This seemed like BS to me, so once I got the heads back I take them to a local machine shop that does a lot of Subaru work (I did not use them originally because they aren't very reputable). They tell me the valve seats were probably cut with a carbide bit which chattered a bit, leaving high areas on the seat. They lap them and tell me they now seal. I take the heads home and they still leak acetone, but not nearly as much as they had in the past.

So am I just crazy for testing valves with acetone? I've never had known good heads leak before.

Thanks for any help.

[machinedave]

If a valve passes a vacuum test but you can see light through then you might want to look at the valve lash if it is a solid lifter setup.

[wyrmrider]

I've done it, put a little dykem in, or alcohol, you can also use to cc

[Grant]

If a valve passes a vacuum test but you can see light through then you might want to look at the valve lash if it is a solid lifter setup.

The tests were done with the heads fully assmebled except for the cams. So the valves were all shut.

[machinedave]

If a valve passes a vacuum test but you can see light through then you might want to look at the valve lash if it is a solid lifter setup.

The tests were done with the heads fully assmebled except for the cams. So the valves were all shut.

Okay. I've never noticed any light shining through after we do valve jobs. I don't see how it could seal if light shines through.

[Kbails]

if it leaks thru I would assume that the seat inquestion needs to be touched up. try using paint rubbing compound for buffing paint to lap with. the lapping compound speads out making the actual contact point hard to read. after using the rubbing compound there should be a polish line on the valve/seat make sure the line is contacting all the way around. it works good I was leary about it until I tried it. give it a try see what it looks like it

[mag2555]

Did you do the leak down test with the motor hot , or cold?
If hot then lapping and even a valve job may not get things fully sealed up as the valve guide to seat alingment is not truly perpendicular as it should be .
What might seal up fine on theses heads when cold can become a nightmare when hot!

[frozenh2o]

As a point of perspective, the most severe valve seal test I've seen was published by Honda in the 1960's. You fill the combustion chamber with motor oil, then with a blow gun you shoot compressed air at 28 psi into the ports, directed at the valves. Any air bubbles at the valve means a leak.

That is a tough test, but for tiny engines making tiny power, everything counts and Honda knew it.

[Kevin Johnson]

Did you do the leak down test with the motor hot , or cold?
If hot then lapping and even a valve job may not get things fully sealed up as the valve guide to seat alingment is not truly perpendicular as it should be .
What might seal up fine on theses heads when cold can become a nightmare when hot!

http://www.speedtalk.com/forum/viewtopi ... 15#p371611

[Schurkey]

I've used paint thinner (mineral spirits) in the combustion chamber, looking for liquid that would get into the port past the valve seal. Haven't done it by filling the port and looking for liquid in the chamber.

I figure the valve--in operation--is sealing the chamber, not the port. Also eliminates leakage (and mess) past the valve guide/valve seal.

I can accept that the head will expand with heat, move around with fastener torque, and is under considerable pulsating pressure when in operation. But valve jobs are typically performed on room-temperature castings that aren't stressed with a torque plate, and valves are ground at room temperature. How can light get past a room-temperature valve face/valve seat seal unless it wasn't done properly?

I vacuum test valves when I'm done. Your shop allows as little as 20 inches of vacuum? Perhaps this is a matter of different equipment providing different results. I don't have an owner's manual for my vacuum tester; the guy I bought it from said that anything over 50% (15") was "good". I don't believe him. This is what I'm used to:

My vacuum tester, with the vacuum pad plugged using a suction-cup. This is the ultimate vacuum my machine can generate.


And the vacuum as generated when connected to a GM 3.4L DOHC cylinder head, in this case the exhaust port shared by two exhaust valves. I have valve seals in place on the valve stems. I've lost an indicated 2% of the vacuum, but the machine wasn't capable of pulling 100% vacuum to begin with. I presume that the valve seals leak some unknown amount of air.


P.S. Does anyone know what happened to Hoff Manufacturing? I'd love to contact them for some spare parts and literature on this machine.

[pdq67]

Might take a very small piece of 600 grit or whatever emery paper and put it in the seat along with the valve and very gently spin and lap the seats. Then turn the little piece a paper over and do the same thing with the valves. Use WD-40 for a lube....

Heck, "crocus" clothe might do it instead of the emery paper??

"Shade-tree" as all git-out, but it works.

Been there, done that..

Onna ol' pdq67's, "feelie" deals is all...

pdq67

[Grant]

Did you do the leak down test with the motor hot , or cold?

Cold; the motor was on a stand.

Might take a very small piece of 600 grit or whatever emery paper and put it in the seat along with the valve and very gently spin and lap the seats. Then turn the little piece a paper over and do the same thing with the valves. Use WD-40 for a lube....

That isn't nearly as shade-tree as what I just tried: lapping the valves with toothpaste. The local autoparts store only had "valve grinding compound" which seemed too coarse given that the valves were almost sealing. The toothpaste has worked on the two valves I've tried it on. I hope it will work for the rest.

Schurkey, thanks for the info. Sounds like the original shop I sent them to was dicking me around. I think its kind of astounding they weren't willing to fix their own work after paying for shipping. I think I'll ask for a small refund to cover the lapping they'll need to finally seal.

[ZEOHSIX]

Sounds like your in over your head and need to visit a machine shop to me. You can buy a test plate for under $30 at Goodson and use a vacuum brake bleeder to test a head.....if it's leaking acetone it will certainly leek combustion gases why not just delegate some of your race program to a auto machine shop if your serious about winning and use the time to develop an aspect of your program you equipped to work on?

[Grant]

Sounds like your in over your head and need to visit a machine shop to me. You can buy a test plate for under $30 at Goodson and use a vacuum brake bleeder to test a head.....if it's leaking acetone it will certainly leek combustion gases why not just delegate some of your race program to a auto machine shop if your serious about winning and use the time to develop an aspect of your program you equipped to work on?

That was the original plan, but the two shops I've tried so far don't seem to have very high standards when it comes to valve seal. Got any recommendations for good Subaru engine builders?

[modok]

I had one customer that would go in the bathroom and turn out the lights and use a tiny flashlight in the port to inspect the valves WITH NO SPRINGS And HE was hard to please, and he did a lot of subarus too! You can detect .0005 gaps with light, gaps that the force of the spring would always bend the valve enough to close.
After probably 30 valve jobs i got to the point I could get a perfect seal 95% of the time with stones, and i also got pretty good with the serti too, 8 years ago.
Runout is tested by nsew method or a runout gauge, contact is checked with Prussian blue paste. Final check done with vaccum test. lapping is not required, but sometimes beneficial to the seat, but not to the valve, use a junk valve to lap.
TODAY, my co-worker had a seat he could not get to seal and finally he found the seat was loose. Not the first time, nor the last.

old subaru 2.5 engines I've seen the cylinders .004" out of round if they have ever been overheated. Look up how to use nsew method, get prussian blue paste, measure cylinders.