New head gaskets won't seal

[89Ru]

2007 Outback 2.5i EJ253 with AVLS.  100k miles.  Basic problem: Can't get heads to seal with 15 psi into cooling system.  Dropped 1/2-1 psi shop air in one hour.  Filled with coolant.  Dripping coolant on driver side in multiple places.  Next step is to use some kind of sealant on the gasket.  What I'm asking is which sealant.  Details below.  

History:  Bubbles in overflow tank and gurgling noises in cabin.  Never overheated under my watch but I just bought this car, have only 1500 miles on it.  Did radiator exhaust gas test twice, once before sale and once after for two minutes, both clear.  Drives great btw.  Hot heat in cabin.  Radiator level drops and would not draw coolant from overflow tank on cooldown.  Coolant temp fluctuates as expected with significant air in system, between 180 and 210.  Blue exhaust gas tester solution stayed blue until exposed to tailpipe fumes, then turned yellow.  Tried new radiator, new water pump and hoses, bled air I don't know how many times.  Finally got fed up and pulled engine out, pulled heads off, valve rockers, and cam cover and sent to local machinist.  Existing gaskets: unknown vendor MLS (multi level steel) with copper spray.  Machinist said warp at .002" and .004" and machined them down.  New gaskets: I used Fel-Pro solution: PermaTorque MLS with LaserWeld, gasket #26415PT with the blue coating.  Reused head bolts.  Rinsed bolt holes with oil x2.  Skipped the bolt stretching steps and just did the torque to 7.4, 22, and 44 ft-lb.  Then 80-90 degrees, 45 degrees, and lastly 45 degrees on center bolts.  Cam covers resealed with permatex ultra grey.  Reinstalled rockers and let sit for 24h.  Left timing belt off and cams loose.  Did leak test as mentioned with air into throttle body coolant hoses.  Tied heater pipes together with heater hose.  Tied water pump intake to coolant crossover pipe with scrap hoses.  Won't hold 15 psi shop air with sealed cooling system.  Filled with coolant and slow drip off DS head with 15 psi.  Unbolted.  Rechecked flatness of heads and short block in eight locations with t square and .002 feelers, pretty much unchanged from before installing.  Took heads back to machinist and he just laughed saying no one in this state can measure Ra smoothness, he has been doing this for 40 years and does them all the same.  I showed him the Felpro handout stating machine to 60 Ra (600 Rz).  He suggested a can of copper spray.  He did say he has some other type of cutters but they come out "the same." 

So where to go from here?  Do I need a new machinist?  I've seen some use permatex aviation gasket sealer.  Subie dealer says use composite gasket p/n  11044AA633.  Since this car already had MLS gaskets with copper spray I'm thinking of abandoning Felpro and going with Subaru OEM gasket.  In that case, which one?  Sealant or not?  I have a bunch of pictures to upload when I get a minute. 

Edited October 23, 2018 by 89Ru
clarity

[idosubaru]

I wonder if the gaskets need to be run/heated up to fully expand/seal/crush?

Why are you pressure testing it after installing them?

Did it look like the heads or gaskets had been worked on before?

FelPro/Subaru, sand paper/blanchard grinder (perferred resurfacing machine), no spray and spray.....are not distinctions that matter.  people do all of the above, or none of the above without issues so those aren't the likely culprits. 

i use Subaru and no spray at all, they're supposed to be installed dry. but i know people use that stuff. 

Edited October 22, 2018 by idosubaru

[GeneralDisorder]

NO SEALANT. Pull it apart and check everything. You have done something horribly wrong. 

I can measure Ra finish. I bought a meter for it. And I'm not even a machinist. 

Use the Subaru head gasket part number ending in 770. 

GD

Edited October 22, 2018 by GeneralDisorder

[89Ru]

Ido: I did the pressure test to avoid having to install in vehicle and pull out again if it leaked.  Have no idea if putting fire in the hole would get it to seal up.  My guess is someone else worked on this since I don't think it has OEM gaskets.  I'm assuming that MLS with copper sealant spray isn't OEM for the EJ253.  

GD: The tool for surface roughness is a bit pricey for me lol.  Everything is apart.  Felpro gaskets aren't deformed or scratched that I can tell.  I'm using new torque wrenches from precision instruments.  Marked the bolts so I wouldn't overtorque when doing the angles.  A bit of stick slip when torquing down.  I rinsed the bolt holes with engine oil and cleaned the bolts with brake cleaner and dipped the threads and washers in engine oil before installing.  

Will get 770 gasket and bolt back together dry.  I don't have an old head bolt to make into a thread chaser but I can buy a new bolt and make one.  Just use a grinder to machine 3-4 slots parallel to long axis of bolt, fill with grease?  This is my first time doing head gaskets so any tips are greatly appreciated.  

Edited October 23, 2018 by 89Ru

[forester2002s]

8 hours ago, 89Ru said:

Skipped the bolt stretching steps...

Not sure what you've skipped.

Subaru head-bolts are NOT torque-to-yield.  that's why you can reuse them.

I've always followed the complete torqueing sequence when installing new head-gaskets.

[89Ru]

16 minutes ago, forester2002s said:

Not sure what you've skipped.

I've always followed the complete torqueing sequence when installing new head-gaskets.

Perhaps this is one of several errors.  I skipped steps 3,4 and 5 below.  Was following some advice that stated these steps were for new bolts only.  I started on step 6 with existing bolts.  

One other question.  How many passes do you make?  I found that the early steps take 2-3 repeat passes in alphabetical order to get all bolts to the specified final torque for each step. 

[GeneralDisorder]

Skip NO steps. Regardless of bolt newness. 

Repeat each of the ft/lb torque sequences till they do not turn any further. If that takes three times or 50 times - once you run through the sequence and they no longer turn you can move on. Obviously this does not apply to the angle torque values. 

If you get creaking (stick-slip), STOP and take it apart. Yes chase the threads with an old bolt. And then you need to lube them with Amsoil Engine Assembly Lube. Lube the bolt threads of one bolt, run it in and out of each hole with the head off a couple times - relubing between each. Apply assembly lube LIBERALLY to each bolt and between the bolt head and top of the washer (not between the washer and the head). Make sure you use the small washer bolts in the corners and the big washer bolts in the center. 

You must use something like the Amsoil lube. Regular engine oil won't handle the load and will creak. Once you achieve ZERO creaking you are doing it right. Creaking means false torque readings and in all likelyhood insufficient clamping. 

My point was your machinist is a hack. I bought an Ra meter (used) for about $350. It's a neat little tool and a proper machine shop would own one (maybe several) in order to ensure they achieve proper specified surface roughness for any given application. This doesn't only apply to cylinder heads - lots of things need to have specific surface qualities. 

GD

Edited October 23, 2018 by GeneralDisorder

[89Ru]

Thanks for the specific details.  I definitely had some head bolt creaking with engine oil.  I've got some Sta-lube from O'Reilly's.  

I wish my machinist was willing to invest in some new tools.  He wouldn't buy a helicoil kit last time I needed stripped threads fixed.  

12 hours ago, GeneralDisorder said:

You must use something like the Amsoil lube. Regular engine oil won't handle the load and will creak. Once you achieve ZERO creaking you are doing it right. 

[GeneralDisorder]

Yeah time to find another machinist. Although heli-coil kits are generally something stocked at the mechanic level. All the machine shops I use make these things happen without them being my problem. I dropped it off so I could go back and do what *I* do - if I wanted to figure it out I would just do it myself. I have a lathe, and a mill, and I'm not afraid to use them. I just don't have the time. 

GD

[89Ru]

Can't figure out how to get pics over 2mb to show up yet.  Do these look too rough for MLS gaskets to seal?  I'm going to take the heads to different machine shop on my next day off and see what they can do.  

http://i444.photobucket.com/albums/qq166/89ru/IMG_20181020_204143202_zpsjrpy4hnq.jpg

http://i444.photobucket.com/albums/qq166/89ru/IMG_20181020_204052240_zpsywe7ojti.jpg

http://i444.photobucket.com/albums/qq166/89ru/IMG_20181020_204010939_zps7rscg61f.jpg

http://i444.photobucket.com/albums/qq166/89ru/IMG_20181020_203945342_zpsdu8ry94o.jpg

http://i444.photobucket.com/albums/qq166/89ru/IMG_20181020_203934441_zpstrjeafnh.jpg

http://i444.photobucket.com/albums/qq166/89ru/IMG_20181020_203915589_zpsnonsqao7.jpg

http://i444.photobucket.com/albums/qq166/89ru/IMG_20181020_203858052_zpsltupd2ln.jpg

Edited October 24, 2018 by 89Ru

[GeneralDisorder]

Nice. The guy resurfaced those on a belt sander. It's probably borderline for surface Ra. A proper surfacing should have been done either by lapping or with a CBN fly-cutter.

GD

[idosubaru]

On 10/22/2018 at 11:36 PM, GeneralDisorder said:

 Apply assembly lube LIBERALLY

How much of a concern is excessive lube coming off the bolt during install and getting between the gasket/mating surfaces?  The thicker lube is probably better than oil also for this reason?

45 minutes ago, GeneralDisorder said:

Nice. The guy resurfaced those on a belt sander. It's probably borderline for surface Ra. A proper surfacing should have been done either by lapping or with a CBN fly-cutter.

GD

Yep - tell tale grooves, i've seen shops do this as well.

[GeneralDisorder]

Haven't had any problems with excessive lube (and we use quite a bit) causing any gasket contamination. 

Yeah - that's an OLD machinist who doesn't understand or believe in gasket technology of the last 30 years. Even my Fel-Pro .015" shim gaskets for my Vortec come with BIG BOLD instructions that the surface finish MUST be better than what was originally required even at the factory for the composition gaskets. I have to deck the block and the head to get a proper seal with the single layer shim gaskets. And both my machinist and my cylinder head shop use a CBN cutter to accomplish this. 

NEVER go back to that guy. I wouldn't even surface a tractor head on a belt sander. That's not going to yield a true surface - in fact those heads might be junk. Need to check them not just for flatness and Ra but also for the surface being square to the head. Have to check it with a height gauge on a granite surface plate. 

GD

[89Ru]

Found a new shop that claims to machine to mirror finish with carbide cutters on a PCM mill machine.  Will drop off heads tomorrow morning.  I asked about measuring surface roughness and didn't get a clear answer.  I'll ask to check if the surface is square with a height gauge.  Two other shops in the area gave this guy a recommendation so going with the majority.  Crossing my fingers that these can be reworked.  New gaskets arrive tomorrow.  

1 hour ago, GeneralDisorder said:

And both my machinist and my cylinder head shop use a CBN cutter to accomplish this. 

Need to check them not just for flatness and Ra but also for the surface being square to the head. Have to check it with a height gauge on a granite surface plate. 

GD

 

[89Ru]

Got heads back from new machinist.  He used a diamond cutter.

 

Pressure test setup

Edited October 31, 2018 by 89Ru

[89Ru]

Ground slots into a head bolt and chased all bolt threads with assembly lube.  Lubed bolts well and ran them in and out several times w/o heads on.  Oiled cylinder walls.  Might have slopped some oil on the fire ring for the driver side.  After remachining heads I reinstalled with new dealer 770 gaskets.  Used no sealant.  Torqued heads following procedure without skipping steps.  Zero stick slip on the threads this time.  Assembly lube did the trick.  

Have 15 psi into dry cooling system, test setup as in previous post picture.  Went out with the family for Halloween.  Came back and its losing pressure, just over 2 psi in 7 hours which is better than last time around with the sanded heads, about 2 psi in 2 hours I think.  I soap bubbled all the joints last time around and nothing leaked except coolant at the gaskets.  

Kind of wondering if this will seal up after it gets up to temp.  Short of picking a gasket sealant (horrors) I am out of ideas.  

[GeneralDisorder]

There's so many hose joints on that rig - almost no way that could seal. Cute though. I speak from some experience being a former Compressor/pump technician. That would be a nightmare to try to seal perfectly. Doubt you have a HG leak with that many other potential (more likely) leak points. 

Sounds like you did it right this time. Run it. Should be fine. Heads look good. We don't do any such testing we just send it.... knowing our process is setup for success. 

GD

Edited November 1, 2018 by GeneralDisorder

[1197sts]

Can someone tell me which manual the above referenced head torque sequence if from?

[89Ru]

11 hours ago, 1197sts said:

Can someone tell me which manual the above referenced head torque sequence if from?

2007 Legacy Outback FSM, SOHC section 

https://sl-i.net/FORUM/showthread.php?18087-Subaru-Factory-Service-Manuals-(FSM)-Every-Model-USDM-EU

[1197sts]

Thank you!

[Old timer]

Hey did you check the deck for being flat.

I've seen that before on Ford 302 I was working on .

Haven't done a Subaru yet,but if and when I do I'll have the block decked with the heads.

Then everything is flat and a good set of the new fel  pro head gaskets should last for ever,

[idosubaru]

1 hour ago, Old timer said:

Haven't done a Subaru yet,but if and when I do I'll have the block decked with the heads.

GD says not to do the block, or not without a very dedicated process that most places aren't capable of.  I wont' try to regurgitate his mechanical reasoning but he's well versed on the technical attributes and failure modes of subaru block/head/mechanical/machining...  he's described it elsewhere on this forum if one was interested.

[GeneralDisorder]

The deck is never a problem from the point of flatness. The block is just too rigid to warp in that way really. Unlike vintage American stuff, the machine work is never a problem on the Subaru engines. Like as not that Ford 302 was manufactured with the decks wrong. The Subaru deck can have pitting in the area of the fire ring, but unless you completely dissasemble the engine, decking it with conventional techniques is pretty much impossible due to contamination of the oil passages, cylinders, etc. And point in fact we rarely deck used blocks even when they are completely torn down. It's just not an issue. 

GD

[89Ru]

Deck is flat without any pitting.  Car went back on the road last week.  No bubbles in coolant overflow tank or gurgling noises behind dash in cabin.  Overflow tank level is drawn down overnight as expected during burp period unlike before where it would stay high.  Rad level is constant unlike before where it would drop.  Coolant is staying green not dirty gray.  No leaks at head or valve cover.  Small pesky coolant leak at lower hose, rad side, which undoubtedly contributed to pressure loss during leak testing of cooling system.  I'm gonna send the machine shop a Christmas card for saving these heads from the scrap pile.  Thanks to all who posted key advice above.  

[GeneralDisorder]

It occured to me after posting last time, that your pressure test rig might have experienced pressure drop simply from the compressed air falling in temperature (and therefore pressure). The air is hot leaving the compressor, and will drop in pressure after sitting in that engine. 

For radiator hoses - buy them from Subaru. They fit the best and last the longest, and they are relatively cheap. Like $25 each. 

GD