Hi,

So I finally replaced my broken rb20 t3 turbo for a subaru td04l turbo, the swap turned out great (made a t25 to td04 adapter), car runs much better, though there are some idle and cold start issues:

1) I noticed some oil under the car, upon closer inspection the left engine mount and the block is covered in oil there - like it's leaking from the oil supply for turbo hole on the side of the block (stock hardline and fitting where replaced with a new high pressure hose with a banjo fitting and bolt). what could be the problem, provided that the bolt was screwed in very good and there where no leaks up until today (for 2 weeks or so)?

2) car won't start when engine is cold from time to time (won't start after the night). tried replacing the idle valve - nothing changes, unplugging the connector doesn't do much either (fuel pump works, engine is cranking, but looks like no spark). any ideas?

3) on idle during warming up tach goes back and forth from 1.5k to 2k rpm, then sets down at 1.2rpm and after hitting the throttle goes down to 900 or so rpm. is the idle valve broken or some other problem?

any help appreciated

About #1, I had a hard line going to my turbo that cracked after a few years. You couldn't see the crack, but it was gushing oil from the nut. It's better to use flexible lines in my experience. I think the vibration cracks hard lines after a while.

I don't have any useful ideas about the other two.

Edit:

I see you had a flexible line. Perhaps it failed?

The line was custom made at a local shop specializing in hydraulics for heavy machinery. the line itself isn't wet, oil's on the block..but I'll double check, maybe the banjo fitting's gone bad.

small update: ran the ecu diagnostic mode III, the result was a code 13, so it looks like either the engine temp sensor gone bad or some wiring issue. also measure the resistance on the air regulator under the TB, shows about 75-76ohms, whilst standard should be 70ohms. so I guess it's slight stuck but working.

Code 13 is a bad code to have. The ECU is HEAVILY reliant upon the ECT and if it's getting a faulty signal on that circuit, the car will run poorly. Get that sorted out ASAP.

As for the oil leak, I'd have to see pictures to understand what you're talking about.

The cold start issue sounds like a bad air regulator. They almost never test bad from a resistance standpoint. What happens is they get stuck shut/open/halfway shut. Without the extra air passing through during cold start conditions the engine can't start, or at least can't maintain idle. When you first start the car up, the engine should idle at about 1500rpm, then slowly idle down to about 1200, and then sit there for a bit, and then drop down to normal.

Thanks, I'll be replacing the sensor and checking the wiring as well.

So there is no way to test the air regulator? replacing it is the only option?

strange, but the engine behaves like you described when I can get it to start. does that mean the aac and air regulator are working?

It's likely that it's working intermittently then. That's why it will start those times and not others.

Finally got to replace the engine temp sensor - now car starts all the time. Also charged the battery - seemed to be discharged from all those attempts to get the car running. code 13 gone, ecu shows code 55.

but idle is at 600rpm right after start up, and goes up to 900 after warm up. IAC valve gone bad?

Oil is still leaking a bit, can't find the time to take of the intake stuff to take a closer look.

Bad air regulator. The idle control valve under the throttle body. It's stuck shut. That's really common.

ska69 wrote: So I finally replaced my broken rb20 t3 turbo for a subaru td04l turbo, the swap turned out great (made a t25 to td04 adapter), car runs much better, though there are some idle and cold start issues:

Wow. What's the response like? I'm betting it's almost N/A with that tiny thing.

ska69 wrote: 1) I noticed some oil under the car, upon closer inspection the left engine mount and the block is covered in oil there - like it's leaking from the oil supply for turbo hole on the side of the block (stock hardline and fitting where replaced with a new high pressure hose with a banjo fitting and bolt). what could be the problem, provided that the bolt was screwed in very good and there where no leaks up until today (for 2 weeks or so)?

ska69 wrote: The line was custom made at a local shop specializing in hydraulics for heavy machinery. the line itself isn't wet, oil's on the block..but I'll double check, maybe the banjo fitting's gone bad.

Can you elaborate on that, or post a pic? The stock oil feed cannot be replaced with a banjo, at least not straight. The block has an 1/8-BSPT fitting, which means it seals with a tapered thread. Banjo bolts have straight threads, and seal with crush washers against very flat surfaces, which the block is not. The two fitting methods are incompatible, so I'm curious how they bridged the two standards, that involves a banjo on the block side. I can imagine some ways to do it, some will be leak-free, some will be not. Also, 1/8 BSPT is not very well known globally, I don't know how well it is known over there, can be they mistook it for something else.

ska69 wrote:ecu shows code 55

Code 55 is good. Code 55 means all systems are GO.

float_6969 wrote:Bad air regulator. The idle control valve under the throttle body. It's stuck shut. That's really common.

Thanks. I'll be replacing it soon with a working one

blownhemi wrote: Wow. What's the response like? I'm betting it's almost N/A with that tiny thing.

I like it. I really can't say much more as I haven't had much time to test it out. boost is set at 0.5 bar at the moment and perfectly suits me at least for now. If and when the turbo will get to small for me I'll just go up to a td05

blownhemi wrote: Can you elaborate on that, or post a pic? The stock oil feed cannot be replaced with a banjo, at least not straight. The block has an 1/8-BSPT fitting, which means it seals with a tapered thread. Banjo bolts have straight threads, and seal with crush washers against very flat surfaces, which the block is not. The two fitting methods are incompatible, so I'm curious how they bridged the two standards, that involves a banjo on the block side. I can imagine some ways to do it, some will be leak-free, some will be not. Also, 1/8 BSPT is not very well known globally, I don't know how well it is known over there, can be they mistook it for something else.

sorry, don't have any pics at the moment. Thread on the banjo and the stock adapter fitting was the same. I had to go with such banjo as the stock hardline has been jammed and later cracked during removal. so a new flexible line and a banjo bolt with crush washers seemed like a perfect solution.

are you suggesting that the oil leak is caused by inaporpriate oil flow hardware? if so, what would be the right way to do the job?

blownhemi wrote: Code 55 is good. Code 55 means all systems are GO.

yep

ska69 wrote: sorry, don't have any pics at the moment. Thread on the banjo and the stock adapter fitting was the same. I had to go with such banjo as the stock hardline has been jammed and later cracked during removal. so a new flexible line and a banjo bolt with crush washers seemed like a perfect solution.

are you suggesting that the oil leak is caused by inaporpriate oil flow hardware? if so, what would be the right way to do the job?

That was my first thought, when I've read your oil setup, but I didn't quite understand. Now, if I understand correctly, you are still using the small factory elbow on the block, and the banjo bolt is screwed into the stock elbow? The other end of the elbow is a metric 12x1.0 thread (IIRC), so it would look strange, but it would kind of work. Except in that case the problem is, the rim of that elbow is not flat enough in its original shape for a crush washer to seal properly.

There's more than one way to do it right. How you can get it done right the easiest way, is very dependent on what hardware you have access to over there, or what you're willing to order off ebay/shops. Google/search around, see how people have replaced the factory hardline, that would give you some ideas. The pieces:
- start: BSPT 1/8 in the block
- end: M10x1.0 (?) banjo on the td04l
- a restrictor somewhere in between
From here on, it's just a simple puzzle to be finished.

blownhemi wrote: That was my first thought, when I've read your oil setup, but I didn't quite understand. Now, if I understand correctly, you are still using the small factory elbow on the block, and the banjo bolt is screwed into the stock elbow? The other end of the elbow is a metric 12x1.0 thread (IIRC), so it would look strange, but it would kind of work. Except in that case the problem is, the rim of that elbow is not flat enough in its original shape for a crush washer to seal properly.

Thanks for your reply. No, I'm not using the factory elbow (http://img6.imageshack.us/img6/2389/blockfitting1.jpg) - it was removed and a banjo bolt with the length and thread was fitted with crush washers. I guess it's leaking from there - still haven't got a chance to take a closer look.

So I have no restrictor in the line (is it necessary for a journal bearing turbo at all?)

I found this in another thread here (http://i288.photobucket.com/albums/ll18 ... thread.jpg) but this would lead removing the banjo and going with a different solution, I guess.

ska69 wrote:Thanks for your reply. No, I'm not using the factory elbow (http://img6.imageshack.us/img6/2389/blockfitting1.jpg) - it was removed and a banjo bolt with the length and thread was fitted with crush washers. I guess it's leaking from there - still haven't got a chance to take a closer look.

So I have no restrictor in the line (is it necessary for a journal bearing turbo at all?)

I found this in another thread here (http://i288.photobucket.com/albums/ll18 ... thread.jpg) but this would lead removing the banjo and going with a different solution, I guess.

Generally speaking, you'd need to know what oil pressure the turbo needs, and what oil pressure the engine supplies. That's the only sure way to tell if a restrictor is needed. But generally, a restricotr is needed for both types of turbos, only journal bearings need bigger holes in the restrictor, than ball bearings. That tiny td04l will definitely need one. BTW, that factory elbow is the restrictor (~1,5mm) in a completely stock setup.

NOTE: 1st picture is correct, 2nd picture is INCORRECT. The block thread is BSPT (Tapered), not BSPP (Parallel). BSPP might work if you use some high-temp, high-pressure Loctite on the thread to seal, but I'd just do it right, and get a BSPT connection.

In the US people usually buy an NPTF-AN adapter for the block to begin (NPTF is similar enough to BSPT to seal), then do all AN fittings and lines up to the turbo. Mostly because all these parts are readily available from shops around there. Maybe the easiest for you would be to get an BSPT-AN (or NPTF-AN) adapter off a shop/fleabay, and have the rest made ready by a hydraulics shop. They will not know what AN (or BSPT or NPTF) means, but AN has 100% equivalents in the JIC fitting standard, and hydraulics people usually know JIC. Here's a conversion chart. e.g.: -4AN is the same as 7/16 JIC.

blownhemi wrote:Generally speaking, you'd need to know what oil pressure the turbo needs, and what oil pressure the engine supplies. That's the only sure way to tell if a restrictor is needed. But generally, a restricotr is needed for both types of turbos, only journal bearings need bigger holes in the restrictor, than ball bearings. That tiny td04l will definitely need one. BTW, that factory elbow is the restrictor (~1,5mm) in a completely stock setup.

ok, thanks! looks like a decent solution.

also could it be done without getting rid of the banjo flex line I have? 1/8 bspt to an/jic female and a banjo bolt?

as for the restrictor - would it be ok, if there is a restrictor in the banjo bolt on the td04l (found this http://i26.photobucket.com/albums/c127/ ... G_0304.jpg and a description saying that oem impreza td04 oil feed banjo has only one 1.5mm hole.)?

ska69 wrote: ok, thanks! looks like a decent solution.

also could it be done without getting rid of the banjo flex line I have? 1/8 bspt to an/jic female and a banjo bolt?

To be able to keep your existing banjo ended line, you'd need an adapter. And an adapter of a rare kind. It has to be 1/8 BSPT male on one end, and have a metric inside thread on the other end, with a relatively large flat rim, that the washer can seal against.

Something like this. Except this states NPT, and you'd need NPTF (this MAY be NPTF, there's a lot of confusion and inaccuracy out there regarding these). But the very best would be BSPT male to metric female adapter. I'm not a fan of attaching BSPT to NPTF, but it seems to work for people.

Or you could try one of these. Screw the banjo bolt into the NPT thread. M10x1.0 thread pitch is 1.0mm, 1.8NPTF-27 has a 0,94 mm pitch. it might work, but check before assembly that the bolt doesn't bottom out.
(At least I'm going under the assumption that your banjo is M10x1.0...)
In both cases, I'd polish up that flat rim on a fine sandpaper, just to be sure.

ska69 wrote:as for the restrictor - would it be ok, if there is a restrictor in the banjo bolt on the td04l (found this http://i26.photobucket.com/albums/c127/ ... G_0304.jpg and a description saying that oem impreza td04 oil feed banjo has only one 1.5mm hole.)?

Should work. At least I really hope so, because that's what I'm gonna use on my Honda.

Finally found something locally - turns out local hydraulics shops carry BSPT fitting. Which is good however they don't have any 90 degree elbows - so the hose would be bent once installed. Is this a bad thing? I can only think of adding a 90 degree elbow somewhere after the BSPT fitting. any other ideas?

There's more than one way to solve this correctly. Just use the correct fittings, correct hoses, don't bend a flexible line more than allowed, keep things away from exhaust heat, and no more or no less restrictions than needed.
Inside these parameters, every solution works. Mine exits straight from the block, loops under the compressor housing/inlet pipes, and comes up on the other side of the turbo, into the top of the CHRA. (I had a longer line made, so it fits multiple turbo locations...) Maybe you could do it this way, too, have a smaller piece fo flexible line made to connect your existing line to the BSPT fitting you screw into the block?

I think I got this sorted out: Oil was leaking from between the block and banjo fitting because the fitting wasn't in line with the block. So a local car shop machined an adapter between the fitting and block and all seems ok now. I hope at least. lol.