A couple of weeks ago, someone started a thread about DIY headers. Since I'm susceptible to that sort of thing, I got a book about exhaust design, and I've been trying (successfully so far) not to order Covell's "Working With Tubing" video, that Mueller suggested. The conclusion of that other thread seemed to be : don't DIY unless you really, really have to, or really, really want to. So for this thread, I'd like to find out if there's a need for custom headers for an EJ25/914 conversion, or if there is any reasonably affordable stock solution that works. And also to see what folks who are already driving conversions are using : stock manifolds, custom headers, ????


So far, my impression from looking at pictures is that the available Subie headers make too gentle a bend as they come down from the exhaust flange, ending up about as low as the stock oil pan. Not a problem in a Subie, but for us that's getting awfully close to pavement. Although sharp bends are usually undesireable, right near the heads they seem to be less of a problem. Also, I have no idea how they relate to various engine mount bars that people are using.

Another issue is primary length. For a 914, I think it would be more fun to emphasize torque at the mid RPM band, as opposed to the top end, so longer, smaller-diameter primaries would be the way to go.

Finally, the Smith book ("The scientific design of exhaust and intake systems", Philip Hubert Smith) talks about two categories of exhaust : "Interference" where reflections from one cylinder makes a significant contribution to scavenging from other cylinders, and "Independence", where the scavenging for each cylinder is almost entirely due to reflections from that cylinder. Most headers I see in piuctures look like the Interference type. Given that header design is so complicated and trial-and-error in nature anyway, I would think that the Independence approach would be less prone to failure. The key to the Independence approach is that the primaries do not merge in the standard Y connection into a slightly larger pipe, but enter an "exhaust box" (it's an old book, so the terminology is probably out-of-date) which has a fairly large volume, say 12-15 times larger than the displacement of one cylinder. Each pipe enters the box directly with no merge. So when a pulse comes down one pipe, it only increases the pressure in the box a small amount, resulting in very weak reflections going up the other pipes; but the original pipe sees a strong (negative pressure) reflection from the exhaust box. So there's really only two parameters to tune : primary length and diameter. Shortly after the exhaust box comes the "silencer", which is pretty much a standard glass-pack muffler.

Anyway, this last part was about how a custom design might look. But lets hear it : what are all you Converts out there (thinking of) using ?

Mostly due to simpliticy, I'm planning on using a set of the Outback Motors collectors, and then making a turn up and over the axles and having a pair of mufflers on either side of the transmission. I'll have a tube that crosses over above the transmission (I believe that there is enoguh space.. otherwise I'm not sure what to do).

The figure below shows the crossover as a red circle, which goes across to the other side.

I want the maximum ground clearance possible since I'm doing some funky stuff with diffusers and stuff.

-Tony

The Smith book, as you say, is an old book (I also have a copy). From what I've been able to see, the "independent" scheme has pretty much fallen out of favor in all engines that are extremely exhaust tuned (look at motorcycles and F1 cars for perfect examples). It was the very thing to do in the 1960s in F1 engine design, but it hasn't been seen much in the last 20 years. The independent design doesn't even rely on a single collector, but can also be used in open "stack pipe" systems. Drag racers still use these today, but they're almost alone. F1 cars don't, partially for space constraints, but what little you can see of the designs they produce hint at an "interference" design.

Bikes have flopped back and forth over the years, and there have been some "mostly independent" designs made even recently, but this is often for styling reasons, not power curve reasons.

It's a fascinating subject, and one that's not really gone over enough, in my view.

I first thought about using the suby's headers, but after a close look, it was obvious that it wouldn't happen. so i decided to go all out (headers, ITBs, etc...)
the system i'm working on right now will most probably be a 4 into 1 setup somewhat similar to, IIRC, the triad headers made for the Type4...
The setup will be designed around midrange torque and throttle response...

QUOTE (TonyAKAVW @ Mar 30 2006, 12:54 PM)

Mostly due to simpliticy, I'm planning on using a set of the Outback Motors collectors, and then making a turn up and over the axles and having a pair of mufflers on either side of the transmission. I'll have a tube that crosses over above the transmission (I believe that there is enoguh space.. otherwise I'm not sure what to do). The figure below shows the crossover as a red circle, which goes across to the other side. I want the maximum ground clearance possible since I'm doing some funky stuff with diffusers and stuff. -Tony

Tony, are you going to use an existing design for "inspiration" ? Do you have an idea how far along the pipes the crossover should be ? I still get kind of dizzy (no pun intended..) thinking about the firing order, and the timing of pulses and reflections and whatnot.

I know the "old school" trick is to install the pipes w/o a crossover, then run the engine hard until you can see part of the pipes glowing red - and that's where the crossover goes.

Also, are you going to cut and join pre-bent J-tubes, or have your pipes bent to fit ?

The outback stubs are premade. Just weld on the pipes to the muffler.
This won't make an ideal exaust system but it should be better than the nasty stock setup that hangs way to low and is small and full of bad angles and rude joints - the engine somehow still makes 165Hp.

My design is solely based on my need for ground clearance. Because the transmission is in the way, there isn't a lot of room for putting in crossovers in any optimal location unless it just happens to be where there's a clear spot. If you don't need the extra ground clearance, then you can do a crossover anywhere under the car.

So my justification for putting it where I did was that its about the only place for it. I've really got no understanding of exhaust system design, so my setup is pretty much a hack. My understanding though is that yo get something that works well, you build, dyno, repeat until something works the way you want it. At this point in my project I'm just going to go with this. If someone came up with a proven exhaust system I would be eager to try it out, but I don't have the resources to make one right.

I know that Scott Thacher's exhaust system is essentially a home-made triad system. Mongrel GT is using a dual in-dual out muffler with a set of Outback collectors. Both of those setups consume more space below the axles than I am willing to sacrifice. It would be very interesting to dyno both of those cars and see what difference there are.

-Tony

QUOTE (lapuwali @ Mar 30 2006, 01:05 PM)

The Smith book, as you say, is an old book (I also have a copy). From what I've been able to see, the "independent" scheme has pretty much fallen out of favor in all engines that are extremely exhaust tuned (look at motorcycles and F1 cars for perfect examples). It was the very thing to do in the 1960s in F1 engine design, but it hasn't been seen much in the last 20 years. The independent design doesn't even rely on a single collector, but can also be used in open "stack pipe" systems. Drag racers still use these today, but they're almost alone. F1 cars don't, partially for space constraints, but what little you can see of the designs they produce hint at an "interference" design. Bikes have flopped back and forth over the years, and there have been some "mostly independent" designs made even recently, but this is often for styling reasons, not power curve reasons. It's a fascinating subject, and one that's not really gone over enough, in my view.

My impression, based on the previous DIY header thread and on other discussions, is that real-world header design always comes down to a (whole lot of) trial and error. Our own Jake R., no slouch when it comes to tweaking and testing, devoted a lot of time to a header design and (to paraphrase) decided it wasn't worth it. Maybe that Simon Cowell avatar got me feeling all insecure, but I started getting the sense that to do it right, you'd really need to do some serious computer modelling, instrument the system with a butt-load of sensors, and still go through the trial and error. In another life, that would be, uhh, fun.


So my thinking was that even though the Indpendence approach might not yield the ultimate in performance, it would at least be less likely to produce one of those unhappy, unpredictable coincidences that can make a header system totally suck. But maybe the pitfalls of designing a Interference type system aren't really that catastrophic. If there's nothing else on the market that's appropriate, it only has to be better than stock manifolds.

BTW, I've got to say I'm intrigued by the idea of "Interference Pipes" that Smith talked about. The idea is to branch off of a single primary right near the exhaust port, run about 18" of pipe, and then weld a cap on it. So there's no net flow of gas through that branch, but a pressure pulse travels down and reflects off the end, going back towards the port. The main drawback, of course, is all the extra weight and complication, but the theory is cool - just like transmission line theory in the EE world. For a DIY project, you could just clamp on a piece of thin-walled mild steel tube and tune the response by chopping off an inch at a time. And everybody who saw it would be sure to call for Whiskey Tango Foxtrot...


BTW, how come nobody else has written a book on this in the past half century or so ?

QUOTE (Porcharu @ Mar 30 2006, 02:16 PM)

The outback stubs are premade. Just weld on the pipes to the muffler. This won't make an ideal exaust system but it should be better than the nasty stock setup that hangs way to low and is small and full of bad angles and rude joints - the engine somehow still makes 165Hp.

What do these guys cost, BTW ?

QUOTE (jsteele22 @ Mar 30 2006, 01:35 PM)

BTW, how come nobody else has written a book on this in the past half century or so ?

There haven't been, to my knowledge, any books dedicated to intake and exhaust design (not seriously, anyway, mostly how-tos on how to replicate a common V8 header, for example), at least not for the general public. Perhaps there are some SAE papers available. There have been a number of books published that cover the topic, however. Bell wrote a couple of books that cover the subject, though even those are 20 years old or more. There's also Taylor's general IC engine books.

Scientific studies of the problem are very sparse in general literature, however. Again, the SAE, or the British equivalent, would be most likely to offer anything. Something like Ricardo's work really needs to be published again, and I suspect Taylor's work was an attempt at this. The field has gotten so big, and generally so focused on just meeting regulations, that I doubt it will happen.

Tuners are rarely all that science-minded, and even then are frequently not interested much in theory, and even then are loath to publish for commercial reasons. Ricardo's own organization has grown into a very lucrative consulting operation which sells its research.

The fact that Smith is still in print after all this time suggests there's a ready market for such a book, but will require someone bright enough to realize that most readers will never be in a position to buy their tuning services (geographical, say), or will go to them to have them done, and very few people will "steal" the research to enrich themselves. Advertising that's not only free, but you get the (tiny) amount of money from the book sales, too.

One reason I suspect Jake decided not to research headers is that the Tangerine header is likely close to optimal for the Type IV. Any gains to be had there are so small that it wouldn't be worth the expense of pursuing them. Easier to just buy one from Chris.

The Tangerine would also likely be an excellent design to base a Subaru header around. I think Chris has even joked about getting a Soob engine into his shop. He may well be thinking about it seriously. The main problem for him would be that most high-power Subarus are turbocharged, so there's a lot less market for good headers than there would be otherwise.

QUOTE (jsteele22 @ Mar 30 2006, 01:53 PM)

QUOTE (Porcharu @ Mar 30 2006, 02:16 PM) The outback stubs are premade.  Just weld on the pipes to the muffler. This won't make an ideal exaust system but it should be better than the nasty stock setup that hangs way to low and is small and full of bad angles and rude joints - the engine somehow still makes 165Hp. What do these guys cost, BTW ?

$150 for a pair.

QUOTE (lapuwali @ Mar 30 2006, 03:38 PM)

The Tangerine would also likely be an excellent design to base a Subaru header around. I think Chris has even joked about getting a Soob engine into his shop. He may well be thinking about it seriously. The main problem for him would be that most high-power Subarus are turbocharged, so there's a lot less market for good headers than there would be otherwise.

Yeah, and there's no single target : different engine positions, mounting bars, shifters, trannies, etc. And as good as the Tangerine design seems to be for a 914, it's probably way off the sweet spot for a Subie : I'm guessing stuff like cam timing, valve port design, intake runners, etc. are different enough to make the "ideal" header parameters quite a bit different. The one thing I don't have a feel for yet is when a set of headers "sucks", how bad are they really?

Maybe, maybe not. If Jake finds it adequate to use a single design header for everything from FP race engines to street 2056s to 2.5L monsters, all with substantially different cam timing, strokes, etc. then it's unlikely that same header wouldn't work decently on a Subaru engine. It may not be optimal, but it would likely be much better than the sucky Subaru stock systems. I've also always been led to believe that cam timing and esp. primary length need to be paired, but perhaps that's just not so.

Chris has mentioned that the primaries on the European header are the wrong length, and one would think that would depend heavily on cam timing and/or stroke and/or SOME other major parameter, but perhaps not.

I'm guessing this kind of thing really starts to make a difference when you're running so much valve overlap that exhaust tuning makes a really big difference. This kind of thing doesn't usually start to appear until you're way up the tuning scale, above 100hp/liter. An NA 2.5 Soob only makes 66hp/liter stock with it's stock exhaust, so it's quite a ways from being radically tuned (still way better than the 50hp/liter Type IV). Bike engines now are pretty heavily exhaust tuned, and they're all approaching 200hp/liter these days. Their torque curves are very lumpy by car standards, and exhaust primary length makes a noticable difference to the lumpiness and peak torque.

I have no idea what kind of power they're getting out of the FP Type IVs these days, but I'd guess it's still not in the 100hp/liter range. Those are certainly going to be the most radically tuned Type IVs in use right now, I'd guess.