Leaking exhaust manifold possible CEL- fix/weld ideas/tips, or just swap

So got a Mazda 3 for Jr and it has CEL. P0421. No other codes. Possible dying cat but no bad smell, smoke etc. Before I spring for that I've been checking all I can. 

I have swapped and checked O2 sensors and they seem fine. Replaced plugs, reset and drove, came back, everything looks original and feels and drives fine. Only 103k on a '06.
Rear o2 sensor fluctuates while steady driving so its showing there's an issue with gasses after the cat.

One of the possible reasons I've researched is  exhaust leak so I backpressured the system from the tailpipe and hit it with soapy water. Well, looky here.... Two pinhole leaks where the manifold pipe goes into the flange. It doesn't look like its coming from the manifold gasket itself.

I have a small welder, (sort of know how to make blobs with it) so... should I try running a bead around them? Any issues with expand/contract? I read about JB weld or muffler putty to clog it up? (just want to pass smog for registration now)

How about going in order of complexity, slather some putty, see if it closes it, if it doesn't, then weld?

If all that fails buy the new part? Or is it a fools errand and I should just spend $700 on a new part and not waste my time...

Whatever you do, if you plan to weld do not throw JBweld on it.

I think for the size I'm going to try some extreme heat JB in the seam between the tubes and the flange. If that stop sit up then I'll pass. If the cat is dead I'll have to replace anyway.  Probably not worth going right to uninstall, weld, reinstall (new gaskets) to find out it still needs replacing. 
 

From experience, you may need to finish welding a new manifold assembly, so you may as well pull it and weld up the one you have.

+allofit for not slathering goops on it to attempt to seal it up.  It will not seal to rusty metal and it will not hold up to the heat.  The pipes may be dull red while the flange is relatively cool what with being bolted to a 200-odd degree chunk of aluminum. 

And then you will not be able to weld it very well because you contaminated the area.  Maybe if you got an oxygen-enhanced torch setup that could get the area orange to burn off any of the magic goops first, but I would not count on it.

With how clean the fasteners look i would just pull the manifold and weld it.

Otherwise considering the whole shebang (manifold and precat) is under $400 on cock auto and fleabay and my time is worth $200 an hour I would just pull the whole thing and swap in a new one since its the same amount of work as pulling the old one, repairing it, and putting it in just to do the whole job again when the manifold isnt the issue.

Worst case Ontario you could try a sparkplug anti-fouler on the rear O2 and see if that solves the issue.  Sometimes it does, sometimes it doesnt.  

I've seen plenty of data with exhaust leaks like that- and I don't think that's causing the light.  Or at least I don't think it would fix it- if it was the cause, then it would have set the mixture to a point of partial misfire to damage the cat.

But if that was the cause, there would be more errors than just a cat efficiency error.  To be *THE* problem, the rear sensor would be more consistently lean as opposed to switching.

BTW, having the rear O2 sensor move around isn't necessarily bad- if it were once every 10-20 seconds, that may not be bad- some calibrations I worked on, I had it do that very intentionally.  But if it's setting the code, then the cat is probably dead enough.  

In reply to alfadriver :

I will defer to your data, but assuming those leaks were there for a long time, the engine would have been getting overfueled to compensate for the exhaust leak for that long.  Rich cylinders and a small amount of "air injection" overworking that tiny cat for how long?

In reply to Pete. (l33t FS) :

Given that the system can compensate for the leak (noting that the rear O2 sensor can deal with it), that lean/rich problem isn't much.  Unless there's a lot of towing time, it wouldn't be near the temp limits were it would get damaged from extra exotherm.

In reply to alfadriver :

Good to know.

At a recent tech seminar they were going on about how modern ultrafine control over fueling has allowed them to use smaller and smaller converters... with the subtext that they are nowhere near as abuse tolerant as the monstrosities used in the 70s and 80s.

Pretty sure I'll be buying a new one now or later. Hence reconsidering a double or triple install time. I was going to try the non fouler but with CA smog may have a visual issue even with no codes and it'd cost me another $50 to fail.

In reply to alfadriver :

only code is p0421 but here was a snapshot at steady 2500rpm and it'd be similar cruising the freeway. just have torque lite. Bank 1x1 never shows anything, even with new ntk sensor, I have a federal smog car and sensor 1 is 5 wire, not cali 4 wire.

Pete. (l33t FS) said:

In reply to alfadriver :

Good to know.

At a recent tech seminar they were going on about how modern ultrafine control over fueling has allowed them to use smaller and smaller converters... with the subtext that they are nowhere near as abuse tolerant as the monstrosities used in the 70s and 80s.

Actually, modern catalyst are much more abuse tolerant than older ones, and they will be getting even better in the near future as new emissions requirements rolls into the fleet.

In reply to Nomad :

Seeing that data... well...

Both things need fixed.  The catalyst seems dead enough based on how fast it's switching.  But it's also not going all that rich when it does go rich- it should be going closer to .8V instead of .6V.   Since the catalyst needs to come out, just have those holes patched up.

The reason you don't see the front O2 signal like that is that it may not exist in that form anymore- it's probably a WBO2 sensor, so the output is very different.  If it were a regular O2 sensor and was that dead, it would be giving you a number of sensor codes.

In reply to 93gsxturbo :

"Worst case Ontario" sounds like a fun tv show I would watch. 

In reply to alfadriver :

Yep. Plugged the holes, cleared and drove it. Same pattern. Single code appeared again. Thanks for the guidance all.

At least I've "exhausted" all other 0ptions through testing it. Just ponied up for the manifold with integral cat with gaskets. $456 all said. I'll swap it in and hope for the best. New OEM part should work and have some sort of factory backing.

Yeah, the upstream is a wideband on this federal epa car. I think the cali version is different.